Ocean Engineering 229 (2021) 108621

Contents lists available at ScienceDirect

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

Propulsion power prediction for an inland container vessel in open and

restricted channel from model and full-scale simulations
H. Islam a, C. Guedes Soares a, *, J. Liu b, c, X. Wang b, c
a
Center for Marine Technology and Ocean Engineering (CENTEC), Instituto Superior Tecnico, Universidade de Lisboa, Portugal
b
Intelligent Transportation Systems Research Center, Wuhan University of Technology, Wuhan, PR China
c
National Engineering Research Center for Water Transport Safety, Wuhan, PR China

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

Keywords: The paper presents simulation results of an inland container vessel designed to be operated in the inland waters
CFD of China. Initially, calm water simulations are performed in open water with varying draft and speed to deter­
Inland vessel mine the possible propulsion power required by the vessel for regular operations. Next, static drift simulations
Restricted waters
are performed with a heave and pitch-free motion at varying drift angles and drafts. Encountered hull resistance
Static drift
Scale effect
and linear derivatives are determined from the drift results to assess the vessel’s maneuvering capabilities.
Power prediction Finally, simulations are performed in restricted water following possible operating channel geometry to assess
OpenFOAM the vessel’s performance in restricted operating conditions. The required propulsion power of the vessel has been
calculated for each case and compared. Cases that showed maximum resistance was re-stimulated in full scale to
discard the scaling effect from predictions. The results show that a significant increase in resistance is observed
when the vessel operates in shallow and narrow channels, which limits its possible operating speed. The study
concludes that model-scale simulations are efficient for preliminary studies. However, for the required power
prediction, full-scale simulations should be considered.

1. Introduction et al., 2019), as well as appropriate autonomous systems for collision

avoidance (Perera et al., 2015; Zhang et al., 2015), other aspects such as
The present-day challenge in the water transportation industry is to the economy (Kretschmann et al., 2017; Santos and Guedes Soares,
minimize energy consumption while ensuring maximum protection for 2018) and safety (Zhang et al., 2016; Wu et al., 2020) are also of
the marine environment, the vessel, and the crew. The maximization of concern. It is generally accepted that the transition to autonomous
vessel efficiency and economy are also among key targets. With the shipping has to be done gradually starting with coastal and inland
everlasting threat of climate change, researchers are continuously transportation and with intelligent (Perera et al., 2012) and
looking for ways to make the way of life more sustainable. The most remote-controlled ships, before moving to fully autonomous oceangoing
significant contributor to CO2 emission is the transportation industry, ships.
which is essential for the transfer of humans and goods. However, in As part of the above-referred effort, a project targeting a 64 TEU
terms of emissions, surely some modes of transportations are better than inland container vessel in China is under development to transport
others. Despite transporting more than 90% of global goods, shipping is goods and containers with trends of intelligent, remote, or autonomous
responsible for only 3% of global CO2 emissions (Shipping and World control. This paper focuses on the hydrodynamic study of the vessel to
Trade, 2020; EU Emission Report, 2020). As such, countries with ensure appropriate propulsion power installation in the vessel to ensure
abundant inland waterways, like the Netherlands (Rhine River) and smooth operations with minimum environmental impact.
China (Yangtze River) are focusing on developing reliable and efficient Different research groups have been developing RANS models since
inland water transportation systems. the late 1980s, and the field has matured significantly in the last the two
At the same time, there is a trend to increase automation in ships, decades. With the introduction of standard hull forms and related
moving towards autonomous ships. While initial efforts have focused on experimental data in the Gothenburg 2000 workshop (Larsson et al.,
guidance and control of the vessels (Moreira et al., 2007; Hinostroza 2003), global benchmarks were set and validation of CFD models

* Corresponding author.
E-mail address: [email protected] (C. Guedes Soares).

https://doi.org/10.1016/j.oceaneng.2021.108621
Received 27 September 2020; Received in revised form 4 January 2021; Accepted 11 January 2021
Available online 21 April 2021
0029-8018/© 2021 Elsevier Ltd. All rights reserved.
H. Islam et al. Ocean Engineering 229 (2021) 108621

