Ocean Engineering 293 (2024) 116603

Contents lists available at ScienceDirect

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

Research paper

Experimental study on multi-body coupling system for twin-barge

hydraulic lifting shipwreck salvaging system
Wencheng Wu a, Mingyue Liu a, b, *, Yufeng Kou a, Longfei Xiao a, b, Ziyang Zhu a, Zhe Jiang c,
Yu Liu c
a
State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai, 200240, China
b
SJTU-Sanya Yazhou Bay Institute of Deepsea Science and Technology, Sanya, 572024, China
c
Shanghai Salvage Company, Shanghai, 200090, China

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

Handling Editor: Prof. A.I. Incecik The twin-barge lifting method, applied in the overall salvaging scheme of ‘SEWOL’, is one of the main ap­
proaches for salvaging large-tonnage shipwrecks. A model experiment for the twin-barge lifting shipwreck
Keywords: salvaging system was conducted for three typical salvaging stages. Based on the experimental results, this study
Twin-barge lifting shipwreck salvaging investigated the coupled dynamic response characteristics of the multi-body system consisting of the shipwreck
Multi-body coupling
and twin barges. The coupling effects between the multi-body system were discussed to understand the complex
Motion response
dynamic behaviour in twin-barge salvaging system, utilizing spectrum and statistical analysis. Additionally, this
Shielding effect
Sling tensions study considered the influence of the environment direction on the barge motion. It was revealed that the multi-
body system exhibits significant shielding, which greatly affects the heave, roll, and pitch motions, particularly
for beam and quarter seas. The maximum sling tension, including the case of single lifting sling broken, was also
analysed to ensure safety under extreme conditions. Finally, it is advisable to perform salvage operations in head
seas considering motion response amplitude, and to conduct the salvage in beam seas from the perspective of
lifting slings safety. This study is of great significance for improving the safety of large-tonnage wreck salvage
operations, promoting the development of research on multi-body system.

1. Introduction of the twin-barge salvaging system, it is essential to study its motion

response and interactions.
With the development of the shipping industry, the number of large- Three main processes must be considered in the twin-barge salvaging
tonnage ships has been increasing rapidly. Large-tonnage ships may of a shipwreck involving three different media (soil, water, and air): (1)
capsize and sink because of dangerous sea conditions, human errors, and off-bottom stage, (2) underwater lifting stage, and (3) partial salvage
wars (Rogowska et al., 2010; Velayudhan et al., 2019). Accordingly, stage (Xin et al., 2023). The twin-barge lifting salvaging system con­
shipwreck salvage operations play a crucial role in protecting waterways sisting of two barges and one shipwreck is a multi-body system under the
and the environment (Monfils et al., 2006; Henkel et al., 2014; Xin et al., influence of current-wind-wave. The reliability of this system is pri­
2023). The common methods of shipwreck salvaging include drainage, marily associated with the tensions of the lifting slings and coupled
salvage pontoons, and floating cranes (Navy, 2013). However, these motion responses. Because the shipwreck and barges are connected by
methods are limited to small-tonnage shipwrecks because of the poor slings, the relative motions among the three bodies are inevitably
load bearing capacity of a single salvage vessel. Therefore, it is necessary coupled with the sling tensions, which may cause large dynamic loads
to adopt two barges for salvaging to reduce the load on each barge. for the lifting system (Persaud et al., 2019).
Recently, Shanghai Salvage Company proposed a twin-barge hydraulic As the shipwreck salvage system is highly complex, certain as­
lifting technique for salvaging a large-tonnage shipwreck ‘SEWOL’ from sumptions and simplifications are adopted to obtain acceptable nu­
a 44 m deep seabed (Zhang et al., 2023). Nevertheless, it is difficult to merical simulation results. Jia et al. (Jia and Li, 2018) established a
accomplish a twin-barge salvaging operation under harsh current, wind, quasi-static model of a wreck overturning operation to calculate the
and wave conditions. Consequently, to ensure the safety and reliability cable tension under different conditions based on a time-domain

* Corresponding author. State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai, 200240, China.
E-mail address: [email protected] (M. Liu).

https://doi.org/10.1016/j.oceaneng.2023.116603
Received 20 August 2023; Received in revised form 13 December 2023; Accepted 16 December 2023
Available online 29 December 2023
0029-8018/© 2023 Elsevier Ltd. All rights reserved.
W. Wu et al. Ocean Engineering 293 (2024) 116603

analysis method. This study revealed that the maximum tension of ca­ when low-order and high-order boundary element methods are applied.
bles is strongly and linearly related to the cable-winded speed and The motion response of the multi-body system is larger than that of a
approximately linearly related to the mass of the ship. However, the single barge, indicating that the interaction between three barges is far
quasi-static model is limited to two-dimensional problems and cannot more complex than that of one barge. With the advancement of
solve the kinematic coupling effect in three-dimensional (3D) problems. computing technology, viscous flow theory has gradually been applied
A coupled numerical model of a twin-barge salvaging system based on to multi-body interaction (Chen et al., 2002). Harada et al. (2019)
the potential theory was proposed by Yao et al. (2018a) for obtaining the investigated the two-dimensional gap resonance between FLNG and
static distribution of cable tensions. To further study the dynamic shuttle tanker by employing a numerical method based on viscous flow
characteristics of cable tensions, Yao et al. (2018b) established a theory. Lu et al. (2011) demonstrated the mechanism of wave forces at
modified numerical model of a twin-barge lifting shipwreck salvaging gap resonance through the utilization of a viscous fluid model.
system based on a special multi-body coupling response calculation Furthermore, to explore the gap wave response and wave loading of
method. The simulation model can predict the dynamic tensions of ca­ FLNG-LNG interactions under various lateral spacings and wave fre­
bles, and particularly generate the shielding effect between barges. quencies, unsteady Reynolds-Averaged Navier Stokes (URANS) was
However, it is difficult to fully show the multi-body shielding phenom­ selected by Jin et al. (2018). Compared with the experiments, particu­
enon because the model treats one body as moving freely in space, while larly at higher wave frequencies, the predictions of gap wave resonance
setting the other bodies as globally fixed for calculation. achieved through viscous flow exhibit superior computational accuracy
Based on the potential flow theory, Zhang et al. (2016) performed than that of potential flow. Additionally, Duan et al. (2020) conducted a
time-domain calculations to study the coupled motion of two ships with systematic analysis of hydrodynamic interactions between two ships
lifting slings under the impact of winds, waves, and currents. This work navigating parallel to each other in shallow waters, employing the
highlights the importance of considering a multi-body system when Taylor Expansion Boundary Element Method (TEBEM). Compared with
studying the motion responses of a twin-barge salvage system. Never­ other computational methods, the TEBEM results are similar to those
theless, few studies have been conducted on multi-body systems in the obtained using the high-order boundary element method and better than
field of salvaging operations. Zhou et al. (2023) reported numerous those obtained using the constant panel method. Moreover, Ok et al.
studies on the hydrodynamic responses of other multi-body systems, (2017) presented three-dimensional numerical simulations for six de­
such as side-by-side lightering (floating liquefied natural gas system grees of freedom motions of a single vessel and side-by-side vessels.
(FLNG) and liquefied natural gas carrier (LNG)) (Zhao et al., 2018), Their analysis took into account the effects of viscous damping and
ship-to-ship encountering (Xu et al., 2017), and two ships overtaking vortex shedding, highlighting the advantages of these phenomena.
(Yuan et al., 2019). Xu et al. (2016) predicted the motion responses of There are many hypotheses in the numerical simulations, and the
three barges in close proximity to each other and analysed the hydro­ reliability of their results needs to be further verified. Therefore, for a
dynamic coupling effect between multi-body systems without consid­ multi-body system, the wave model testing can provide a relatively
ering mechanical interactions. Subsequently, Xu et al. (2012) indicated precise reflection of the strong nonlinear coupling effect (Sun and Li,
that the hydrodynamic interaction in multi-body systems is significant 2023). Several experimental studies have been carried out regarding

