Ocean Engineering 111 (2016) 369383

Contents lists available at ScienceDirect

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

Experimental and numerical study of the slamming load on the bow

of a chemical tanker in irregular waves
Shan Wang, C. Guedes Soares n
Centre for Marine Technology and Ocean Engineering (CENTEC), Instituto Superior Tcnico, Universidade de Lisboa, Portugal

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

Article history: This paper presents calculations of ship motions, slamming occurrence probability and slamming loads
Received 28 November 2014 on the bow of a ship hull in irregular waves. The results are compared with the experimental data from
Accepted 16 November 2015 model tests of a 170 m chemical tanker with Fn 0 in head seas. Ship motions are calculated by using a
Available online 1 December 2015
partially nonlinear time domain code based on strip theory. The probabilities of slamming occurrence at
Keywords: the bow are studied numerically and statistically, and compared with the experimental data. The
Slamming occurrence experimental data are analysed statistically, to determine the relationships between the measured
Slamming load pressure and the entry velocity and with wave parameters. Two estimated signicant slamming events
Bow slamming are simulated by using the Arbitrary-Lagrangian Eulerian (ALE) algorithm, based on the calculated
Irregular wave
relative entry velocities in the numerical procedure.
Ship motions
& 2015 Elsevier Ltd. All rights reserved.
Arbitrary-Lagrangian Eulerian

1. Introduction
When a ship travels in rough seas, it will impact with water
because of large vertical relative motions between the ship and the
wave surface. This hydrodynamic impact phenomenon is dened
as slamming. This is one of the most complicated dynamic phe-
nomena of a ship operating in rough sea. The impulsive pressure
loads induced by slamming may cause local damage on the bottom
structure or endanger global ship strength due to the contribution
of the whipping bending moment to the wave bending moment. In
rough seas, this impact force is so large that many ships have
reported local structural damages due to the slamming loads,
especially in head sea waves with high forward speed. In the initial
stage of ship design, the complete information on slamming,
including where the slamming occurs, the frequency of slamming
occurrence, the characteristics of slamming loads and the exible
responses due to the slam-induced loads, are required. This work
will focus on the frequency of slamming occurrence and the
slamming loads due to the impact between the ship hull and the
water surface.
A slam will occur when the relative vertical motion between a
ship section and wave surface is equal to the still water draft of the
section and at the same time the relative vertical velocity is
negative, i.e. the ship section enters the water. The other condition
to be satised is that on the previous time moment the section
n
Corresponding author. Tel.: 351 218417957.
E-mail address: c.guedes.soares@centec.tecnico.ulisboa.pt (C. Guedes Soares).
must have been out of the water. Some authors dene that a slam
occurs when the relative velocity is higher than a critical value
(Ochi and Motter, 1973). For the SWATH cross-structure, Zarnick
and Hong (1986) developed a method to estimate the expected
number of water impacts per unit length. The occurrence of a
cross-structure slam is dependent upon at least three conditions:
entry of the cross structure into the water; second an entry velo-
city exceeding some threshold velocity; a small angle between the
cross-structure and free surface. Today the most widely accepted
probabilistic slamming model requires only the rst two condi-
tions to be true, e.g. Psaraftis (1978), Faltinsen (1993) and Senja-
novic and Parunov (2001). Guedes Soares and Pascoal (2005) have
studied the probability distributions of green water on deck and
showed that based on a relative motion formulation the predic-
tions of the theory agreed well with the measurements. However
their results were obtained with a model with basically vertical
sided bow, while the ones of Buchner (1998, 2002) account for a
different are and introduce the corrections for that effect. Guedes
Soares et al. (2007) performed an experimental programme to
determine the instants when impact occurs for a specic bow
shape, and it was found that wave impact at the bow is highly
correlated with the local wave steepness.
Of interest are the impact loads such as bow-are slamming,
bottom slamming, stern slamming, and green water. These impact
loads are of a transient nature and can cause severe structural
damages. Although the loads are widely varying in their char-
acteristics magnitude, rise time, duration, etc., all involve the
impact at high relative velocity between the free surface of nearly