became easier. After the Gothenburg 2010 (Larsson et al., 2011) and operation scenario, the paper focuses more on the practical application
Hino and Kenkyūjo (2005) and Tokyo (2015) (Hino et al., 2021) of CFD in assessing vessel propulsion power performance after consid­
workshops witnessed gradual improvement in the prediction accuracy ering different operation scenarios.
of CFD models, which came above 3% in 2010. In the last decade, a very
large number of publications have been made that focus on the verifi­ 2. Method
cation and validation of different popular CFD models. Apart from the
workshop proceedings, which includes results from a large number of 2.1. The numerical solver
participants, some of the relatively recent CFD studies in resistance
prediction for standard ships were done by Sadat-Hosseini et al. (2013), The open-source CFD toolkit, OpenFOAM (Open Field Operation and
Kim et al. (2013), and Islam and Guedes Soares (2019). Over the years, Manipulation) is used for the present studies. The solver has been under
overall expertise in calm water CFD simulations have developed development for more than 15 years and is widely popular among re­
significantly and CFD has proven to be a suitable substitute for towing searchers and in the offshore industry. OpenFOAM contains several
tank simulations. However, studies in academia are mostly related to the packages to perform multi-phase turbulent flow simulations for floating
validation of large seagoing vessels, and works related to inland vessel objects and also allows relatively easy customization and modification
propulsion power prediction are still quite limited. of solvers because of its modular design. The solver has been detailed by
Along with the development of resistance prediction, ship maneu­ Jasak (2009). A brief description of the solver is provided below.
vering studies using CFD have also been developing in parallel. Like The governing equations for the solver are the Navier-Stokes equa­
resistance studies, progress and standards in maneuvering studies are tion (1) and the continuity equation (2) for an incompressible laminar
generally set at SIMMAN (2008, 2014) workshops. Planar motion flow of a Newtonian fluid. In vector form, the Navier-Stokes and con­
mechanism (PMM) simulations are particularly popular for maneu­ tinuity equation are given by:
vering studies and a large number of publications are available in this ( )
∂v
regard. Among some of the recent works, Wang et al. (2011) used the ρ + v ⋅ ∇v = − ∇p + μ∇2 v + ρg (1)
CFD tool FLUENT to simulated oblique motion for KVLCC2 in deep and
∂t
shallow water. Kim et al. (2015) presented PMM simulation results for
∇⋅v = 0 (2)
the KCS model using in-house code SHIP_Motion and predicted hydro­
dynamic derivatives from simulation results. Hajivand and Mousaviza­ Here, vis the velocity, p is the pressure, μ is the dynamic viscosity, g is the
degan (2015a, b) also performed PMM simulations using STAR-CCM+
acceleration due to gravity and ∇2 is the Laplace operator.
and OpenFOAM (static drift only) for the DTMB 5512 model and pre­
The volume of fluid (VOF) method is used to model the fluid as one
dicted hydrodynamic derivatives from the simulation results. Yao et al.
continuum of mixed properties. This VOF method determines the frac­
(2016) presented static drift, turning, and pure sway simulation data
tion of each fluid that exists in each cell, thus tracking the free surface
using OpenFOAM for tanker model KVLCC2. Islam and Guedes Soares
elevation. The volume fraction was obtained by equation (3):
(2018) also performed PMM simulations for the KCS model using
OpenFOAM. However, as can be observed, most of the referred works ∂α
+ ∇ ⋅ (αv) = 0 (3)
discuss seagoing vessels and studies related to inland vessels are ∂t
comparatively scarce. Liu et al. (2015) provided a summary of the
evaluation and prediction of maneuverability of inland vessels, referring where v is the velocity field, and α is the volume fraction of water in the
to the notable works in the area. Liu (2017) also performed extensive cell, varying from 0 to 1, with 0 representing a cell full of air and 1
studies on the impact of rudder configuration on inland vessel maneu­ representing a cell full of water.
verability. However, general studies focusing on inland vessel hydro­ The unstructured collocated finite volume method (FVM) is used to
dynamic derivatives are still limited. discretize the governing equations. Time integration was performed by a
Another very important aspect for inland vessels is the restriction of semi-implicit second-order, two-point, backward-differencing scheme.
channel width and draft. Dedicated shallow and confined water work­ Pressure-velocity coupling was obtained through the PIMPLE algorithm,
shops like MASHCON (MASHCON, 2009–19) discusses extensively the a combination of SIMPLE and PISO. For the present study, turbulence
shallow water effects, blockage effects, lockage entrance, ship-ship in­ has been modeled with the Reynolds-averaged stress (RAS) SST k-ω two-
teractions. Shallow water affects the maneuvering capability of ships, equation model. To accommodate the motion of the vessel, the mesh
which may be critical in inland navigation (Zhou et al., 2016a; Xu et al., morphing technique has been used.
2020). Ship-ship interactions are also very important in inland naviga­
tion (Zhou et al., 2016b; Wnek et al., 2018). 2.2. The ship model
However, in general, most studies from the field focus on studying
large vessels passing through coastlines or port channels, like large The simulations were performed for a 2.065 m model that represents
crude carriers (Toxopeus et al., 2013; Lee and Hong, 2017; Hoydonck a 69 m long 64 twenty feet equivalent unit (TEU) inland container ship
et al., 2018) and container ships (Terziev et al., 2018; Zeng et al., 2019). model. Simulations are performed on a bare hull with no appendages.
Thus, studies related to the impact of channel restrictions on the resis­ The vessel follows a novel design aiming at an intelligent inland vessel
tance for inland vessels are still limited. with multiple advanced technologies, intended to be operated between
The present study aims to predict the propulsion power requirement Huzhou and Shanghai, China. Apart from a blunt bow, the vessel fea­
for a 64 TEU autonomous inland container vessel designed to be oper­ tures a rim-driven thruster propulsion system, and ship-shore coordi­
ated in the waterways of Zhejiang Province. The paper presents CFD nation assisted navigation. A representative hull diagram and the ship’s
simulation results for three different types of operations of the inland principal dimensions (for both full scale and model scale) are shown in
vessel, including cruising in open waters with design speed, maneu­ Fig. 1 and Table 1.
vering, and passing through restricted channels. Four different loading
conditions have been considered for the study, and a large number of 2.3. Mesh generation
simulations have been performed using a RANS based CFD model with
OpenFOAM to predict the vessel’s resistance and behavior in different For all the simulations, meshes are generated using the mesh gen­
conditions. Power prediction for the full-scale ship has been made using eration utility of OpenFOAM. The blockMeshDict is used to generate the
the resistance data and a recommendation for installed power has been initial simulation domain, which follows the ITTC-2011 and ITTC-2014
made. Instead of using CFD for an in-depth analysis of a particular recommendations of one ship length at the front, two ship lengths after