Fig. 1. Configuration of the twin-barge lifting shipwreck salvaging system.

2
W. Wu et al. Ocean Engineering 293 (2024) 116603

multi-body systems. Zhao et al. (2017) conducted an experimental converted to prototype value. The earth-fixed (right-handed) and body-
investigation into couple dynamic responses of a typical side-by-side fixed (right-handed) coordinate systems used in the model experiments
connection system for an FLNG and LNG carrier. The main aim of this are illustrated in Fig. 3. The direction of the Z-axis in the earth-fixed
research was to identify the load distribution characteristics within a coordinate system is vertically upward. The directions of the waves
classical multi-body system and the factors that affect it. Subsequently, and currents are opposite to the X-axis direction and remain constant
Bai et al. (2021a) carried out twin-barge floatover installation model relative to the earth-fixed coordinate system. The body-fixed coordinate
tests to analyse the hydrodynamic characteristics of the structure. They systems of barge A, barge B, and the shipwreck are identical. The origin
pointed out that the shielding effect and interaction between the twin is located at the initial centre of gravity. The positive x-axis points to­
barges had a significant impact on the motion amplitude of the system wards the bow of the ship. The positive y-axis points towards the port
and even on the phase difference. Moreover, Hamzah (2014) conducted
experiments on the hydrodynamic interactions between two floating
offshore structures in waves. Additionally, Vieira et al. (2018) carried
out model experiments to examine the offloading operations of FLNG
systems, with a specific focus on the analysis of the influence of free
surface resonance on side-by-side mooring systems. An experimental
investigation of the hydrodynamic response involved in topside transfer
operations from a T-shaped barge onto twin barges was presented by Bai
et al. (2021b). However, previous studies mainly focused on the problem
of collocation between two floating bodies. The prediction of dynamic
response of multi-floating structures is far from enough. Furthermore,
there have been relatively few model experiments conducted specif­
ically for salvaging systems related to twin-barge lifting. Therefore,
there is a compelling need for model experiments to be conducted in the
context of twin-barge lifting salvaging systems.
This study conducted model tests of the twin-barge lifting salvaging
system to analyse its motion response characteristics and coupled hy­
drodynamic interactions. The motion performance of the system,
particularly the shielding effect, was analysed and discussed based on
the response amplitude operator (RAO). Additionally, the amplitude
frequency characteristics of the salvaging system under various wave
conditions were obtained by analysing the amplitude and power spectral
density (PSD) of the motion response. Finally, the maximum tension of
the lifting slings and the risk analysis of sling breakage were studied.

2. Experimental setup and conditions

2.1. Experiment setup

For the model test of salvaging a 1000 TEU container ship at a water
depth of 300 m, a twin-barge lifting shipwreck salvaging system for a
large-tonnage wreck was designed, as shown in Fig. 1. The process of
lifting the shipwreck is divided into three typical stages: initial lifting
stage (the shipwreck just leaves the seabed at 300 m deep water, Fig. 1
(a)), middle lifting stage (the shipwreck is lifted to a depth of 150 m),
and final lifting stage (the shipwreck partially emerges with a draft of
5.9 m, Fig. 1(b)). It is worth noting that, in the initial lifting stage, the
shipwreck is situated 2 m above the seabed (0.04 m in model scale),
taking into account the motion of the ship. The two barges are arranged
in parallel. The lifting slings are hung from one side of the barge and
connected to the shipwreck through the bottom cables. Two hydraulic
lifting systems are arranged horizontally on the main deck of the twin
barges, working synchronously to lift the shipwreck as a whole. To
compensate for the severe barge motion caused by rough seas, a hy­
draulic hoisting and buffering system with a compensation function is
arranged at the end of each lifting sling to maintain the stability of the
salvaging system. The model arrangement for each stage is depicted in
Fig. 2.
The experiments were conducted in the Deepwater Offshore Basin at
Shanghai Jiao Tong University. The wave basin is 50 m long, 40 m wide,
and 10 m deep (maximum effective depth). The two sides of the basin
are equipped with multi-flap wavemakers to generate various types of
waves. On opposite sides of the wavemakers, wave-absorbing beaches
are installed to minimize waves reflected from the boundaries of the
basin. The properties of the wreck salvage system, hoist buffer systems
and mooring system were modelled using Froude similarity with a
scaling ratio of 1:50. The data presented in this paper has all been Fig. 2. Model arrangement for the three typical stages in the basin.

3
W. Wu et al. Ocean Engineering 293 (2024) 116603

Fig. 3. Coordinate system definition of the salvaging system.