http://dx.doi.org/10.1016/j.oceaneng.2015.11.012
0029-8018/& 2015 Elsevier Ltd. All rights reserved.
370 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383
incompressible seawater and the hull structure. The transient
impact loads can be highly non-linear and may be strongly affec-
ted by the dynamic response of the hull structure. The structural
analysis of ship hulls due to the impact loads has relied on sim-
plied approaches for years, considering the hydrodynamic load-
ing calculated by assuming a rigid structure, and the loading is
then applied to the structure in a quasi-static manner.
Many methods have been developed to calculate the slamming
pressure on a rigid body that penetrates the wave surface with a
prescribed velocity. Among the earliest was the method of von
Krmn (1929) who estimated the pressure on sea plane oats.
Wagner (1932) proposed an asymptotic solution for water entry of
two-dimensional bodies with small local deadrise angles within
the assumption of potential-ow problem without gravity and no
air cavity formed during the impact. Armand and Cointe (1987)
and Howison et al. (1991) developed this work by accounting for
the effect of nonlinear jet ow in the intersection region between
the body and free surface using asymptotic matching expansions.
When the wedge impacts vertically with water at a constant
velocity, Dobrovol'skaya (1969) derived an analytical solution by
transferring the potential ow problem for the constant water
entry into a self-similar ow problem in complex plane, which
takes advantage of the simplicity of the body geometry and is valid
for any deadrise angle within the limits that compressibility and
hydroelastic effects do not matter and air entrapment does not
occur. Zhao et al. (1996) generalised the work of Wagner (1932) by
applying linearised free-surface boundary conditions on the hor-
izontal plane at the splash-up height and imposing the body
boundary condition on the actual position of the body. They solved
the problem numerically using a boundary-integral equation
method. For wedges and a section with are, their results agree
well with experimental data and calculations from a fully non-
linear boundary element method presented by Zhao and Faltinsen
(1993) for the slamming force and the body pressure distribution.
Initial asymmetric wedge-impact ows with horizontal as well as
vertical impact velocity were examined by Judge et al. (2004),
which employed the method of two-dimensional vortex dis-
tributions to model the initial-boundary-value problem. Their
numerical calculations agreed well with the experimental study of
drop tests for small degrees of asymmetry and small ratios of
horizontal to vertical impact velocity. Hermundstad and Moan
(2005) studied numerically and experimentally the bow are
slamming on a RoRo vessel in regular oblique waves. The relative
motion of ship and wave was calculated by a nonlinear strip the-
ory, and the slamming loads prediction was conducted by using a
simplied 2D BEM method, which is based on a generalised two-
dimensional Wagner formulation and solved by the boundary
element method. Kapsenberg and Thornhill (2010) developed an
approximation method based on momentum theory enhanced
with pile-up effects due to the wave surface, to predict the impact
pressure and force induced by bow-are slamming. Recently,
much work about the two-dimensional and three-dimensional
bodies impacting a calm water surface was investigated by using
SPH and ALE algorithm, e.g. Veen and Gourlay (2012), Wang and
Guedes Soares (2013), and Wang and Guedes Soares (2014a). The
two-dimensional slamming pressure calculated by Veen and
Gourlay (2012) was applied to the estimation of the slamming
loads on a ship in head waves. The general conclusions from their
results are that this method is capable of predicting the time
history of total slamming force, slamming pressure distribution
and pressure peaks on ship-sections and three-dimensional bod-
ies, when proper related parameters are used. Application of these
methods on water impact problems is still a big challenge for some
issues involved. The numerical convergence is very sensitive to
mesh size and contact stiffness of the model. At the beginning of
water impact, uid compressibility matters for bodies with very
small deadrise angle and it is not usually considered in
numerical model.
Wang and Guedes Soares (2014b) studied the probability of
slamming occurrence of a chemical tanker subjected to irregular
waves. Their work focused on the bottom slamming probability,
and the results showed that the extreme seas referred causes
severe slamming problem at the bow and stern of the chemical
tanker even without forward speed. The present work focuses on
the analysis of the slamming induced loads on the ship sections at
the bow. The relative ship motions and velocities between a che-
mical tanker and irregular seas surface are calculated numerically
by a nonlinear time domain method based on strip theory, which
are implemented in an ALE algorithm to predict the impact loads
on the ship hull. Model tests are conducted for the vessel with
Fn 0 to study the seakeeping performance under extreme seas.
The measurements include the wave surface elevation, ship
motions, vertical bending moment and also the wave-induced
pressure on the bow and stern of the ship hull. Statistical analysis
of the experimental data is conducted. The numerical relative
motions are compared with the measured ones for several posi-
tions, achieving good agreement. The predicted slamming pres-
sures from numerical method are compared with the results from
pressure transducers at the bow in the model test.
2. Methodology
In the ship design stage, the exible responses of the ship due to
the wave-induced loads are of great signicance. An analysis of
slamming induced responses may require an accurate prediction of
not only of the slamming loads but also of the conditions under
which slamming occurs, where the slamming occurs and the fre-
quency of the occurrence. In this paper, ship motions in three irre-
gular sea states are numerically calculated by using a fully nonlinear
time domain code. Based on the calculated relative vertical motions
and velocities between the ship hull and wave surface, the slamming
probabilities on the bow of the vessel are estimated and discussed.
The slamming pressures on concerned ship sections are numerically
calculated by ALE algorithm, by considering the ship sections enter
into calm water with a constant entry velocity. Numerical results of
ship motions, slamming probability and slamming pressures on the
bow are compared with the experimental data from model tests.
Statistically analysis of the experimental data is performed, regarding
the relation between the peak pressures and the impact velocities and
the free surface elevation. It is assumed that the ship hull has negli-
gible deformations due to the slam-induced loads and the slamming
loads are not affected by the whipping vibrations of the hull. Unless
otherwise specied, data are presented in full scale.
2.1. Probability of slamming occurrence
Consideration is given to a ship advancing at a constant velocity
U in irregular waves. Slamming occurrence in head seas depends
on relative ship motion with respect to the wave surface.
When the heave and pitch motions are considered, the relative
vertical motion of the ship is given as follows:
RM 3  x5  Ut 5  a 1
where 3 is the heave motion, 5 is the pitch motion of the ship at
a coordinate system as shown in Fig. 1. a is the irregular water
surface elevation, which can be obtained by linear superposition
principle. For a long-crested irregular sea described by a sea
spectrum S, the wave elevation can be given as follows:
XN  
a A sin ej t kj x j
j1 j
2
 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383
Fig. 1. Coordinate system of the ship motion.
where Aj , ej , kj and j (j 1,N) are the amplitudes, encounter
frequencies, wave numbers and phase angles of the regular waves,
respectively. The encounter frequency between the ship and
waves is e 0 0 2 U=g, where 0 is the wave frequency.
Then, the relative vertical velocity is given as follows:
d3 d and irrotational. Equations of motion were obtained by the equi-
VR  x 5  U 5  w
dt dt
where w means the z-component of the undisturbed wave
velocity.
As mentioned in Faltinsen (1993), in order for slamming to
occur at a point on the surface of the ship, it is necessary that the
relative vertical motion at the same longitudinal position of the
ship is larger than the vertical distance d from the still water
surface to the point, which means the slamming probability P can
be given as follows:
Pslamming PRM 4 d
The random nature of the slamming phenomena has motivated
many researchers to approach the slamming problems using sta-
tistical tools. One of the widely used statistical works was done by
Ochi (1964). His work, however, is only applicable to bottom
slamming where the forward part of the vessel has emerged from
the free surface and re-enters with a relatively high velocity. Ochi
(1964) sets the threshold velocity equal to
V cr 0:093gL1=2
where g is the gravitational acceleration and L is the ship length,
and then the slamming occurrence is given as follows:
Pslamming PV R 4 V cr
Statistical estimates of slamming pressure in an irregular sea
can easily be obtained by assuming that the magnitude of the
impact pressure is proportional to the square of VR. The irregular
waves are assumed as short-term stationary and can be described
by a sea spectrum S(). If the maxima (peak values) of the relative
velocity of the ship follow a Rayleigh distribution, the probability
of slamming occurrence can be estimated statistically as the joint
probability between the bow emergence and exceedance of the
threshold velocity during the re-entry. The expression was given in
Jensen (2001) as follows:
!!
2
V 2cr d estimation of ship motions in high sea states. The method was
Pslamming exp  7
2 2v 2 2r
where d is the draft at the slamming location, v and r are the
standard deviations of relative vertical velocity and relative ver-
tical motion, respectively, which are expressed by
Z 1  
V R 2
2v S  d
 8
0 a
Z   Only heave and pitch motions are considered for the numerical
1  2
2r S RM  d
 9
0 a
371
where 2v and 2r are the variance
 of
 the relative
 velocity and
motion respectively. V R =a  and RM =a  are the transfer
functions of the relative velocity and motion respectively.
2.2. Numerical ship motions