2
H. Islam et al. Ocean Engineering 229 (2021) 108621

Table 2
Specifications of mesh used for the different simulations.
Simulation Simulation Mesh Minimum Minimum y+
Condition domain Resolution Cell size (X cell size value
(Million) x Y x Z;
before
Layers)

Calm water Half 1.5 0.007 × 0.00175 65

0.0067 x
Fig. 1. Simplified representation of the 64TEU hull form used in simulations. 0.00375
Static Drift Full 3.5 0.007 × 0.00175 65
(Average) 0.0067 x
0.00375
Table 1 Restricted Full 2.7 0.00586 × 0.00136 40
Principal dimensions of the 64TEU vessel, for both full scale and model scale. water (Average) 0.00544 x
Full-scale Model-scale 0.00375

Scale 1:1 1:34.5

Lpp (m) 69.00 2.00 2.4. Computational resources
Lwl (m) 71.26 2.065
Bwl (m) 12.60 0.365
D (m) 3.45 0.1 The simulations are performed on two computers, an intel Xenon
T (m) 2.60 0.0754 workstation with 2.2Gz processor with 64 threads, and 16 GB of RAM,
Displacement (m3) 2040300 49.69 and an Intel Core i9 computer with 2.90Gz processor 24 threads, and 64
S w/o rudder (m2) 1154.8 0.9702
GB of RAM. However, for each simulation, only 12 cores were used to
CB 0.874 0.874
CM 0.984 0.984 ensure efficient memory handling. Each simulation was run for 50 s
(simulation time), with an average time step of 0.01s, and took a
physical time of 40 h (on average).
the stern, one ship length at the side, and three ship lengths in the
vertical direction. However, the lateral and vertical domain size was 3. Results
different for the shallow water cases. The initial domain is then refined
six times using topoSetDict and refineMeshDict. Next, snappy­ The paper focuses on resistance prediction for an inland container
HexMeshDict is used to incorporate the hull form into the blockMesh. vessel model at different operating conditions. Initially, calm water
Four layers are added around the hull for proper capturing of the viscous simulations are performed at four different speeds at four different
force. The initial simulations in open calm water are run for the half-hull drafts. The present work is a continuation of the work done by Islam
(port side) with a heave and pitch-free motion. The static drift and et al. (2021), where a detailed verification and validation study for the
restricted water simulations are performed using the full hull (both calm water simulations has been presented. The verification or uncer­
starboard and port side). Although depending on the water depth and tainty study was performed for three Froude numbers and revealed that
drift angle, some details in the mesh are adjusted, while the overall mesh the resistance prediction has relatively higher uncertainty in high-speed
topology in all the simulations remains the same. cases, whereas, sinkage prediction has higher uncertainty for low-speed
The general mesh used for calm water simulations has a resolution of cases. This is understandable since at lower speed the sinkage value
1.5 million cells, for the half domain. The mesh resolution was selected remains small and relatively higher mesh resolution is required to cap­
based on the grid dependency study reported for the vessel in Islam et al. ture that vertical motion properly. As for trim results, they showed
(2021). For static drift, the overall mesh resolution is higher as the full relatively low uncertainty at all vessel speeds. For validation of the re­
domain (both starboard and port side of the hull) is simulated. The mesh sults, the results were compared with experimental and StarCCM + re­
resolution varies depending on the drift angle as the final refinement sults. However, it was mentioned that the experimental study had high
area is enlarged to cover the drifting hull. Thus, the resolution is higher uncertainty and the CFD results were found to be in better agreement
for larger drift angles. This was done to ensure even mesh resolution with each other. A summary of the results presented by Islam et al.
around the hull for all the cases. The average mesh resolution for the (2021) is shown in Tables 3 and 4. The highest mesh resolution from the
static drift cases is around 3.5 million. Finally, the restricted water study has been used for the presented simulations.
simulations are performed with a narrow channel with shallow drafts. Next, drift simulations are performed with varying drift angles for
The mesh resolution varies slightly due to the changing depth and width. the same speed, with a heave and pitch-free motion. From the results,
The refinement level for the cases was maintained the same as the maneuvering derivatives are also calculated. Then, simulations are
previous two cases, thus mesh distribution near the hull remained the performed in a restricted channel to assess how to channel depth and
same as before. Even mesh resolution was maintained in the vertical width influences the vessel’s encountered resistance.
direction from the hull bottom to the bottom boundary. The average Simulations for open water and static drift were performed with just
mesh resolution is around 2.7 million. A rough description of the used heave and pitch free motion, while keeping the other four motions
meshes is provided in Table 2, and the general mesh assembly for the restricted. As for restricted water cases, static simulations were per­
cases are shown in Figs. 2–4. The y + values shown in Table 2 corre­ formed with all motions restricted. Since an inland vessel has been
spond to y + values for the highest speed simulated for that type of considered, which should only operate in calm waters, the impact of
simulation. Since multiple cases were simulated, setting mesh quality other motions, in general, should be minimum (Islam and Akimoto,
based on the highest speed ensures that the y + criterion will be met for 2018). Furthermore, the vessel has been designed for enhanced stability
all the other cases. However, for all the cases, even for the slowest speed, compromising some efficiency aspects. Thus, in general, the vessel has a
the y + value remained above the value of 30. This is important since low motion response. Although in restricted waters all motions become
wall function has been used in the simulations and following the Law of important, adding more motions would also include more uncertainty
the Wall, y + values should remain between 30 and 300 for reliable regarding the results. And without proper validation data, concluding
results. from the results would have become more difficult. Thus, only static
simulations have been performed for restricted waters.
Finally, specific cases are repeated with full-scale simulation to