Table 1
Main parameters of the barge and wreck.
Parameter Symbol Unit Barge Wreck

Overall length LOA m 169.00 139.10

Breadth B m 39.80 22.60 Fig. 4. Mooring system.
Depth D m 10.90 11.80
Draft T m 7.50 8.00 intentionally offset when the barges are onshore.
The two barges are fixed with eight mooring lines each, as displayed
side, and the direction of the z-axis is vertically upward. in Fig. 4. Table 4 lists the properties of the mooring lines, where Wdry is
The main parameters of the barge and wreck are listed in Table 1. the weight in air and EA is the axial stiffness. The mooring lines,
Two static states of the shipwreck were considered in the experiment: a) particularly the anchor chain, are purposely designed to be truncated
submerged in water and b) partially emerged (draft of 5.9 m). Tables 2 horizontally because of the excessive horizontal span of the mooring
and 3 present the gravity and buoyancy parameters of the shipwreck and system. The truncated mooring system is designed to match the proto­
barges, respectively. To ensure the barges maintain an upright floating type restoration curves. The coordinates of the anchor points and fair­
condition during the experiment after connecting with the shipwreck via lead of the truncated mooring system are listed in Tables 5 and 6,
lifting slings, the designed centre of gravity and buoyancy are respectively, where the origin is located at the midpoint of the waterline
of the barge. As shown in Fig. 5, the measured horizontal stiffness of the

Table 2
Gravity and buoyancy parameters of shipwreck.
Stage Parameter Value (t) Centre of gravity/buoyancy (m) Radius of gyration (m)

x y z Rxx Ryy Rzz

Submerged in water Gravity 9607.66 61.30 0.03 7.07 8.36 34.65 36.74
Buoyancy 1240.99 59.35 0.03 9.29 / / /
Partially emerged Gravity 10038.88 60.68 0.03 7.36 8.26 33.49 35.59
Buoyancy 305.00 57.54 0.02 2.95 / / /

Table 3
Gravity and buoyancy parameters of the barges.
Stage Barge Parameter Value (t) Centre of gravity/buoyancy (m) Radius of gyration (m)

x y z Rxx Ryy Rzz

Submerged in water A Gravity 36875.63 79.83 2.44 7.43 11.57 42.73 43.08
Buoyancy 41150.01 78.80 0.02 3.93 / / /
B Gravity 36898.76 79.82 − 2.43 7.42 11.58 42.73 43.09
Buoyancy 41151.57 78.80 − 0.03 3.93 / / /
Partially emerged A Gravity 36184.77 80.13 2.90 7.52 11.46 42.71 43.02
Buoyancy 41169.71 78.79 0.03 3.93 / / /
B Gravity 36219.71 80.11 − 2.88 7.51 11.47 42.69 43.02
Buoyancy 41170.03 78.80 − 0.04 3.93 / / /

4
W. Wu et al. Ocean Engineering 293 (2024) 116603

Table 4
Properties of the mooring line.
Segment Length (m) Diameter (mm) Wdry (kg /m) Wsub (kg /m) EA (MN) Pretension (t)

Prototype Steel cable 1350.00 56 13.10 11.00 146 30.00

Anchor chain 200.00 76 126.00 118.00 124 /
Truncated Steel cable 548.88 / 16.00 15.92 117 30.76
Anchor chain 200.00 / 53.93 53.85 24 /

Table 5 barge A and SB1-SB8 in barge B, are arranged from the bow to the stern,
Coordinates of the anchor point of the truncated mooring system. respectively. In the model experiment, the lifting sling can be replaced in
some cases by a wire rope and an equivalent linearly elastic spring, with
Line no. Barge A Barge B
sufficient accuracy for practical use (Tikhonov et al., 1997), whose
X (m) Y (m) Z (m) X (m) Y (m) Z (m) properties are given in Table 7.
1 750.50 18.95 − 300.00 750.50 − 18.95 − 300.00 The hydraulic hoisting force buffer system operates similarly to the
2 412.79 631.35 − 300.00 412.79 − 631.35 − 300.00 case when a nonlinear spring is added to one end of the lifting sling.
3 71.73 724.41 300.00 71.73 − 724.41 300.00
Fig. 7 illustrates the relationship between the lifting force and cylinder
− −
4 − 84.60 727.22 − 300.00 − 84.60 − 727.22 − 300.00
5 − 422.89 630.54 − 300.00 − 422.89 − 630.54 − 300.00 travel of the buffer system. Movement of the plunger commences under
6 − 760.60 19.76 − 300.00 − 760.60 − 19.76 − 300.00 a force of 450.16 t, reaching maximum stroke at 806.88 t, resulting in a
7 − 669.56 − 324.13 − 300.00 − 669.56 324.13 − 300.00 maximum stroke length of 1.1 m. The lifting sling exhibits significantly
8 656.69 − 321.32 300.00 656.69 321.32 300.00
− −
high inherent stiffness, whereas the hydraulic lifting buffer system
demonstrates notably lower stiffness. Consequently, the lifting buffer
system typically presents a ‘strong-weak-strong’ stiffness characteristic,
contrasting with the often-simulated enhanced nonlinear stiffness in
Table 6
laboratories. In the test, a mechanical mechanism is employed to
Coordinates of the fairlead of the mooring system.
replicate the lifting buffer system, as depicted in Fig. 6(b). The buffer
Line no. Barge A Barge B cylinder’s stiffness is emulated using multiple springs, and the initial
X (m) Y (m) Z (m) X (m) Y (m) Z (m) tension of the spring is adjusted to simulate the starting tension
1 75.00 − 13.67 3.40 75.00 13.67 3.40
threshold of the buffer cylinder. Subsequently, the stroke bolt is used to
2 75.00 13.67 3.40 75.00 − 13.67 3.40 simulate the effective stroke of the buffer cylinder. The linear approxi­
3 71.73 16.21 3.40 71.73 − 16.21 3.40 mate buffer system with a corresponding prototype stiffness of 324.30 t/
4 − 84.60 19.02 3.40 − 84.60 − 19.02 3.40 m is presented in Fig. 7. The deviation of the stiffness simulation is
5 − 85.10 12.86 3.40 − 85.10 − 12.86 3.40
deemed acceptable, given that the purpose of simulating the buffer
6 − 85.10 − 12.86 3.40 − 85.10 12.86 3.40
7 − 84.60 − 19.02 3.40 − 84.60 19.02 3.40 system is to determine whether it will exceed the buffer limit during the
8 71.73 − 16.21 3.40 71.73 16.21 3.40 experiment.
The configuration of the measurement system is illustrated in Fig. 8.
During the experiment, tension force transducers were mounted at the
mooring system is in good agreement with the design value. top of lifting slings to measure the tensions. The 3 sets of optical motion
There are 16 sets of hoist buffer systems arranged at two barges as measurement system were applied to measure the six degrees of freedom
shown in Fig. 6(a) and Fig. 8, which include slings, lifting devices, and motions of the barges and shipwreck. When the shipwreck was under­
hydraulic hoisting force buffer systems. The 16 sets of lifting slings are water, two underwater inclinometers were installed to measure the
precisely lowered into the water for a distance of 1.5 m from the ship’s pitch and roll motions of the wreck. The sampling frequency of the
side. They establish a connection with the lifting devices on the floating measurement system is 40Hz.
barge by passing beneath the submerged wreck. The slings, P1–P8 in