Fonseca and Guedes Soares (1998a, 1998b) calculated nonlinear
ship responses using a partially nonlinear time domain code based
on strip theory. The model is based on the strip theory and the
assumption of small amplitude incident waves and small unsteady
motion of the hull and the dominant nonlinear contributions is
associated to the buoyancy forces (hydrostatics and Froude Kry-
lov). Therefore the calculation is considered accurate, when the
radiation and diffraction terms are kept linear. It means that the
relative motions are of large amplitude, although the waves are
assumed with small amplitude. The hull is assumed to be slender
and forward speed is small, and the ow is assumed to be inviscid
3 librium equations between the hydrodynamic external forces and
the mass and gravity forces, combining the linear and nonlinear
terms. These equations of motion were solved in the time domain
by a numerical procedure. The exciting forces due to the incident
waves are decomposed into a diffraction part and the Froude
Krylov part. The diffraction part is kept linear. Since this is a linear
problem and the exciting waves are known, it can be solved in the
frequency domain and the obtained transfer functions can be used
to generate time series of the diffraction force and pitch moment.
The FroudeKrylov force is calculated by the integration at each
4
time step of the associated pressure on the wetted surface of the
hull. The radiation force is represented by innite frequency added
mass, radiation restoring coefcients and convolution integration
of memory functions. When the relative vertical motion is larger
than the free board, the force associated with the green water on
deck is calculated by momentum theory.
The numerical results from this partially nonlinear time
domain method were compared with available experiments in
5 Fonseca and Guedes Soares (2002, 2004a, 2004b); however, there
were some discrepancies mainly related to the high speed effects
and simplications in the numerical method. They assumed linear
radiation/diffraction and the memory functions and hydrodynamic
6 coefcients were calculated for mean water level. This assumption
holds valid for ship motions in small and moderate seas. In
extreme seas, the ship's wetted surface area changes a lot, lar-
gely resulting in change of the hydrodynamic coefcients. To
improve the partially nonlinear time domain method, Rajendran
et al. (2015a, 2015b) included the body nonlinearity in the calcu-
lations of radiation/diffraction forces, and updated the high fre-
quency added mass at each instant time. By comparing the ship
motions with experimental data from model tests of a cruise
vessel with a low Froude number in very high irregular seas, it was
proved that inclusion of body nonlinear radiation and diffraction
forces in the partially nonlinear code signicantly improved the
regarded as a fully nonlinear time domain method, and is used in
this work. Relative motions at bow calculated using the fully
nonlinear time domain method were found to be slightly under-
estimated. Inaccuracy in the estimation is attributed to inability of
the code to include wave run-up at the bow and nonlinearity of
the waves. It should be noted that this method is only applicable to
estimation of heave, pitch and surge motions of a ship in waves.
and experimental calculations of the relative vertical motions
between the ship hull and waves in this work.
372 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383
2.3. Slamming loads
The time histories of the relative motions and velocities are
calculated from the numerical procedure mentioned above, and
then the relative vertical velocities between the concerned ship
section and the wave surface can be obtained for possible slam
durations. It is assumed that only heave and pitch motion con-
tributes mostly to the water impact loads. With the calculated
relative vertical velocity and geometry of a ship section, the
slamming pressure on the wetted surface of the ship hull can be
predicted by an Arbitrary-Lagrangian Eulerian (ALE) algorithm
implemented in a commercial software LS-DYNA. The slamming
event is simulated as a 2D ship section impacting with calm water
at a constant velocity, namely the calculated relative vertical
velocity at the moment of impact.
For an ALE solver, the uid is solved by using an Eulerian for-
mulation, while the structure is discretized by a Lagrangian approach.
A penalty coupling algorithm enables the interaction between the
body and the uids. The remap step in the ALE algorithm applies a
donor cell HIS (Half-Index-Shift) advection algorithm to update uid
velocity and history variables. The interface between the solid struc-
ture and the uids is captured by Volume of Fluid method. The dif-
ferential equations governing this are numerically solved. This is an
explicit nite element method and is unconditionally stable. The
detailed description of this ALE algorithm can be found in Aquelet
et al. (2005). To ensure the computational stability, very small time
step is required, as discussed in Wang and Guedes Soares (2014a).
Convergence study was conducted and comparisons between
numerical results and experimental data were made in Wang and
Guedes Soares (2012, 2013, and 2014a), for 2D ship sections and 3D
buoys. The general conclusions are that the numerical method is
capable of predicting the time history of impact force, pressure his-
tories and pressure distributions on the structural bottom surface.
However, the most obvious disadvantage is the high computational
efforts associated with the small time step. Their work indicated that
the numerical predictions of the pressure values at the keel of a
section had some discrepancies with the related experimental data,
but good agreements were achieved on the upper part. The numerical
peak pressures at the keel were smaller than the measured values
generally, due to instability of simulations at the initial moment of
water impact. In this work, two sections at the bow are simulated by
ALE solver, and the slamming pressures on the upper of bow are
calculated numerically and compared with the measured values from
sensors in the model test.
3. Model test
The numerical calculations of slamming probability, relative
motions and velocities, and the slamming pressures are compared
with the experimental data from the model tests of a chemical
tanker with a Froude number of 0 (Fn 0) and subjected to head
waves. From the tests, the vertical bending moment at the middle
ship, the ship motions in all six degrees of freedom at the centre of
gravity, the pressures at the bow and stern on the ship and the
wave elevation from the gauges are recorded in time series.
The model tests of a chemical tanker are conducted in the
seakeeping basin of the Ocean Engineering Division of the Tech-
nical University Berlin at a model scale of 1:70. The basin is 110 m
long, with a measuring range of 90 m, the width is 8 m and the
water depth is 1 m. An electrically driven piston type wave gen-
erator is installed on one side of the basin. The wave generator is
fully computer controlled and a software is implemented which
enables the generation of transient wave packages, deterministic
irregular wave states with dened characteristics as well as tai-
lored critical wave sequences. During the tests, the model is xed
Table 1
Main particulars of the chemical tanker.
Items Full scale Model scale
Length overall (Loa)(m) 170.00 2.428
Length between perpendiculars (Lpp)(m) 161.00 2.300
Breath (m) 28.00 0.400
Depth (m) 13.00 0.186
Draft (m) 9.00 0.129
Displacement (ton) 30,666 0.089
Fig. 2. Cross sections of the chemical tanker under waterline.
with an elastic suspension system using a triangular towing
arrangement pulling the model without inducing a moment. The
longitudinal motions are restricted by a spring in front of and a
counter weight behind the model. With this arrangement, heave
and pitch motions as well as the measure forces and moments
remain unrestrained.
The main particulars of the investigated model are given in
Table 1. The geometries of the transversal cross ship sections are
illustrated in Fig. 2. The origin for all the following parameters related
to the positions on the model is located at the aft perpendicular at
keel level. The model is made of breglass reinforced plastic (GRP)
and segmented at the mid-section (Lpp/2). Both segments are con-
nected with three force transducers, which record the longitudinal
forces during the model tests. Based on the measured forces and the
given geometrical arrangement of the three force transducers, the
resulting vertical wave bending moment and the longitudinal forces
are obtained. The model is also equipped with 38 pressure transdu-
cers in all at bow and stern, while the work focuses on the pressure
loads at the bow due to steep large wave. Fig. 3 plots the sketch of the
pressure sensors at the bow, and a detailed description of the posi-
tions on the ship hull is given in the report of Clauss et al. (2011). To
obtain the relative motion between the model and water elevation, a
total amount of nine wave gauges are installed at the target locations
on the undisturbed water surface. The longitudinal arrangement with
locations of the nine wave gauges is shown in Fig. 4, which shows
that wave gauge 9 is placed at the position of the mid-ship section.
With respect to the width of the model, the gauges are located
sideways, parallel to the ship. The tests were conducted in head waves
at zero speed condition. A series of regular and irregular sea states
were investigated. Among them, three irregular cases, represented by
parameters peak period (TP), signicant wave height (Hs), and peak-
edness parameter (), are considered in this work,  as listed
 in Table 2.
The steepness of wave series is dened as H s = 1:56Tz2 , where Tz is
the average zero-up crossing period. For each sea state, six groups of
wave sequences with the same Tp, Hs, and are generated by JONS-
WAP spectrum within about 30 min full time, for long term statistical
analysis. The phases are random and differ within each group but are
constantly plugged for each set. For example, for the sea states listed
in Table 2, three runs are performed for each one, and run 1 has
another random phase than run 2 and 3, but same phases are used for
the three tests of run 1 with different sea states. In this work, only one
group of data is analysed for each sea state, and the same phase is
adopted.
 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383 373