3
H. Islam et al. Ocean Engineering 229 (2021) 108621

Fig. 2. Simulation mesh for calm water simulations in open water; overall simulation domain (half-hull), mesh distribution at the side, on the free surface, and the
hull form.

Fig. 3. Simulation mesh for static drift simulations in open water; overall simulation domain (full-hull), mesh distribution at the free surface, in the vertical direction,
and on the hull form.

obtain a more reliable prediction for the required installed power. All coefficients are measured, including sinkage and trim (except for the
model scale simulations are performed for four draft conditions, lowest draft case). For the lowest draft (0.0435 m), dynamic simulations
whereas, the full-scale simulations are repeated just for drafts 3 and 4. A show high oscillation due to very low displacement volume and rela­
summary of the simulated cases is shown in Table 5. Further details tively large mesh volume.
regarding the simulations are provided in individual sub-sections of the Furthermore, initial dynamic simulations for the case reveal very low
results. movement, thus, the simulations are later performed in static condi­
tions. All resistance results are converted to full hull (both port and
starboard side) resistance. For coefficient calculation, the following
3.1. Calm water simulation in open waters
equation (4) is used for the drag coefficient,
/
The calm water simulations are performed in an open channel (with c = F 0.5 × ρ × A × v2 (4)
relatively large width and depth, one ship length as width on each side,
WS

and one ship length in depth) with a heave and pitch-free motion. Here, c is the coefficient, ρ is the water density, AWS is the wetted surface
Considering different loading conditions, simulations are performed for area of the vessel, and v is the vessel speed. The simulation results are
four different draft conditions at varying speeds. As mentioned before, shown in Table 6. A relative comparison among the results is also shown
the 1.5 million mesh resolution (half-hull) used for the simulations was in Figs. 5–7.
systematically verified using a detailed uncertainty study presented by The drag coefficient results show that the hull has the least amount of
Islam et al., 2021. resistance in the draft 2 (0.0580 m) condition, and overall all the drafts
A total of eight simulations are performed for varying Froude num­ show relatively low resistance at Froude 0.155 (0.7 m/s). The hull shows
ber, drafts (0.0435 m, 0.0580 m, 0.0754 m, and 0.0870 m) and a similar level of resistance at draft 1 and draft 4. However, while draft 1
displacement volume (0.7884 m2, 0.8839 m2, 0.9702 m2, and 1.0239 shows higher viscous resistance, draft 4 shows higher pressure
m2, respectively). For all the cases, pressure, viscous, and total drag

4
H. Islam et al. Ocean Engineering 229 (2021) 108621

Fig. 4. Simulation mesh for restricted water simulations; overall simulation domain (full-hull), mesh distribution on the hull form, in the vertical direction, and at the
free surface.

Table 3
Summary of the verification study (uncertainty analysis) for the 64TEU inland vessel (model scale) simulations using OpenFOAM. Islam et al. (2021).
Total Resistance Coefficient, Ct Sinkage (cm) Trim (deg)

Froude Num. 0.111 0.155 0.200 0.111 0.155 0.200 0.111 0.155 0.200

Output values Ø1 (M1) 5.94E-03 5.35E-03 6.29E-03 − 0.100 − 0.200 − 0.330 − 0.0794 − 0.0415 0.02673
Ø2 (M2) 5.68E-03 5.57E-03 6.61E-03 − 0.106 − 0.190 − 0.330 − 0.0778 − 0.0414 0.0257
Ø3 (M3) 6.13E-03 6.00E-03 7.02E-03 − 0.113 − 0.203 − 0.343 − 0.074 − 0.04134 0.0175
Convergence ε21/ε32 − 0.58 0.50 0.75 0.86 − 0.77 0.03 0.42 1.67 0.13
Corrected Uncertainty U1c 1.72% 1.60% 10.91% 281.64% 4.64% 0.00% 0.55% 0.06% 0.22%
Normalized Correction Factor-based Approach
Corrected Uncertainty U1c 0.02% 1.43% 37.59% 1119.26% 0.65% 0.14% 0.25% NA 3.82%

Table 4
Comparative results among CFD and EFD estimation for total drag estimation (model scale) for the model scale 64TEU inland vessel. Islam et al. (2021).
Draft: 0.0435 m 0.0754 m