Fig. 5. Test results of horizontal stiffness in the X- and Y-directions.

5
W. Wu et al. Ocean Engineering 293 (2024) 116603

Fig. 6. Layout of 16 sets of hoist buffer systems.

Table 7
Parameters of the sling (model).
Stage Length Equivalent Weight Axial Ultimate
(m) diameter (g/m) strength strength
(mm) (kg/cm) (kg)

Initial lifting 6.35 2.40 120.60 3.10 13.40

Middle lifting 3.35 2.40 120.60 5.90 13.40
Final lifting 0.47 2.40 120.60 42.00 13.40

Fig. 8. Arrangement of the measurement system and buffer system.

Table 8
Parameters of environmental conditions.
Hs Tp (s) Hmax γ Vc (m/ Vw (m/
(m) (m) s) s)

White noise 1.0 5–20 / / / /

Current-wave-wind 1.1 9 2 3.3 0.77 10.80
(CWW)

observed deviations of the trapezoidal spectrum in Fig. 9 are acceptable

as long as there are wave energies in a wide range of frequencies,
Fig. 7. Stiffness of hydraulic hoisting force buffer system. because the main purpose is to obtain motion RAOs. During wind speed
calibration, wind speeds were collected over a 1-min duration, and their
2.2. Test conditions average value was calculated. This process was repeated three times,
with the average of these measurements reaching the target wind speed
A series of model tests were conducted to investigate the motion (10.80 m/s in prototype), confirming the successful simulation of wind.
response characteristics and coupled hydrodynamic interactions of the The surface current velocity measurement results are presented in
twin-barge salvaging system. The JONSWAP spectrum was used to Table 9. The surface velocity met the specified velocity in the relevant
simulate the wave condition. The designed operating environmental marine environmental conditions. The turbulence level (I) was calcu­
conditions are presented in Table 8, where Hs is the significant wave lated by I = Vstd /Vmean = 9.2% < 10%, indicating an ideal turbulence
height, Tp is the peak period, Hmax is the maximum wave height, γ is the level for the simulated current.
spectral peak factor, Vc is the velocity of current, and Vw is the velocity of A primary source of uncertainty in the presented model test stems
wind. Prior to tests, the wave spectra, the velocities of wind and current from the current-wind-wave. Although the statistical values of wind and
were calibrated. Figs. 9 and 10 depict the results of the wave calibration. current velocity in the experiment can be maintained relatively consis­
The trapezoidal spectrum employed in the white noise test was without tently, the time series data cannot achieve complete uniformity.
current. The JONSWAP spectra were calibrated with current, following Therefore, the recordings were started 1.5 min after the wave-making
the specifications outlined in Table 8, ensuring the alignment of the command was given, allowing the model to attain a stable state of
wave spectra with the specified targets under the influence of current. motion. The recording duration for each wave realization is more than
The JONSWAP spectra are in good agreement with the target. The 8.5 min (1 h in prototype) to mitigate uncertainty. Furthermore, the

6
W. Wu et al. Ocean Engineering 293 (2024) 116603

Table 10
Test matrix.
Test Wave Environment Remark
no. direction

A01 180◦ White noise Initial lifting stage (the

A02 180◦ CWW shipwreck just leaves the
A03 135◦ White noise seabed at 300 m deep water)
A04 135◦ CWW
A05 90◦ White noise
A06 90◦ CWW
A07 90◦ CWW, single sling failure
with the highest maximum
tension
B01 180◦ White noise Middle lifting stage (the
B02 180◦ CWW shipwreck is lifted to a depth
B03 135◦ White noise of 150 m)
B04 135◦ CWW
Fig. 9. Calibration of white noise wave spectrum (without current). B05 90◦ White noise
B06 90◦ CWW
B07 90◦ CWW, single sling failure
with the highest maximum
tension
C01 180◦ White noise Final lifting stage (the
C02 180◦ CWW shipwreck is partially
C03 135◦ White noise emerged with a draft of 5.9 m)
C04 135◦ CWW
C05 90◦ White noise
C06 90◦ CWW
C07 90◦ CWW, single sling failure
with the highest maximum
tension