Fig. 3. Left-sketch of the pressure sensors at the bow; right-installed pressure sensors at the bow of the ship model.

Fig. 4. Sketch of the model test setup and the locations of the wave gauges.

Table 2
Irregular sea states studied in this work.

Irregular sea states

Sea states Hs (m) Steepness Tp (s) Tz (s)

Model scale Full scale Model scale Full scale Model scale Full scale

Irregular 14 0.2357 16.5 3.3 0.06908 1.90041 15.9 0.1771 12.4

Irregular 17 0.1385 9.7 3.3 0.07130 1.43427 12 0.1329 9.3
Irregular 20 0.1642 11.5 3.3 0.08453 1.43427 12 0.1329 9.3

During the model tests, it was observed that an unusual
behaviour of some pressure transducers occurs, because of the
different temperature compensation behaviour of the sensors. The
model is xed with an elastic suspension system using a triangular
to wing arrangement, and its inuence on the vertical bending
moment is less than 2%. The mutual interference between the ship
model and wave gauges as well affects the measurements. Though
only degrees of freedom that are free in the experiment are heave
and pitch, small roll motions are found in the records of ship
motions. All of these factors may induce experimental uncertain-
ties. The details can be found in the report of Clauss et al. (2011).
4. Results and discussion
4.1. Comparisons of measured and calculated ship motions
In the numerical calculations of ship motion, 32 strips in the
longitudinal direction are used for this chemical tanker. The rst
step of the calculations is to compute the exciting force induced by
the wave eld. Then, it is necessary to represent the wave surface
elevation in time domain. The simulated wave surfaces are com-
pared with the measured values from wave gauges, for the case of
Irregular 14. Fig. 5 compares the numerical wave surfaces with the
measured values obtained from wave gauge 9 and wave gauge
7 under the condition of Irregular 14, showing very good agree-
ment at the mid-ship section and some discrepancies on the peak
values at the bow. As seen in the wave cycle around t1300 s in
Fig. 5(b), the measured positive peak is 32% higher than the
numerical ones, while the measured negative one is 28% lower.
Since it is not in the current scope of the studies to use a higher
order wave model, the current study makes use of the measured
experimental waves, and is directly applied at the centre of gravity
of the ship to preserve asymmetric characteristics of the waves.
For the nonlinear time domain code which is used to calculate the
ship motions in waves, the wave data in frequency domain is
required, so the experimental wave surfaces at amidships are
directly converted to the data in frequency domain by using FFT.
The harmonics are used for spatial transfer of the wave time series
from the measured position to centre of gravity of ship and to the
374 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383

numerical ones. Though the measured data of phase angles and

15
numerical calculations are slightly different, it can be observed that
Wave elevation (m)

Exp.
10
Numerical
the difference between them is steady. For the experimental data, the
5 ship motions in all six degrees of freedom are recorded at the gravity
0 centre of the vessel. Here, the vertical motions of the ship are derived
by Eq. (1), assuming that the vessel moves like a rigid body. For a real
-5
ship, the exible vibration is inuenced by the exciting force and will
-10
1000 1050 1100 1150 1200 1250 1300 in turn affect the loads. This phenomenon is more signicant at the
Time(s) bow. The simplications used in the fully nonlinear time domain
method also contribute to the discrepancies. The fully nonlinear time
15
Exp.
Wave elevation (m)

10
Numerical 10
5

Wave elevation (m)

Exp.
0 Numerical
5
-5

-10
1000 1050 1100 1150 1200 1250 1300 0
Time(s)

Fig. 5. Comparison of the wave elevation at the mid-ship section and the one -5
located at wave gauge 7 under the conditions with Irregular 14. (a) At mid-ship
1000 1050 1100 1150 1200 1250 1300
section; (b) at ship section with wave gauge 7.
Time(s)

10
Exp.
Wave elevation (m)
Exp.
5 Numerical Numerical
5
Heave (m)

0 0

-5
-5
1000 1050 1100 1150 1200 1250 1300
1000 1050 1100 1150 1200 1250 1300 Time(s)
Time (s)
Fig. 7. Comparison of the wave elevation at the mid-ship section and the one
located at wave gauge 7 under the conditions with Irregular 17. (a) At mid-ship
Exp.
section; (b) at ship section with wave gauge 7.
0.1 Numerical
Pitch (rad)

0 3
Exp.
2
Numerical
Heave (m)

-0.1 1

0
1000 1050 1100 1150 1200 1250 1300
Time(s) -1
15
-2
10 Exp. 1000 1050 1100 1150 1200 1250 1300
Numerical Time(s)
5
VM(m)

0 0.1 Exp.
Numerical
-5 0.05
Pitch (rad)

-10 0
-15 -0.05
1000 1050 1100 1150 1200 1250 1300
Time(s)
-0.1
1000 1050 1100 1150 1200 1250 1300
Fig. 6. Comparisons of the heave motion, pitch motion and relative vertical motion Time(s)
at the ship sections with wave gauges 7 under conditions of Irregular 14.
10
Exp.
bow by using linear dispersion. This may result in error in the 5 Numerical
estimation of wave elevation at the bow.
VM (m)

Fig. 6 shows time series of experimental and numerical heave, 0

pitch and vertical motions for the ship section located at wave gauge
-5
7 under the condition of Irregular 14. The vertical motions are cal-
culated by considering the heave and pitch motion. The agreement 1000 1050 1100 1150 1200 1250 1300
between the measured values and the calculations from this non- Time(s)

linear strip theory is generally good. For the heave motion, the results Fig. 8. Comparisons of the heave motion, pitch motion and relative vertical motion
show that the measured positive peak values are lower than the at the ship sections with wave gauges 7 under conditions of Irregular 17.
 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383 375

0.4 0.35
Rayleigh distribution

Probability distribution function

Rayleigh distribution 0.3

Probability distribition function

Numerical data
Numerical data
0.3 0.25

0.2
0.2
0.15

0.1
0.1
0.05

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Maxima of relative vertical velocity(m/s) Maxima of relative vertical velocity(m/s)

Fig. 9. Probability distribution of the maxima of the relative vertical velocity between the ship section with wave gauges 7 and wave surface. (a) Irregular 20; (b) Irregular 17.

Table 3 and Fig. 8 shows time series of experimental and numerical heave,
Comparisons of probability of slamming at the points with pressure sensors on the pitch and vertical motions for the ship section located at wave
ship hull. gauge 7. Similar to the results under the condition of Irregular 14,
good agreement at the mid-ship section and about 15% reduction
Probability of slamming occurrence for the points with pressure sensors
on the numerical peak values at the bow are observed for the wave
Sea states Ship sensors Numerical Experimental Statistically surfaces, while slight differences of peak values and phase angles
are found for ship motions.
Irregular 14 Sensor 1 0.0518 0.0181 0.0961
Sensor 2 0.0957 0.0434 0.1655
Sensor 3 0.0037 0.0017 0.0415
4.2. Slamming probabilities at the bow
Sensor 4 0.0133 0.0025 0.0890
Sensor 5 0.0448 0.0386 0.1341 For the model tests, a group of 30 min time series data is generated
Sensor 6 0.1381 0.0840 0.1791 for each run. The ship hull will encounter a large number of waves
Irregular 17 Sensor1 0.0404 0.0043 0.0592
during the tests. It is not practical to make calculations for each wave
Sensor 2 0.0164 0.0103 0.1165
Sensor 3 0.0041 0.0000 0.0213 encounter event and each ship section. The probabilities of slamming
Sensor 4 0.0140 0.0000 0.0511 occurrence for two concerned ship sections are calculated experimen-
Sensor 5 0.0276 0.006 0.0917 tally, numerically and statistically. The experimental and numerical
Sensor 6 0.0939 0.0192 0.1313 results are obtained based on both Eq. (4) and Eq. (6) by using the fact
Irregular 20 Sensor 1 0.0219 0.002 0.0315
Sensor 2 0.0451 0.0134 0.1010
that the relative motion and relative velocity are statistically indepen-
Sensor 3 0.0026 0.000 0.0080 dent (Faltinsen, 1993), and the statistical values are calculated by Eq. (7)
Sensor 4 0.0278 0.0049 0.0419 based on the numerical values of relative velocity and motion between
Sensor 5 0.0489 0.0198 0.1017 the ship hull and wave surface. All of them are calculated based on two
Sensor 6 0.1376 0.0670 0.1900
conditions: the relative velocity is larger than the critical value and the
relative motion is larger than the vertical distance from the still water
surface to the point. Since the statistical expression is based on the
assumption that the peaks of the relative velocity of the ship follow a
Rayleigh distribution, the numerical probability distributions of the
relative vertical velocity of one ship section (at wave gauge 7) are cal-
culated and shown in Fig. 9. The results indicate that the obtained
numerical probability distributions generally follow the Rayleigh dis-
tribution; therefore, it is believed that the statistical values with Ray-
leigh distribution assumption are reasonable.
Table 3 lists the probability of slamming occurrence for the
points of the concerned pressure sensors on the ship hull. The
values for the sensor 6 are highest, while they are lowest for the
sensor 3 for each sea state. It is consistent with the fact that the
point near the waterline more easily emerges from the wave and
Fig. 10. Geometry of the ship sections and the locations of the pressure sensors in re-enters. The experimental estimations are much lower than the
model scale. numerical calculations and statistical values based on Rayleigh
distribution. It is probably due to the differences of the peak values
domain method does not take into account the interaction between of the wave surface and radiation and scattering effects not
the incident wave eld and the ship hull which generally result in included in the denition of relative motion. The numerical results
wave run-up as the wave encounters the hull. This probably is one of are close to the statistical calculations based on Rayleigh dis-
the reasons for the discrepancies. In addition, the wave elevation at tribution, thus, the expression of Eq. (7) can be used in the
amidships is spatially transferred to the bow using linear dispersion. numerical procedure for the calculation of slamming probability.
Even though nonlinear wave surfaces are applied, linear wave Faltinsen (1993) indicates that an often used criterion is that a
potential is used for calculation of wave pressure which excludes any typical ship master reduces the speed if slams occur more than
inuence of nonlinearity in the waves. 3 of 100 times that waves pass the ship. This means that if slam-
Under the condition of Irregular 17, Fig. 7 plots the comparisons ming probability is calculated to be larger than 0.03, the ship
of wave surfaces between the numerical calculations and the speed should be reduced to a level so that the slamming prob-
measured values obtained from wave gauge 9 and wave gauge 7, ability is less than 0.03. Here, the speed of the ship is zero, so it is
376 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383