Velocity (m/s) EFD CFD_StarCCM+ CFD_OpenFOAM EFD CFD_StarCCM+ CFD_OpenFOAM

0.50 – 5.69E-03 6.21E-03 – 6.27E-03 5.94E-03

0.70 – 5.88E-03 5.74E-03 – 6.45E-03 5.35E-03
0.90 5.14E-03 6.49E-03 6.36E-03 5.47E-03 6.85E-03 6.29E-03

5
H. Islam et al. Ocean Engineering 229 (2021) 108621

Table 5 the calculations have been done assuming a regular diesel engine driven
Summary of the simulated cases in the study. propulsion. The prediction results are shown in Table 7. In Fig. 8, the
Simulation Simulation Degrees of Velocity Drift Domain velocities correspond to Froude number 0.067, 0.111, 0.155, and 0.200,
Condition domain freedom (m/s) Angle depth to respectively.
Ship draft As expected, the results show the highest power requirement for the
ratio, h/D
design speed at the maximum draft. The power requirement for the
Calm water Half (model 2 (Heave 0.3, 0.5, 0 25
scale) and pitch 0.7, 0.9
free)
Static Drift Full (model 2 (Heave 0.9 0, 3, 6, 25
scale) and pitch 9, 12,
free) 15
Restricted Full (model 0 (All 0.5 0 2.0, 1.5,
water scale) motions 1.2, 1.1
restricted)
Calm, drift, Full (Full 0 (All 5.2, 2.88 0, 15 25, 1.1
restricted scale) motions
restricted)

resistance. The relative sinkage also increases with increasing load (or
draft). As for trim, at low draft condition, the vessel shows bow ward
trim at all speeds. With increasing draft and speed, the vessel shows
stern wards trim, as expected.

3.1.1. Power prediction (conversion from model scale)

Next, from the predicted resistance, approximate power calculation Fig. 5. Relative comparison among the total drag coefficient for different
is done for the full-scale ship. To do so, the residual or wave-making draft conditions.
(pressure) resistance Cw is taken from the model scale simulation re­
sults and the viscous resistance Cv is calculated using the ITTC-1957
guidelines (Molland et al., 2011; ITTC-2002). The value of form factors
for the vessel (1+k) is calculated to be 1.454, 1.377, 1.45, and 1.498, for
draft 1, 2, 3, and 4, respectively. Then total drag resistance for the
full-scale ship is calculated by combining the wave-making and viscous
drag coefficient (CT = Cw + Cv ). The effective drag force for a full-scale
vessel is then calculated as F = CT × 0.5 × ρ × AWS × v2 . Here, AWS is
the full-scale vessel wetted surface area and v is corresponding full-scale
velocity. The required effective power can be calculated as P = F × v.
Finally, the required installed power is calculated by assuming a
quasi-propulsive coefficient of 0.55, 2.0% shaft loss, 5.0% gearbox loss,
and a 15% reduction to maintain the maximum continuous rating of the
propeller (Marine Insight, 2019).
Since the vessel will be operating in inland waters, no sea margin is
considered. The air drag is also ignored. Although there is a possibility
that the proposed vessel would be propelled by electric motors, specific
Fig. 6. Relative comparison among sinkage results for different draft (2, 3,
detail regarding losses in electric power train is yet to be available. Thus, 4) conditions.

Table 6
Calm water simulation results for the 64TEU inland water vessel model.
Froude Number Velocity (m/s) Viscus drag coefficient, Cv Pressure drag coefficient, Cp Total drag coefficient, Ct Sinkage (cm) Trim (deg)

Draft 1: 0.0435 m
0.067 0.3 6.37E-03 1.72E-03 8.09E-03 – –
0.111 0.5 4.95E-03 1.26E-03 6.21E-03 – –
0.155 0.7 4.22E-03 1.51E-03 5.74E-03 – –
0.200 0.9 3.73E-03 2.64E-03 6.36E-03 – –
Draft 2: 0.0580 m
0.067 0.3 5.59E-03 1.76E-03 7.35E-03 − 0.025 − 0.194
0.111 0.5 4.40E-03 1.30E-03 5.71E-03 − 0.065 − 0.1786
0.155 0.7 3.72E-03 1.38E-03 5.09E-03 − 0.140 − 0.15
0.200 0.9 3.60E-03 2.34E-03 5.93E-03 − 0.240 − 0.1036
Draft 3: 0.0754 m
0.067 0.3 5.82E-03 1.79E-03 7.61E-03 − 0.044 − 0.1023
0.111 0.5 4.47E-03 1.47E-03 5.94E-03 − 0.100 − 0.0794
0.155 0.7 3.72E-03 1.63E-03 5.35E-03 − 0.200 − 0.0415
0.200 0.9 3.56E-03 2.74E-03 6.36E-03 − 0.321 0.0233
Draft 4: 0.0870 m
0.067 0.3 5.78E-03 2.12E-03 7.90E-03 − 0.057 − 0.0388
0.111 0.5 4.47E-03 1.85E-03 6.32E-03 − 0.12 − 0.0155
0.155 0.7 3.70E-03 1.94E-03 5.65E-03 − 0.23 0.0318
0.200 0.9 3.53E-03 3.18E-03 6.71E-03 − 0.39 0.1206