range of 0.20 rad/s to 1.28 rad/s. The surge RAOs of barges A and B
decrease with increasing incident wave frequency. The surge RAO curve
reaches its maximum in head seas, and the largest value of the sway RAO
curves occurs in beam seas, because the change in environment direc­
tion causes variations in the longitudinal and lateral wave forces on the
Fig. 10. Calibration of irregular wave spectrum (with current).
ship. When the shipwreck is submerged, sub-peaks appear in the sway
RAO curves of barges A and B for head and quarter seas, indicating the
occurrence of another resonance phenomenon. Combining the yaw and
Table 9 sway RAOs under the same conditions, the resonant frequencies were
Calibration of current velocity. found to be equal. Therefore, the coupling between sway and yaw results
Target (m/s) Measured (m/s)
in the occurrence of a second resonance phenomenon of the sway
motion.
Vmax Vmin Vmean Vstd
For a multi-body system consisting of the wreck and twin barges, the
0.77 0.94 0.63 0.76 0.07 shielding effect of the salvaging system is significant, causing the heave
response amplitude of barge A to decrease rapidly for the beam sea, as
presented in Fig. 12. This phenomenon is evident in short-period waves
time domain statistical analysis of all the measured data of the irregular
but is negligible in long-period waves because of their strong diffraction
wave test condition is also carried out to reduce uncertainty.
ability. In beam seas, the heave RAO corresponding to natural frequency
The water depth was set to 6 m, which corresponded to a water depth
of barge B is higher than that in other waves. This is because the crest
of 300 m in the prototype. Table 10 presents the test matrix of the model
line of the incident wave is parallel to the length of the barge, which has
tests, including the three typical stages of the salvaging process, three
the most evident wave-induced lifting force on the ship.
wave directions, and two environmental conditions. The wind, wave,
The position of the shipwreck significantly affects the roll natural
and current are collinear. Additionally, the failure of a single sling with
frequency of the barge. As the wreck rises, the slings connecting the
the highest maximum tension in beam seas (90◦ ) is considered.
barge and wreck become shorter and the restoring moment becomes
stronger. This leads to an increase in the barge roll natural frequency.
3. Motion response and lifting tension analysis of the salvaging
Until the wreck is partially out of the water, the roll natural frequency of
system
the barge becomes equal to that of the wreck (approximately 0.63 rad/
s). When the shipwreck is submerged, with the change in the environ­
3.1. Motion response amplitude operators
ment direction from 180◦ to 90◦ , the roll RAOs corresponding to natural
frequency of barges A and B change from being nearly identical to being
The motion RAOs are obtained by transforming the motion responses
substantially different due to the shielding effect of barge B. The closer
of white noise test in the time domain, based on FFT. Fig. 11 displays the
the wreck to the free surface, the more evident the shielding effect.
surge, sway, heave, roll, pitch, and yaw RAO curves of the two barges
Fig. 13 presents the time history of heave motion of barge B and the sling
and wreck in the three lifting stages under head (180◦ ), quarter (135◦ ),
tension of SB8 in case A02. Notably, certain nearly simultaneous peaks
and beam (90◦ ) seas. Since the wave spectrum employed in the white
are marked on both curves with pentagram stars. It can be observed that
noise experiment takes the form of a trapezoidal spectrum, its frequency
the sling tensions change with the heave motion of the barge, resulting
range is inherently broader than the target (0.31–1.26 rad/s). Conse­
in large roll excitation moment. Therefore, the roll RAOs corresponding
quently, calculations of the RAO were conducted within the frequency

7
W. Wu et al. Ocean Engineering 293 (2024) 116603

Fig. 11. Motion RAO curves of barges A and B (A01SurgeA means the surge motion of barge A in case A01).

8
W. Wu et al. Ocean Engineering 293 (2024) 116603

to natural frequency of barges A and B are still significant even when

encountering head seas. When the shipwreck is partially emerged, the
change in the environment direction greatly affects the roll motion of
the barges. The maximum value of the roll RAO curves occurs in beam
seas, and the peak values in head and quarter seas are substantially
reduced to approximately 1/5 of that in beam seas. In head seas, the
heave motions of the barge and shipwreck are in phase, and their rela­
tive motions are small, which reduce the roll excitation moments
induced by the sling tensions. Moreover, the motion of barge B slows
down owing to the wave reflection, leading to a smaller roll motion
amplitude of the barge. Additionally, owing to the gap resonance phe­
nomenon in the multi-body system, barges A and B exhibit multiple
Fig. 12. A snapshot about shielding effect of barge B to barge A in white peaks of pitch motion RAO curves. Because of the symmetry of the
noise test. salvaging system, the yaw moment can be small in beam and head seas.
Therefore, the yaw response is low for environment at 90◦ and 180◦ .
As indicated in Fig. 14, the roll RAO corresponding to natural fre­
quency of the shipwreck is the smallest for head seas and exhibits a sharp
increase at 135◦ and 90◦ , particularly in the final lifting stage. Similar to
that of the lifting barge, the roll RAO of the shipwreck is sensitive to the
environment direction and increases rapidly as the environment direc­
tion changes from 180◦ to 90◦ .

3.2. Motion response characteristics in sea conditions

During salvaging, the heave, roll, and pitch motions of the barge and
shipwreck have considerable effects on the instantaneous tensions of the
lifting slings. To successfully conduct a salvaging operation, it is
necessary to investigate the motion response of the salvaging system
under the influence of the current, wave, and wind (CWW). In the data
processing stage, the time domain statistical analysis of all the mea­
Fig. 13. The time history of heave motion of barge B and the sling tension of surement data of the CWW test condition is carried out, and the statis­
SB8 in case A02. tical values of each measurement signal are obtained. Based on the
experimental results, the maximum motion values of the twin barges

Fig. 14. Roll motion RAO curves of shipwreck.

Table 11
Maximum value of the roll, pitch, and heave motions of the twin barges and shipwreck.
Wave Barge A Barge B Shipwreck
direction
Initial Middle Final lifting Initial Middle Final lifting Initial Middle Final lifting
lifting stage lifting stage stage lifting stage lifting stage stage lifting stage lifting stage stage

Roll (◦ ) 180◦ 0.60 0.72 1.45 0.71 0.61 1.23 0.54 0.56 0.71
135◦ 0.92 0.83 1.68 1.20 1.53 1.41 1.53 1.73 2.64
90◦ 2.08 1.46 3.99 2.24 2.41 5.49 2.18 1.53 6.13
Pitch (◦ ) 180◦ 0.41 0.55 0.66 0.47 0.53 0.57 0.86 0.88 0.54
135◦ 0.66 0.59 0.71 0.87 0.48 0.48 0.70 0.64 0.83
90◦ 0.17 0.13 0.16 0.17 0.20 0.22 0.28 0.29 0.17
Heave (m) 180◦ 0.29 0.29 0.42 0.36 0.36 0.35 – – 0.37
135◦ 0.34 0.55 0.53 0.43 0.41 0.45 – – 0.62
90◦ 0.92 1.25 0.69 0.94 0.98 1.43 – – 1.36