Exp. vertical relative motion Estimated entry point Estimated exit point
10

RVM (m)
0

-10
1600 1650 1700 1750 1800 1850 1900 1950 2000
10
RVV (m/s)

-10
1600 1650 1700 1750 1800 1850 1900 1950 2000
0.03
pressure (bar)

Pressure sensor 6 irregular14

0.02

0.01
0
1600 1650 1700 1750 1800 1850 1900 1950 2000
Time(s)

Exp. vertical relative motion Estimated entry point Estimated exit point
10
RVM (m)

-10
1600 1650 1700 1750 1800 1850 1900 1950 2000
10
RVV (m/s)

-10
1600 1650 1700 1750 1800 1850 1900 1950 2000
0.02
pressure (bar)

Pressure sensor 6 irregular17

0.01

0
1600 1650 1700 1750 1800 1850 1900 1950 2000
Time(s)

Exp. vertical relative motion Estimated entry point Estimated exit point
10
RVM (m)

-10
1600 1650 1700 1750 1800 1850 1900 1950 2000
10
RVV (m/s)

-10
1600 1650 1700 1750 1800 1850 1900 1950 2000
0.04
pressure (bar)

Pressure sensor 6 irregular20

0.02

0
1600 1650 1700 1750 1800 1850 1900 1950 2000
Time(s)

Fig. 11. Experimental results of the relative vertical motion, relative vertical velocity and pressure on sensor 6. (a) Irregular 14; (b) Irregular 17; (c) Irregular 20.

considered that if the calculated slamming probability is larger
than 0.03, the slamming event is signicant. It is found that slams
occur more frequently than the criterion from the point of view of
numerical calculations and statistical values based on Rayleigh
distribution, for most of the points on the ship section located at
the fore part. Therefore, this two ship sections with these pressure
sensors are considered for the slamming load calculation.
4.3. Impact loads
To calculate numerically the impact loads between the ship hull
and the wave surface it is required to determine the relative ver-
tical velocity. Before the numerical calculations, the experimental
data are analysed statistically rst, to study the relationships
between the measured pressure and the entry velocity and also
the wave parameters. Fig. 10 plots the geometries of three ship
sections and corresponding positions of the pressure transducers
in model scale. As illustrated above, the slamming probability of
the ship section located at wave gauge 7 is high, and the pressure
transducers 3, 4, 5, and 6 are just mounted on this position. Thus
the ship section labelled with Section 2 is selected for the sta-
tistical analysis. Fig. 11 shows the relative vertical motions, relative
vertical velocities between pressure transducer 6 and the wave
surface from wave gauge 7, together with the pressure values. The
results from three different sea states are all presented here. It
should be noted that the motions and velocities are given in full
 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383 377

ship scale, while the pressure values are in model scale because impact loads for two cross sections with different geometries, to
the numerical simulations will be conducted in model scale. demonstrate the relationship between the section geometry and
As seen in Fig. 11, the entry and exit points are predicted in the the characteristics of slamming loads. They justied that the
curves of the relative vertical motions. Corresponding to the entry momentum slamming force is larger than the initial impulse for a
points, the entry velocities are marked with red circles. Compared ship section with a small deadrise angle keel and a big are. As
with the time series of the pressure values, it is found that the seen in Fig. 10, the pressure sensors are all placed on the bow are.
time instants of entry velocities are in good agreement with the Correspondingly, the measured slamming pressures are mainly
rise time of the pressures, although for some entry points, there due to the change of the momentum as the hull immerse into the
are no obvious associated pressure peaks, because the corre- water and have longer durations.
sponding entry velocities are too small. It is believed that the The theorem of impulse assumes that the maximum impact
measurements of the model tests for the ship motions and pres- pressure is proportional to the quadratic impact velocity as
 2
sure values are accurate at least from the time point of view. For P max x 12kDwrel x; t=Dt  , where k represents the pressure
some high entry velocities, high impulses occur in the curve of coefcient and depends on the sectional form, especially the local
pressure load. Taking the Irregular 14 case for example, around deadrise angle at the point considered, and Dwrel x; t=Dt is the
t 1860 s, an entry velocity about  6 m/s, results in a pressure impact velocity at the time instant t. Here, impact velocity means
peak of 0.03 bar. the entry velocity which is obtained from the calculated relative
Guedes Soares (1989) proposed that the slamming force is vertical velocity between the ship hull and waves. These two terms
considered to be given by the sum of two components. The rst are used interchangeably in this work. The k value was evaluated
one is an impact component related to the impact of bottom on for any section by Ochi and Motter (1973) and Stavovy and Chuang
the water surface and characterized by a large peak with a small (1976) based on experimental data. Wang and Guedes Soares
duration of milliseconds. The second one is given by the rate of (2012) compared the values from various methods with the
change of the hydrodynamic momentum as the hull enters into numerical calculations for the sections with different deadrise
the water. Fonseca et al. (2006) made two calculations of the angles. For the ship bow-ared section like Section 2 illustrated in

Pressure sensor006 irregular14-1 Pressure sensor004 irregular14-1

0.04 0.02
Measured y = 0.00099*x + 0.0024
y = 0.0021*x + 0.0048
2 linear y = - 0.00017*x 2 + 0.0019*x + 0.0015
y = 0.0005*x - 0.00057*x + 0.0073
0.03 quadratic 0.015 data 1
Peak pressure (bar)
Peak pressure (bar)

linear
quadratic
0.02 0.01

0.01 0.005

0 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Impact velocity(m/s) Impact velocity(m/s)
Pressure sensor006 irregular17-1 Pressure sensor004 irregular17-1
0.03 0.02
y = 0.0023*x + 0.0036 y = 0.0018*x + 0.00021
y = - 7.6e-06*x 2 + 0.0023*x + 0.0036 y = 0.00016*x 2 + 0.0011*x + 0.00077

Measured 0.015 Measured

Peak pressure (bar)
Peak pressure (bar)

0.02 linear linear

quadratic quadratic
0.01

0.01
0.005

0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Impact velocity(m/s) Impact velocity(m/s)
Pressure sensor006 irregular20-1 Pressure sensor004 irregular20-1
0.06 0.02
y = 0.0037*x + 0.0007 y = 0.0019*x + 0.00026
0.05 y = 0.00088*x 2 - 0.0015*x + 0.0063 y = 5.7e-05*x 2 + 0.0015*x + 0.00062
Peak pressure (bar)

0.015 Measured
Measured
Peak pressure (bar)

0.04 linear
linear
quadratic
quadratic
0.03 0.01

0.02
0.005
0.01

0 0
0 2 4 6 8 0 2 4 6 8
Impact velocity(m/s) Impact velocity(m/s)