6
H. Islam et al. Ocean Engineering 229 (2021) 108621

drag coefficient, draft 2 initially shows relatively low resistance

comparing to draft 1, which gradually increases with an increasing drift
angle. As for lateral force and yaw moment, they follow a clear trend
with higher values for the larger draft and larger drift angle. The slope
for the trends varies slightly for different drafts, indicating that the
values of maneuvering derivatives are not the same for each draft.
The sinkage and trim results for the cases are shown in Figs. 12 and
13, respectively. As can be seen from the sinkage results, the sinkage
increases with an increasing drift angle. At the design speed and draft,
for a drift of 15◦ , the sinkage increases by around 0.05 m in full scale,
which might be particularly important while navigating through very
shallow waters. A sudden high drift maneuvering in shallow water might
ground the vessel if maneuvering is not done with care. The trim angle
also shows a change of value with an increasing drift angle, however, the
effect is relatively less. Nevertheless, the combined effect of added
Fig. 7. Relative comparison among trim results for different draft (2, 3, sinkage and trim might prove important for very shallow channels.
4) conditions.
3.2.1. Estimation of static derivatives
maximum draft is roughly 150 KW more, comparing to the design draft, The linear hydrodynamic derivatives are estimated from the static
for the design speed. Overall, the projected required power for the drift simulation results. Hydrodynamic derivatives represent the rate of
64TEU inland vessel is around 1150 KW. change of force or moment with, changing drift angle for linear de­
rivatives, and changing velocity and acceleration for non-linear de­
rivatives. From the static drift results, linear derivatives YV and NV are
3.2. Static drift simulation in open water calculated. Although some non-linear derivatives may also be calculated
from the static results, their results are generally less reliable. The de­
Static drift tests are principally performed for assessing vessel rivatives are calculated from the slope of the changing lateral force and
maneuverability as part of the Planar Motion Mechanism (PMM) tests. yaw moment, vs the changing drift angle (Islam and Guedes Soares,
However, it is not uncommon for vessels to be exposed to drifting motion
while maneuvering, especially during docking and undocking. Thus, the
required propulsion power was studied for drifting cases.
Static drift simulations are performed for model scale ship at the
design speed (Fr. 0.2, 0.9 m/s) for varying drift angles. Six simulations
are performed for different drift angles for each ship draft, with a heave
and pitch free motion except for the smallest draft case (draft 1).
As mentioned in section 3.1, draft 1 shows instability for heave and
pitch-free motion, and thus drift simulations for draft 1 are performed
with all ship motions restricted. From the simulation, total drag resis­
tance, lateral force, and yaw and roll moments are measured. The results
are shown in Table 8. The lateral force and moment for the 0 drift cases
are negligible, thus they are simply replaced by 0.0. Comparison among
the forces and moment for changing drifts and drafts are also shown in
Figs. 9–11.
Fig. 8. Required installed power prediction for the 64TEU inland vessel at
As expected, Figs. 9–11 show that with increasing draft and drift
varying draft and speed (Calculated from model scale results).
angle, the encountered force and moment also increases. For the total

Table 7
Calm water propulsion power requirement for the 64TEU inland water vessel model, converted results from the model scale.
Froude Velocity Viscous drag Pressure/wave drag coefficient, Total drag Required Effective Power Required Installed Power
Number (knots) coefficient, Cv Cp/Cw coefficient, Ct (KW) (KW)

Draft 1: 1.5 m
0.067 3.4 2.90E-03 1.72E-03 4.63E-03 11.50 26.42
0.111 5.6 2.71E-03 1.26E-03 3.96E-03 44.80 102.92
0.155 7.8 2.59E-03 1.51E-03 4.10E-03 126.10 289.72
0.200 10.1 2.50E-03 2.64E-03 5.14E-03 339.60 780.25
Draft 2: 2.0 m
0.067 3.4 2.75E-03 1.76E-03 4.51E-03 12.55 28.84
0.111 5.6 2.56E-03 1.30E-03 3.87E-03 48.94 112.44
0.155 7.8 2.45E-03 1.38E-03 3.83E-03 131.86 302.96
0.200 10.1 2.37E-03 2.34E-03 4.71E-03 348.34 800.33
Draft 3: 2.6 m
0.067 3.4 2.89E-03 1.79E-03 4.68E-03 14.31 32.87
0.111 5.6 2.70E-03 1.47E-03 4.17E-03 57.89 133.00
0.155 7.8 2.58E-03 1.63E-03 4.20E-03 159.03 365.38
0.200 10.1 2.49E-03 2.74E-03 5.23E-03 424.92 976.29
Draft 4: 3.0 m
0.067 3.4 2.99E-03 2.12E-03 5.12E-03 16.49 37.89
0.111 5.6 2.79E-03 1.85E-03 4.63E-03 67.92 156.06
0.155 7.8 2.66E-03 1.94E-03 4.61E-03 183.92 422.58
0.200 10.1 2.58E-03 3.18E-03 5.75E-03 493.31 1133.41