9
W. Wu et al. Ocean Engineering 293 (2024) 116603

and the shipwreck under various test conditions are listed in Table 11. The motion response spectra in the frequency domain were calcu­
When the shipwreck is submerged, the shielding effect mainly acts lated based on the fast Fourier transform. Fig. 15 shows the PSDs of the
on the roll amplitudes of the two barges in the CWW environment at motion responses (i.e., sway, surge, and yaw) of the twin barges and
135◦ and 90◦ . As the wreck gradually rises, the ratio of the roll ampli­ wreck at 90◦ in the final lifting stage, which exhibit similar PSD distri­
tude of barge B to that of barge A increases in beam seas because of the butions in all cases. The low-frequency motions are dominated by the
increasing influence of the shipwreck on the barges. Thus, the shielding sway, surge, and yaw motions, which are associated with the mooring
effect becomes more significant. In the final lifting stage, a multi-floating lines. The results indicate that the low-frequency motions have consid­
body system with very small gaps between floating structures is formed. erable effects on the global response, whereas the wave-frequency re­
Therefore, the shielding effect stands out in the roll and heave motions, sponses have negligible effects on the surge, sway, and yaw motions.
that is, the roll amplitude of barge B is 1.38 times that of barge A, Owing to the significant motion amplitude in beam seas, the PSD
whereas the heave amplitude is 2.07 times that of barge A in beam seas. curves of the roll, pitch, and heave of the barges and wreck for beam seas
In this stage, the roll motion amplitudes of the shipwreck at 135◦ and were examined. As can be observed in Figs. 16–18, two energy peaks
90◦ are much greater than those of the twin barges owing to their small exist in the PSD of the barge heave. The first peak is excited by the
roll inertia moment. This indicates that there is a strong coupling be­ natural frequency of the barge, while the second peak is induced by the
tween the roll and heave of the barge owing to their synchronous wave excitation. This indicates that both the wave and natural fre­
changes. quencies contribute to the heave motion of the barge. The PSD ampli­
By comparing the motion amplitudes of the system at different tude of barge B at the wave frequency is much greater than that of barge
salvaging stages, it can be observed that the maximum roll motion of the A in roll and pitch, signifying a strong shielding effect. After the ship­
barges and wreck in the final stage is approximately twice that when the wreck is partially lifted out of the water, the PSD amplitude of its roll
wreck is submerged. This may be because when the wreck is partially motion increases abruptly because the wave frequency is very close to
emerged, the roll natural frequency of the barges and shipwreck the roll natural frequency of the wreck. Therefore, it can be conjectured
(approximately 0.63 rad/s) is close to the peak period of irregular waves that the multi-body system shielding has a large impact on the heave,
(0.70 rad/s). The resonance phenomenon is evident, leading to an in­ roll, and pitch motions but little influence on the surge, sway, and yaw
crease in the motion amplitude. However, when the wreck is submerged, motions.
the roll natural frequency of the barges is less than 0.37 rad/s, which is
considerably different from the wave frequency. Additionally, Table 11 3.3. Lifting sling tensions
indicates that the roll, pitch, and heave motion amplitudes of the
salvaging system are all small in head seas. Therefore, it is recommended Salvaging operations are mostly conducted in open seas where the
that the twin-barge lifting shipwreck salvaging system should operate in sea conditions are complex and constantly changing. Therefore, deter­
head seas to reduce the motion responses of the wreck and barges. mining whether the sling tension will exceed the permissible value is a

Fig. 15. PSDs of surge, sway, and yaw motion responses (case C06).

Fig. 16. PSD of heave motion (90◦ ).

10
W. Wu et al. Ocean Engineering 293 (2024) 116603

Fig. 17. PSD of roll motion (90◦ ).

key point in the design of salvaging systems. in beam seas, the maximum tension of the same sling is less than that in
Barge B directly interacts with the quarter and beam seas, compared head and quarter seas. The centre of the plots in Fig. 19 leans towards
with barge A, resulting in a larger motion amplitude. Moreover, the the upper-left corner, indicating that the maximum tensions of the slings
tensions of the slings of barge B are typically higher than those of barge in the bow and stern are larger than that in the midship. This is because
A. Thus, it is important to analyse the maximum tension of the slings of the external forces acting on the shipwreck are mainly resisted by the
barge B. As illustrated in Fig. 6(a), slings SB1–SB8 are arranged from the slings at the bow and stern (Fz.SB1 and Fz. SB8). However, when the
bow to the stern. When the wreck is submerged, the maximum tension of wreck is emerged, the lateral motion of the wreck is smaller and the
the lifting slings exhibits similar distributions under different environ­ constraining force is reduced owing to the shielding effect. In this case,
ment directions, as displayed in Fig. 19. Here, Fz. SB1–Fz.SB8 are the the slings in the midship play a more critical role in lifting the wreck.
loads acting on the lifting slings on barge B. It is worth mentioning that Owing to the decrease in the buoyancy of the shipwreck, the maximum

Fig. 18. PSD of pitch motion (90◦ ).

Fig. 19. Maximum tension of each lifting sling on barge B.

11
W. Wu et al. Ocean Engineering 293 (2024) 116603

uniformly distributed and usually small during operation because the

lifting slings are perpendicular to the environment direction. Therefore,
from the perspective of the maximum tension of slings, it is recom­
mended to perform salvaging in beam seas. Although the motion
amplitude of the barge is larger under this condition, it is beneficial for
reducing the sling tension and decreasing the risk of sling failure.
In salvaging operations, there is a relatively high probability of sling-
breakage accidents, which can also result in severe consequences.
Therefore, it is crucial to focus on preventing such accidents. In this
study, the case of a single broken sling was analysed to provide guidance
for the development of safety measures in practical salvaging
engineering.
In each stage of shipwreck salvage operation, the lifting sling bearing
the highest load for each barge in beam seas is set to be broken.
Fig. 20. The statistics of tension for sling SB1-SB8 in case C04 and C06. Consequently, the slings located at the bow (i.e., P1 and SB1) for the
initial and middle lifting stages, and the slings positioned at the stern (i.
tension of the slings in the midship is higher than that when the ship is e., P8 and SB8) for the final lifting stage are marked as broken. The
completely submerged. In the final lifting stage, sling SB3 located in the results of the maximum tension of the remaining slings, along with the
midship is subjected to a tension of over 1000 t, which is dangerous and maximum tension of slings in beam seas without sling breakage, are
should be considered seriously. shown in Fig. 21. Here, Fz. P1–Fz.P8 represent the loads acting on the
From Fig. 19, it is evident that, during the three lifting stages, the lifting slings on barge A. The weight of the shipwreck is redistributed to
maximum tensions experienced by the lifting slings in quarter seas the other slings. It can be observed that the maximum tension of the
(135◦ ) are greater than those in head and beam seas. The maximum sling closest to the broken sling increases significantly, whereas the
value exceeds the effective range of the buffer system (806.88 t). maximum tension of the sling away from the broken sling shows an
However, the maximum tension of the slings is only 87.2% of the obvious reduction. In addition, the number of slings with increased
effective range of the buffer system for beam seas in all three stages, maximum tension is greater than that with reduced tension. The
whereas in head seas, the maximum tension can reach 98.9% of the maximum tension accounts for 79.8% of the sling’s ultimate strength
effective range. From the statistics of tension for sling SB1-SB8 in case (1717 t) in this case.
C04 and C06, shown in Fig. 20, the sling tensions in beam seas are In three lifting stages, the yaw motion experiences a significant