Fig. 12. Relationship between the impact velocity and the peak pressures for the pressure sensor 6 on the bow.
378 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383

Fig. 10, it is attempted to search a relationship between the peak based on experimental data. It is because the relative vertical
pressures and the relative impact velocity, therefore, linear and velocity decreases during the entry process of the model test.
quadratic tting of the measured values is made. Fig. 12 plots the In the present work, the chemical tanker has no sailing speed.
relationship between the impact velocities and the peak pressures The impact between the wave and ship hull is induced by the
obtained from pressure sensor 6, together with the linear and waves. It is believed that the local wave characteristics are of great
quadratic tting curves. Under the same sea states, it is observed signicance to the magnitude of the wave impact. Fig. 13 plots the
that the pressures from sensor 6 are higher than the values from free surface elevation at the entry point versus the pressure peaks.
sensor 4 for an identical value in the x-axis on the tting curves They are negatively linearly correlated. It means that the lower

Pressure sensor 6 irregular14

0.04
y = - 0.0007*x + 0.008 Measured
linear
Peak pressure (bar) 0.03

0.02

0.01

0
-10 -5 0 5 10
Free surface elevation (m)

Pressure sensor 6 irregular17 Pressure sensor 6 irregular20

0.04 0.05
y = - 0.0013*x + 0.0049 Measured y = - 0.0018*x + 0.0039 Measured
linear 0.04 linear fitting
0.03
Peak pressure (bar)

Peak pressure (bar)

0.03
0.02
0.02

0.01 0.01

0
0
-0.01
-10 -8 -6 -4 -2 0 2 4 -10 -5 0 5
Free surface elevation(m) Free surface elevation(m)

Fig. 13. Relationship between the free surface elevation and the peak pressures for the pressure sensor 6 on the bow.

10
Wave(m)

-10
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
0.02
P6(bar)

0.01

0
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
0.02
P5(bar)

0.01

1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
0.02
P4(bar)

0.01

0
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
0.02
P3(bar)

0.01

1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
Time(s)

Fig. 14. Measured pressure values on four pressure sensors with respect to the measured wave elevation for the condition Irregular 20.
 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383 379

10

Wave(m)
0

-10
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
0.02

P6(bar)
0.01

0
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
0.02
P5(bar)

0.01

0
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
0.02
P4(bar)

0.01

0
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
0.02
P3(bar)

0.01
0
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
Time(s)

Fig. 15. Measured pressure values on four pressure sensors with respect to the measured wave elevation for the condition Irregular 17.

Fig. 16. Probability of exceedance of pressure peaks in three irregular sea states.

free surface elevation at the entry point results in higher impact
pressure. Given that the pressure becomes higher with the
increase of the entry velocity, it is believed that the entry velocity
is higher for a lower free surface. It can be also connected with the
phase angle of the local wave surface. The closer the entry point to
a maximum of the wave elevation, the lower is the local vertical
velocity from the water.
The measured time series of the pressure values on the four
sensors in Section 2 are plotted in Figs. 14 and 15, together with the
associated wave surface elevation, respectively for the conditions of
Irregular 20 and Irregular 17. The general trends observed are that the
magnitude of the slamming pressures increases as the wave ampli-
tude increases, and the slamming is less frequent for sensor 3 and
most frequent for sensor 6. It is because the sensor 3 is the farthest
one away and sensor 6 is the closest one to the water surface. For
these two extreme sea states, the peak wave period is the same and
the steepness is very close, while the signicant wave heights differ
by about two metres in full scale. As seen in these two gures, the
difference of the frequencies of the slam events for the two sea states
is not apparent, but the frequencies of severe slam events with high
peaks are obviously higher for the sea state of Irregular 20. It is found
that a smaller number of slam events with high peak pressure occur
compared to that of the events with moderate and small peak value
for each conditions. These cases with high peak pressures are con-
sidered in present work.
Fig. 16 plots the probability of exceedance of the pressure peaks
for the values obtained from sensors 4 and 6. The results differ due
to the various characteristics of the sea states. For the sea state
Irregular 17, the signicant wave height is 8.7% and 16% lower than
that for Irregular 20 and Irregular 14 respectively. This results in
smaller values for the exceedance probability of peak pressures,
e.g. for pressure sensor 6 of Irregular 17, the exceedance prob-
ability of 0.01 bar pressures is about 23% and 33% lower than that
of the other two sea states. As seen in Fig. 16, the probabilities for
small and medium peaks from the case of Irregular 14 are bigger,
but the probabilities of high peaks are smaller than the values
from Irregular 20. As seen in Table 2, though the signicant wave
height from Irregular 14 is higher, the steepness and peak wave
frequency are smaller than those of Irregular 20. Under the iden-
tical sea state, the pressure peaks from sensor 4 are smaller than
the ones from sensor 6 due to the decay of the entry velocities.
Given that the wave characteristics of the condition of Irregular 20
give higher frequencies of the large pressure peaks, this sea state is
considered in this work for the slamming loads calculations.
Two slam events are chosen for the slamming loads calculation.
The rst case is for Section 2 during the time instants between
t1720 s and t1740 s, and the second one is for Section 1 during
the time instants between t1460 s and t 1480 s. As shown in
Fig. 17, the calculated wave elevations a , vertical motions of ship
hull (VM), relative vertical motions (RVM) and relative vertical
velocities (RVV) in the sea states condition of Irregular 20 are
380 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383

Exp. Numerical Exp. Numerical

a (m) 20 20

a (m)
0 0

-20 -20
1720 1725 1730 1735 1740 1460 1465 1470 1475 1480
20 20
VM(m)

VM(m)
0 0

-20 -20
1720 1725 1730 1735 1740 1460 1465 1470 1475 1480
20 20
RVM (m)

RVM (m)
0 0

-20 -20
1720 1725 1730 1735 1740 1460 1465 1470 1475 1480
10 10

RVV(m/s)
RVV (m/s)

0 0

-10 -10
1720 1725 1730 1735 1740 1460 1465 1470 1475 1480
Time(s) Time(s)

Fig. 17. Measured and calculated results of wave elevation, vertical motion of ship hull, relative vertical motion and relative vertical velocity in the sea states condition of
Irregular 20 and all the values are in full scale. (a) Case 1 for Section 2 between t 1720 s and t 1740 s; (b) Case 2 for Section 1 between t 1460 s and t 1480 s.

around t 1474 s, the numerical entry velocity is v  10 m/s.

4 These instantaneous relative velocities from numerical results are
used in the slamming loads calculations as a constant value during
Roll motion()

2 the water impact.