7
H. Islam et al. Ocean Engineering 229 (2021) 108621

Table 8
Static drift simulation results for the model scale 64 TEU inland water vessel model, for Froude number 0.2 and model scale velocity of 0.9 m/s.
Drift Angle (deg) Total drag coefficient, Ct Lateral force coefficient, F’y Roll moment coefficient, M’x Yaw moment coefficient, M’z Sinkage (cm) Trim (deg)

Draft 1: 0.0435 m
0 6.36E-03 0.00 0.00 0.00 – –
3 6.36E-03 3.07E-05 6.03E-05 4.30E-05 – –
6 6.45E-03 7.73E-05 1.24E-04 7.94E-05 – –
9 6.90E-03 1.29E-04 1.90E-04 1.10E-04 – –
12 7.30E-03 1.93E-04 2.53E-04 1.46E-04 – –
15 7.67E-03 2.69E-04 3.15E-04 1.93E-04 – –
Draft 2: 0.0580 m
0 5.93E-03 0.00 0.00 0.00 − 0.240 − 0.104
3 6.10E-03 5.25E-05 1.29E-05 7.54E-05 − 0.24 − 0.106
6 6.43E-03 1.37E-04 2.88E-05 1.42E-04 − 0.266 − 0.103
9 6.85E-03 2.59E-04 5.22E-05 2.13E-04 − 0.303 − 0.098
12 7.52E-03 3.86E-04 8.04E-05 2.88E-04 − 0.334 − 0.089
15 8.21E-03 5.04E-04 1.15E-04 3.57E-04 − 0.39 − 0.081
Draft 3: 0.0754 m
0 6.36E-03 1.04E-06 − 4.35E-07 − 8.14E-07 − 0.321 0.023
3 6.50E-03 8.44E-05 1.42E-05 1.19E-04 − 0.333 0.023
6 6.82E-03 2.27E-04 3.13E-05 2.26E-04 − 0.353 0.025
9 7.52E-03 4.32E-04 5.76E-05 3.34E-04 − 0.393 0.033
12 8.48E-03 6.55E-04 9.12E-05 4.50E-04 − 0.447 0.043
15 9.68E-03 9.07E-04 1.36E-04 5.64E-04 − 0.513 0.055
Draft 4: 0.0870 m
0 6.70E-03 0.00 0.00 0.00 − 0.390 0.121
3 6.95E-03 1.18E-04 1.57E-05 1.47E-04 − 0.400 0.121
6 7.44E-03 3.15E-04 3.53E-05 2.85E-04 − 0.400 0.119
9 8.15E-03 5.53E-04 6.83E-05 4.33E-04 − 0.455 0.127
12 9.26E-03 8.28E-04 9.88E-05 5.61E-04 − 0.500 0.138
15 1.07E-02 1.13E-03 1.43E-04 7.05E-04 − 0.600 0.154

Fig. 9. Prediction of total drag coefficient for different ship drafts at varying Fig. 11. Prediction of Yaw moment coefficient for different ship drafts at
drift angle, at design speed. varying drift angle, at design speed.

Fig. 10. Prediction of lateral force coefficient for different ship drafts at Fig. 12. Prediction of sinkage for different ship drafts at varying drift angle, at
varying drift angle, at design speed. design speed.