Fig. 21. Maximum tension of slings with and without sling breakage.

Fig. 22. Statistical value of the roll and yaw motions of barge B with and without sling breakage in beam seas.

12
W. Wu et al. Ocean Engineering 293 (2024) 116603

increase due to the breaking of the sling, with the maximum value of lifting shipwreck salvaging system. Further research is imperative to
yaw motion response reaching double the level before sling breakage, as investigate different salvage schemes, thus providing a more compre­
indicated in Fig. 22. When the bow slings break, the load distribution hensive understanding of salvage planning.
among the slings shows a pronounced shift towards the bow, resulting in
an uneven distribution of forces between the bow and stern of the CRediT authorship contribution statement
barges. This imbalance leads to an increased amplitude in yaw motion.
Although there is a redistribution of load among the remaining slings Wencheng Wu: Writing – review & editing, Writing – original draft,
after sling breakage, the impact on the total roll excitation moment of Visualization, Methodology, Investigation, Data curation, Conceptuali­
the barges is relatively minor. Therefore, in terms of roll motion, the zation. Mingyue Liu: Writing – review & editing, Methodology, Inves­
influence of sling breakage is limited. tigation, Conceptualization. Yufeng Kou: Writing – review & editing,
When considering the case of sling breakage, it is advisable to pro­ Methodology, Investigation, Data curation, Conceptualization. Longfei
actively identify the sling with the highest maximum tension. Xiao: Supervision, Project administration, Methodology, Funding
Strengthening measures should be taken for that particular sling, and the acquisition, Conceptualization. Ziyang Zhu: Methodology, Investiga­
load-bearing capacity of slings in close proximity to the most vulnerable tion, Data curation. Zhe Jiang: Methodology, Investigation. Yu Liu:
sling should be enhanced to ensure the safety of the wreck salvage Methodology, Investigation.
system in the event of a group of lifting slings breaking.
Declaration of competing interest
4. Conclusion
The authors declare that they have no known competing financial
This paper presents an experimental investigation on the hydrody­ interests or personal relationships that could have appeared to influence
namic behaviour of a twin-barge salvaging system. Many factors affect the work reported in this paper.
the complex motion of the twin-barge salvaging system, including
coupled motions and the shielding effect. The spectrum and statistical Data availability
analysis of the motion response in white noise and irregular waves were
performed. The hydrodynamic characteristics of the multi-body The data that has been used is confidential.
coupling of the salvaging system were analysed for three typical stages
during the shipwreck salvaging process. Additionally, the maximum Acknowledgments
tension of slings was investigated. The following conclusions were
drawn. The authors would like to acknowledge the support of the National
The motion RAOs of the two barges and wreck indicate that shielding Natural Science Foundation of China (Grant No. 52031006), the
has a large impact on the heave, roll, and pitch responses but little in­ Important National Science and Technology Specific Projects (Grant No.
fluence on the surge, sway, and yaw motions. The results of the PSD 2018YFC0309704). The authors are also grateful to the reviewers for
analysis reveal that the low-frequency response is dominant in the surge, their constructive comments and suggestions, which contributed to
sway, and pitch motions, whereas the influence of the wave-frequency improving the manuscript.
response is small. Additionally, the position of the shipwreck and ten­
sions of the slings have coupling effects on the motion of the twin barges. References
When the shipwreck is lifted from the bottom to the surface, the roll
natural frequency of the barge gradually increases and becomes Bai, X., Luo, H., Xie, P., 2021a. Experimental analysis of hydrodynamic response of the
twin-barge floatover installation for mega topsides. Ocean Eng. 219, 108302.
consistent with that of the shipwreck (approximately 0.63 rad/s). Bai, X., Luo, H., Xie, P., 2021b. Experimental investigation on hydrodynamic
Moreover, the change in sling tensions with barges heave motion leads performance of side-by-side three barges in topsides floatover installation in beam
to significant roll excitation moments, resulting in significant roll mo­ seas. Ocean Eng. 236, 109516.
Chen, H.C., Lin, W.M., Hwang, W.Y., 2002. Validation and Application of Chimera RANS
tion response for barges even in head seas. Method for Ship-Ship Interactions in Shallow Water and Restricted waterway[C]//
For the three stages, the maximum tensions of the slings in the 24th ONR Symposium on Naval Hydrodynamics. Fukuoka, Japan.
quarter seas are the highest, whereas those in beam seas are usually Duan, W.Y., Liu, J.Y., Chen, J.K., et al., 2020. Comparison research of ship-to-ship
hydrodynamic interaction in restricted water between TEBEM and other
smaller and uniformly distributed. When the slings with the highest
computational method. Ocean Eng. 202, 107168.
maximum tension are broken, the increase in maximum tension of the Hamzah, F.S., 2014. Experimental Study on the Hydrodynamic Interaction between Two
other slings is negatively correlated with the distance from the position Floating Offshore Structures Moored Together and Subjected to Regular wave[J].
of the broken slings. Universiti Teknologi Petronas.
Harada, T., Nakashima, A., Arai, M., et al., 2019. Motions and safety of a floating
When considering motion response amplitude, it is advisable to liquefied natural gas and shuttle tanker during side-by-side offloading operations.
perform salvage operations in head seas to ensure smoother progress. It Proc. IME M J. Eng. Marit. Environ. 233 (2), 610–621.
is essential to enhance the strength of the slings and expand the buff­ Henkel, L.A., Nevins, H., Martin, M., et al., 2014. Chronic oiling of marine birds in
California by natural petroleum seeps, shipwrecks, and other sources. Mar. Pollut.
ering system’s capacity. Alternatively, if prioritizing the safety of lifting Bull. 79 (1–2), 155–163.
slings to reduce the risk of sling breakage, salvage operations can be Jia, D., Li, F., 2018. Influence of dynamics effect on safety of cable in wreck up-righting
conducted in beam seas. Additionally, some measures should be taken to project. J. Traffic Transport. Eng. 18 (1), 89–100.
Jin, Y., Chai, S., Duffy, J., et al., 2018. URANS predictions on the hydrodynamic
minimize the roll response of the barges in beam seas. In this study, it is interaction of a conceptual FLNG-LNG offloading system in regular waves. Ocean
supposed to avoid conducting operations in quarter seas. Eng. 153, 363–386.
A basic understanding of the coupling motion between the twin Lu, L., Teng, B., Sun, L., et al., 2011. Modelling of multi-bodies in close proximity under
water waves—fluid forces on floating bodies. Ocean Eng. 38 (13), 1403–1416.
barges and shipwreck was provided through a series of model tests. The Monfils, Rean, Gilbert, Trevor, Nawadra, Sefanaia, 2006. Sunken WWII shipwrecks of the
valuable suggestions for the environment direction suitable for wreck Pacific and east Asia: the need for regional collaboration to address the potential
salvage operations were proposed from the perspective of motions and marine pollution threat. Ocean Coast Manag. 49 (9–10), 779–788.
Navy, U.S., 2013. US Navy Salvage Manual Vol. 1, Strandings, Harbor Clearance and
sling tensions. The results of this study can guide the design and oper­
Afloat salvage[J]. Naval Sea Systems Command and JMS Publishing, Washington,
ation of salvaging systems and improve the efficiency and safety of these DC.
operations. Ok, H.T., Lee, S.J., Choi, J.H., 2017. Numerical simulation of motion of single and side-
Although coupled dynamic response characteristics of the twin- by-side vessels in regular waves using OpenFOAM. Ships Offshore Struct. 12 (6),
793–803.
barge salvaging system is investigated by extensive experimental tests, Persaud, R., Li, H.J., Leng, J.X., 2019. Numerical simulation of a passive heave
this study only considers one salvage method, i.e., twin-barge hydraulic compensator for scientific research ships. J. Nav. Architect. Mar. Eng. 16 (1), 45–59.