0 4.4. Roll motions in model tests

-2
As mentioned before, in the numerical procedure, the roll
-4 motion of the ship hull is not predicted; however, slamming loads
1720 1725 1730 1735 1740 are affected signicantly by it in some cases. Wang and Guedes
Time(s) Soares (2013) studied the effects of the roll angles on the slam-
ming pressure on the wetted surface of a bow-ared section. It
was found that, when the roll angle was larger, the pressures on
4 the downward side were higher, while those on the leeward side
were lower, and it was observed that the effects were less for a
Roll motion()

2 smaller roll angle. Hermundstad and Moan (2005) found that the
0
-2
-4
1460 1465 1470 1475
Time(s)
Fig. 18. Roll motions in the sea states condition of Irregular 20 for the two cases.
(a) Case 1 for Section 2 between t 1720 s and t 1740 s; (b) Case 2 for Section 1
between t 1460 s and t 1480 s.
compared with the measured ones and all the values are in full
scale. It is found that the peak values of the entry velocities gen-
erally coincide with the large values of ship's vertical motion and
the wave surface elevation. The numerical wave surface elevations
agree quite well with the measured data, and the calculated ver-
tical motions of the hull have some differences with the measured
ones, resulting in some deviations between the relative vertical
velocities. As seen, in the rst case, around t 1729.5 s, the
numerical entry velocity is v  6.45 m/s, and in the second case,
effects depend on the bow-are angle and the phasing between
the roll motion and the relative vertical motion between the sec-
tion and the water surface. As mentioned before, the numerical
amplitudes of the motions of the ship have satisfactory agreement
with the measured values.
1480 Here, the effects of the roll motion in model test are discussed.
Fig. 18 plots the experimental time series of roll motions of these
two cases. For the rst case, around t 1729.5 s, the roll motion is
about  1, and for the second case, around t 1474 s, the value is
about 1. A negative roll angle may reduce the slamming pressure,
while a positive one may increase it as seen in the coordinate
system in Fig. 1 and the positions of the pressure sensors in Fig. 3.
These two slamming events are simulated in model scale,
combining the section geometries in Fig. 10 with the calculated
entry velocities. As seen in Fig. 10, there is a very big bulb in
Section 2, on which the ow separates and may not reattach to the
section's surface, resulting in zero pressures on the pressure sen-
sors which are placed on the upper part of the section, for the
numerical simulation. This section is modied and denoted as
Section 2 modied. Hermundstad and Moan (2005) calculated
the slamming loads on three alternative simplied geometries.
 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383 381

0.02 0.02
sensor 6 sensor 5

0.01 0.01
Pressure (bar)

0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15

0.02 0.02
sensor 4 sensor 3 Exp.
0.01 0.01 Numerical

0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
Vertical displacement(m)

0.015
sensor 2
0.01

0.005 Exp.
Numerical
Pressure (bar)

0
0 0.05 0.1 0.15
0.015
sensor 1
0.01

0.005

0
0 0.05 0.1 0.15
Vertical displacement(m)
Fig. 19. Comparisons of the measured values and numerical calculations for the pressure values on the pressure sensors. (a) Case 1; (b) Case 2.