13
W. Wu et al. Ocean Engineering 293 (2024) 116603

Rogowska, Justyna, Wolska, Lidia, Namieśnik, Jacek, 2010. Impacts of pollution derived Xu, H., Zou, Z., Wu, S., et al., 2017. Bank effects on ship–ship hydrodynamic interaction
from ship wrecks on the marine environment on the basis of s/s ‘‘Stuttgart’’ (Polish in shallow water based on high-order panel method. Ships Offshore Struct. 12 (6),
coast, Europe). Sci. Total Environ. 408 (23), 5775–5783. 843–861.
Sun, S., Li, W., 2023. Experimental investigation on the coupling response of a multi- Yao, Z., Wang, W., Jiang, Y., et al., 2018a. Functional evaluation of hydraulic tension
floating side-by-side system between a single-point mooring carrier and an buffer system in SEWOL salvage project. J. Dalian Marit. Univ. 44 (1), 48–58.
articulated floating platform. Ocean Eng. 281, 114700. Yao, Z., Wang, W., Jiang, Y., et al., 2018b. Coupled responses of Sewol, twin barges and
Tikhonov, V.S., Lebedev, V.I., Kotov, A.V., 1997. Numerical simulation of longitudinal slings during salvage. China Ocean Eng. 32 (2), 226–235.
tackle oscillations when salvaging sunken vessels from deep water. Ocean Eng. 24 Yuan, Z.M., Li, L., Yeung, R.W., 2019. Free-surface effects on interaction of multiple ships
(7), 605–621. moving at different speeds. J. Ship Res. 63 (4), 251–267.
Velayudhan, A.K.D., 2019. Design of a supervisory fuzzy logic controller for monitoring Zhang, M., Dai, L., Gan, J., et al., 2016. Safety and analysis of shipwreck salvage
the inflow and purging of gas through lift bags for a safe and viable salvaging operation based on AQWA. J. Wuhan Univ. Technol. 40 (3), 466–471.
operation. Ocean Eng. 171, 193–201. Zhang, F., Hou, J., Ning, D., et al., 2023. Performance analysis of the passive heave
Vieira, D.P., de Mello, P.C., Dotta, R., et al., 2018. Experimental investigation on the compensator for hydraulic shipwreck lifting systems in twin-barge salvaging. Ocean
influence of the liquid inside the tanks in the wave behavior of FLNG vessels in side- Eng. 280, 114469.
by-side offloading operations. Appl. Ocean Res. 74, 28–39. Zhao, D., Hu, Z., Chen, G., 2017. Experimental investigation on dynamic responses of
Xin, S., Wang, Y., Wang, L., et al., 2023. Dynamic analysis of the mooring system of a FLNG connection system during side-by-side offloading operation. Ocean Eng. 136,
salvage barge: a coupled time-domain method considering seabed resistance during 283–293.
the off-bottom stage. Ocean Eng. 275, 114078. Zhao, W., Milne, I.A., Efthymiou, M., et al., 2018. Current practice and research
Xu, X., Li, X., Lu, H., et al., 2012. An experimental and numerical study on motions of directions in hydrodynamics for FLNG-side-by-side offloading. Ocean Eng. 158,
three adjacent barges in floatover installation[C]. The Twenty-Second International 99–110.
Offshore and Polar Engineering Conference. OnePetro. Zhou, J., Ren, J., Bai, W., 2023. Survey on hydrodynamic analysis of ship–ship
Xu, X., Li, X., Yang, J., et al., 2016. Hydrodynamic interactions of three barges in close interaction during the past decade. Ocean Eng. 278, 114361.
proximity in a floatover installation. China Ocean Eng. 30, 343–358.

14