Their results showed that the calculated slamming pressure from
the Alternative 1 section is closer to the experimental data, and
the values from the other two alternatives are slightly smaller. The
bulbous bow of Section 2 in the present work is much bigger than
the one in the work of Hermundstad and Moan (2005), so it is not
proper to use a modied section like the Alternative 1. Taking the
computational time into consideration, Section 2 modied in
Fig. 10 is adopted in the calculation of Case 1, while the original
one of Section 1 is used for Case 2. Fig. 19 plots the experimental
and numerical results of the pressures on the two sections.
Experimental data include the entire process of entry and exit of
the slamming events, while only the entry stage is simulated in
the numerical calculations, so only parts of the experimental
results are presented here. The horizontal axis is plotted in terms
of the vertical displacement of the section, and zero point of Case
1 means the moment of sensor 6 impacting with water, and that of
Case 2 means the moment of the keel impacting with water. The
pressures are generally higher at the lower part of the section. The
reason is that the relative velocity is higher when the water sur-
face passes the lower part for the experiments, and the ow
separates from the section surface on upper part of the section
surface due to the curvature in the numerical calculations. It is also
observed that most of the pressures do not have a large peak with
small duration, but have a smaller magnitude with longer dura-
tion. It means that these two slamming events are mainly due to
the rate of change of the hydrodynamic momentum as the section
immerses into the water, which refers to the momentum slam-
ming. It can be explained as the local angles at the keels and the
ares of the sections are large. The bow are water impact pro-
blems have been studied by many researchers, and comparisons of
the impact loads between numerical calculations and measured
values have been presented, e.g. in Sun and Faltinsen (2009) and
Wang and Guedes Soares (2013). The measured values in their
work are obtained from model tests of a ship section segment
rather than an actual ship model which is used in present work.
For the experimental pressure on pressure sensor 6, the initial
peak may be due to the effects of the bulk of the section. As seen in
Fig. 19, the experimental pressures on 6, 5, 2 and the ones on 4 and
1 happen earlier, and the one of 3 occurs later than the numerical
calculations. The reason is that the measured relative velocity
increases slightly and then decreases, while the constant entry
velocity is applied in the numerical calculations. It is also found
that the numerical pressures are higher than the measured ones,
for the results of Case 1. It is reasonable due to the three-
dimensional effects and the negative roll motion of the ship in
model test. For the results of Case 2, though the numerical values
are still higher than measured ones, the differences are smaller. It
is mainly due to the positive roll motion in this slamming event. As
seen in Fig. 19(b), the numerical pressures reach peak values and
begin to decrease after the moment when the vertical displace-
ment is 0.1 m, when Section 1 is totally immersed in the water.
Though the roll motion is neglected and the slamming events are
simulated in two-dimensions for the numerical calculations, the
slamming pressures are predicted satisfactorily comparing with
the measured values.
382 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383
5. Conclusions
A numerical method is presented in this paper, to seek an
effective and acceptable method of predictions of the slamming
loads on ship hulls subjected to extreme sea states. The numerical
calculations are conducted in two steps. First, the ship motions in
irregular waves are calculated by a nonlinear numerical procedure
based on strip theory. Second, the obtained relative entry velo-
cities between the ship hull and wave are applied in an ALE
algorithm for predicting the slamming pressures during the water
entry stage. This method is validated by experimental data from
the model test of a 170 m chemical tanker in head seas at zero
speed condition. Three sea states with different characteristic
parameters are considered in the present work, and the numerical
calculations of the slamming loads are performed for two slam-
ming events in the condition of Irregular 20. Only heave and pitch
motion of the ship hull are considered in the numerical
calculations.
The numerical predictions of the heave, pitch motions and
wave surface elevations are in good agreement with the mea-
surements; however, slight differences of the peak values result in
relatively signicant differences on the probability of the slam-
ming occurrence on the ship hull. It is observed that even for the
ship hull without advancing speed, the slamming event occurs
frequently on the ship's bow.
The statistical study of the measured pressure shows that the
peak pressure increases as the instantaneous entry velocity
increases, and varies inversely with the free surface elevation at
the entry point. For the pressure sensor 6, it seems that the
pressure values are nearly linearly correlated with the entry
velocities. The general trends observed are that the magnitude of
slamming pressures increases as the wave amplitude increases,
and the slamming is less frequent for the upper part of the bow
are. It is also found that a smaller number of signicant slam
events occur compared to that of moderate and small events for
each condition. Two signicant slam events are studied for the
slamming loads calculation.
The predicted time histories of 6 pressure points are compared
with the experimental data. The results show that the numerical
calculations are higher than the measured one due to the three-
dimensional effects and roll motion in the model test. However, it
is believed that the positive roll motion in the test and the dif-
ferences of the measured and numerical entry velocity decrease
the differences of the slamming pressures. The present pressure
histories also justify that the are slamming force is mainly due to
the rate of change of the hydrodynamic momentum as the hull
immerses into water.
With statistical data of waves and the properties of a ship, this
research can be applied to assess ship motions in waves, slamming
probability and slamming loads on the ship hull, even the case
with forward speed. To achieve more accurate and effective pre-
diction of the slamming load, the nonlinear time domain code
needs to be improved on predicting the peak values of the ship
motions, and three-dimensional effects and roll motion should be
considered in the slamming loads calculation.
Acknowledgements
The work has been performed in the scope of the project
EXTREME SEAS Design for Ship Safety in Extreme Seas, (www.mar.
ist.utl.pt/extremeseas), which has been partially nanced by the EU
under contract SCP8-GA-2009-234175. The project partners Det
Norske Veritas AS, Germanischer Lloyd SE, Norwegian Meteorological
Institute, Institute of Applied Physics, Canal de Experiencias
Hidrodinamicas de El Pardo, Meyer Werft, Estaleiros Navais Viana de
Castelo, Technical University of Berlin and Universities of Bologna and
Duisburg-Essen have contributed to the project. In particular the
Technical University of Berlin has conducted the experimental work
that is being compared with the present numerical work. The rst
author is grateful to S. Rajendran, for his help in the use of the ship
motion program. The rst author has been funded by the Portuguese
Foundation for Science and Technology (FCT, Fundao para a Cincia
e a Tecnologia, Portugal) under the Contract no. SFRH/BD/98909
/2013.
References
Armand J.L., Cointe, R., 1987. Hydrodynamic impact analysis of a cylinder. In: Pro-
ceeding of the Fifth International symposium on Ocean, Offshore and Arctic
Engineering, Tokyo, Japan, Vol. 1, pp. 609634.
Aquelet, N., Souli, M., Olovsson, L., 2005. Eulerlagrange coupling with damping
effects:application to slamming problems. Comput. Methods Appl. Mech. Eng.
195, 110132.
Buchner, B., 1998. A new method for the prediction of non-linear relative wave
motion. In: Proceedings of the 17th International Conference on Offshore
Mechanics and Arctic Engineering, ASME, New York, paper OMAE98/SR-0592.
Buchner, B., 2002. Green Water on Ship-type Offshore Structures (Ph.D. thesis).
Delft University of Technology, Delft, Netherlands.
Clauss, G.F., Klein, M., Dudek, M., 2011. Model test report-Test Campaign I (TCI)-
Chemical Tanker (TUB), EXTREME SEAS project report (condential).
Dobrovol'skaya, Z.N., 1969. On some problems of similarity ow of uids with a free
surface. J. Fluid Mech. 36, 805829.
Faltinsen, O.M., 1993. Sea Loads on Ships and Offshore Structures Solutions. Cam-
bridge University Press, Cambridge, UK.
Fonseca, N., Antunes, E. and Guedes Soares, C., 2006. Whipping response of vessels
with large amplitude motions. In: Proceedings of the 25th International Con-
ference on Offshore Mechanics and Arctic Engineering, paper OMAE2006-
92412.
Fonseca, N., Guedes Soares, C., 1998a. Time domain analysis of large amplitude
vertical motions and wave loads. J. Ship Res. 42 (2), 100113.
Fonseca, N. and Guedes Soares, C., 1998b. Non-linear wave induced responses of
ships in irregular seas. In: Proceedings of the 17th International Conference on
Offshore Mechanics and Arctic Engineering(OMAE 98), ASME; Vol II.
Fonseca, N., Guedes Soares, C., 2004a. Experimental Investigation of the Nonlinear
Effects on the Vertical Motions and Loads of a Containership in Regular Waves.
J. Ship Res. 48 (2), 118147.
Fonseca, N., Guedes Soares, C., 2004b. Experimental Investigation of the Nonlinear
Effects on the Statistics of Vertical Motions and Loads of a Containership in
Irregular Waves. J. Ship Res. 48 (2), 148167.
Guedes Soares, C., 1989. Transient response of ship hulls to wave impact. Int.
Shipbuild. Prog. 36, 137156.
Guedes Soares, C., Pascoal, R., 2005. Experimental study of the probability dis-
tributions of green water on the bow of oating production platforms. J. Off-
shore Mech. Arct. Eng. 127 (3), 234242.
Guedes Soares, C., Pascoal, R., Anto, E.M., Voogt, A., Buchner, B., 2007. An approach
to calculate the probability of wave impact on an FPSO bow. J. Offshore Mech.
Arct. Eng. 129 (2), 7380.
Hermundstad, O.A., Moan, T., 2005. Numerical and experimental analysis of bow
are slamming on a RoRo vessel in regular oblique waves. J. Mar. Sci. Technol.
10, 105122.
Howison, S.D., Ochendon, J.R., Wilson, S.K., 1991. Incompressible water-entry pro-
blems at small deadrise angles. J. Fluid Mech. 222, 215230.
Jensen, J.J., 2001. Load and Global Response of Ships, Elsevier Ocean Engineering
Book Series, 4. Elsevier, The Netherlands.
Judge, c, Troesch, a, Perlin, m, 2004. Initial water impact of a wedge at vertical and
oblique angles. J. Eng. Math. 48, 279303.
Kapsenberg, G.K, Thornhill, E.T., 2010. A practical approach to ship slamming in
waves, 28th Symposium on Naval Hydrodynamics. Pasadena, California.
Ochi, M.K., Motter, L.E., 1973. Prediction of slamming characteristics and hull
response for ship design. Trans. SNAME 81, 144190.
Ochi, M.K., 1964. Prediction of occurrence and severity of ship slamming at sea. In
5th Symp. Naval Hydrodynamics, Bergen, Norway.
Psaraftis, H.N., 1978. Some new aspects of slamming probability theory. J. Ship Res.
22 (3), 186192.
Rajendran, S., Fonseca, N., Guedes Soares, C., 2015a. Effect of surge motion on the
vertical responses of ships in waves. Ocean Eng. 96, 125138.
Rajendran, S., Fonseca, N., Guedes Soares, C., 2015b. Simplied body nonlinear time
domain calculation of vertical ship motions and wave loads in large amplitude
waves. Ocean Eng. 107, 157177.
Senjanovic, I., Parunov, J., 2001. Slamming and whipping analysis of large container
ship. In: Proceedings of the Eleventh International Offshore and Polar Engi-
neering Conference, Stavanger, Norway, June, pp. 1722.
Stavovy, A.B., Chuang, S.L., 1976. Analytical determination of slamming pressures
for high speed vessels in waves. J. Ship Res. 20, 190198.
 S. Wang, C. Guedes Soares / Ocean Engineering 111 (2016) 369383
Sun, H., Faltinsen, O.M., 2009. Water entry of a bow are section with a roll angle. J.
Mar. Sci. Technol. 14, 6979.
Veen, D.J., Gourlay, T.P., 2012. A combined strip theory and smoothed particle
hydrodynamics approach for estimating slamming loads on a ship in headseas.
Ocean. Eng. 43, 6471.
von Krmn T., 1929. The impact on seaplane oats during landing. National
Advisory Committee for Aeronautics. Technical note No. 321, pp. 309313.
Wagner, H., 1932. Uber stossund gleitvergange an der oberache von ussigkeiten.
Z. Angew. Math. Mech. 12, 193215.
Wang, S., Guedes Soares, C., 2012. Analysis of the water impact of symmetric
wedges with a multi-material eulerian formulation. Int. J. Marit. Eng. 154 (Part
A4), 191206.
Wang, S., Guedes Soares, C., 2013. Slam-induced loads on bow-ared sections with
various roll angles. Ocean. Eng. 67, 4557.
Wang, S., Guedes Soares, C., 2014a. Numerical study on the water impact of 3D
bodies by an explicit nite element method. Ocean. Eng. 78, 7388.
383
Wang, S., Guedes Soares, C., 2014b. Experimental and numerical study on bottom
slamming probability of a chemical tanker subjected to irregular waves. In:
Guedes Soares, T.A., Santos, T.A. (Eds.), Maritime Technology and Engineering.
Taylor & Francis, London, pp. 10651072.
Zhao, R., Faltinsen, O.M., 1993. Water entry of two-dimensional bodies. J. Fluid
Mech. 246, 593612.
Zhao, R., Faltinsen, O.M., Aarsnes, J.V., 1996. Water entry of arbitrary two-
dimensional sections with and without ow separation. In: Proceedings of
the 21st Symposium on Naval Hydrodynamics, pp. 408423.
Zarnick, E.E., Hong, Y.S., 1986. Relative bow motion and frequency of slamming of
SWATH cross-structure. DTNSRDC/SPD Report 1174-01, David W. Taylor Naval
Ship Research and Development Center, Bethesda, Maryland.