8
H. Islam et al. Ocean Engineering 229 (2021) 108621

rough understanding regarding how power requirement changes with

changing velocity in case of operating in restricted waters.
An important aspect for restricted water simulations is the study of
effects related to hull wake, propwash, the influence of the number of
propellers, current, and reflected waves in the shallow and narrow
channel. However, the present study mostly focused on resistance pre­
diction and studies related to these influences were ignored. Another
aspect is that since the restricted water simulations were performed at
low Froude numbers, the hull wake and the interaction of waves with
the boundaries were limited. Furthermore, the studied vessel is designed
with a full-directional thruster instead of propellers and rudders. Which
makes the study of wash from single or twin propellers become some­
what redundant. Thus, for the propulsion power assessment, only bare
hull was simulated without considering propeller and rudder effects.
All restricted water simulations are performed with a static hull,
Fig. 13. Prediction of trim for different ship drafts at varying drift angle, at sinkage, and trim motion restricted. Although vessel squat in shallow
design speed. channels is a topic of interest, it is not studied in the present work since
the principal target of the work is propulsion power prediction and
2018). The results are shown in Table 9. dynamic simulation in shallow water is both challenging and resource
Following theoretical hydrodynamics, for high stability, Yv should consuming. Furthermore, the studied cases here fall within low depth
have a negative value and Nv should have a small negative or positive Froude number, thus the impact of sinkage and trim is limited. The effect
value. From the results, it can be interpreted that the vessel has rela­ of changing channel width was also not studied in the model scale study.
tively low lateral stability, and the stability reduces with an increasing The results for calm water simulations in restricted waters are shown in
draft. However, the vessel shows reasonable stability in yaw moment. Table 10.
Nevertheless, the vessel shows relatively high stability comparing to sea- The results from Table 8 show that with shallower water, the viscous
going tankers and container vessels, which is expected for inland vessels. resistance decreases whereas the pressure resistance increases. The re­
sults also show that at draft 3 (design draft) and 4, the total drag coef­
3.2.2. Estimation of power requirement during maneuvering ficient shows an optimal value at velocity 0.5 m/s, and the drag
To analyze the power requirement for propulsion during maneu­ increases with increasing or decreasing velocity. A trend line for the
vering, the required propulsion power is also calculated from the drag total drag coefficient at velocity 0.5 m/s for changing depth is shown in
force coefficients for the static drift cases. Power is estimated following Fig. 15. The figure shows that with reducing depth, the total drag co­
the same method as described in Section 3.1.1. The results for the final efficient increases, and maximum drag is observed for draft 1 (in most
required installed power is shown in Fig. 14. The results indicate a much cases), while the rest of the drafts show relatively close values. Since
higher power requirement at drifting conditions, which increases with draft 1 has the lowest draft, the simulations for draft 1 were the shal­
drifting angle and draft. As such, the vessel will face a reduction in speed lowest cases.
as it maneuvers with larger drift angles. The power prediction also
suggests that high drift maneuvering would be particularly difficult if a 3.3.1. Estimation of power requirement in restricted waters
negative current is present in the area or channel of maneuvering. Following the previous cases, the total required installed power is
also calculated for the restricted water cases. However, to account for
the shallow water effect while estimating the viscous drag coefficient for
3.3. Calm water simulation in restricted waters
a full scale, the form factor was calculated by adding a term as followed
(Millward, 1989):
Finally, considering that the vessel will be operating in inland wa­
{ ( / )1.72 }
terways which are, in many cases, restricted in width and depth, simu­
Cv = Cf * (1 + k)shallow = Cf ITTC * (1 + k) + 0.644 T h (5)
lations are also performed in restricted channels. The vessel is intended
to operate in a class-III channel, following the definition from the
“Standards For Inland River Navigation-2014". The depth of waterways Here, T is the ship draft and h is the channel water depth. The full-scale
within this class is generally around 2.0 m–2.4 m and the width would results are shown in Table 11.
be 30 m for a one-way line and 60 m for a two-way line. However, The results from Table 11 shows that the required propulsion power
assuming that the vessel, in general, would avoid such shallow paths in a varies significantly based on ship speed and bottom clearance. For the
fully loaded condition, simulations are performed considering the two- same bottom clearance, the power requirement increases significantly
way width and channel depth to vessel draft ratios of 2.0, 1.5, 1.2, with an increase in vessel speed. This indicates that if power is installed
and 1.1. In the model scale, the channel width is set as 1.74 m (60 m, full based on the design speed for open waters, the vessel won’t be able to
scale), and channel depth is varied following the depth to draft ratios. attain beyond a certain speed in restricted channels. Furthermore, if the
Since power requirement increases significantly in the case of restricted channel has a negative current greater than 5 knots, the vessel
operating in restricted waters, a reduced vessel model-scale speed of 0.5 might not be able to propel forward at all.
m/s (5.6 knots, full scale) is considered for the study. Some simulations A comparative image for power requirement at changing draft and
are also performed with higher and lower vessel speeds to generate a depth for the vessel at speed 5.70 knots is shown in Fig. 16. The results
show that even at h/D = 2, the power requirement is almost twice that of
Table 9 open water case (Fig. 8), and the power requirement increases sub­
Linear hydrodynamic derivatives of the 64TEU inland vessel. stantially for very shallow cases (h/D = 12.0, 1.10). However, the effect
Yv Nv is not just because of the channel depth, but also because of the channel
Draft_1 6.00E-05 4.00E-05
width, which has been considered to be 30 m for the presented cases.
Draft_2 0.0001 7.00E-05
Draft_3 0.0002 0.0001
Draft_4 0.0003 0.0001

9
H. Islam et al. Ocean Engineering 229 (2021) 108621

Fig. 14. Estimated power requirement for the 64TEU vessel while maneuvering with static drift at 10.1 knots.

Table 10
Restricted channel calm water simulation results (static simulations, heave, and
pitch motion restricted) for a model scale 64TEU inland container vessel.
Depth to Velocity Viscus drag Pressure drag Total drag
draft ratio (m/s) coefficient, Cv coefficient, Cp coefficient, Ct
(h/d)

Draft 1: 0.0435 m
1.10 0.1 4.39E-03 3.09E-02 3.53E-02
1.10 0.5 3.75E-03 3.20E-02 3.57E-02
1.20 0.5 5.30E-03 2.53E-02 3.06E-02
1.50 0.5 7.15E-03 1.51E-02 2.23E-02
2.00 0.5 6.45E-03 8.38E-03 1.48E-02
Draft 2: 0.0580 m
1.10 0.1 5.21E-03 2.81E-02 3.33E-02
1.10 0.5 4.03E-03 3.28E-02 3.68E-02
1.20 0.5 5.84E-03 2.34E-02 2.92E-02
1.50 0.5 6.20E-03 1.31E-02 1.93E-02
2.00 0.5 5.49E-03 7.93E-03 1.34E-02 Fig. 15. Total drag coefficient vs water depth to ship draft ratio for a model
Draft 3: 0.0754 m scale 64TEU inland container vessel, at speed 0.5 m/s.
1.10 0.1 1.21E-01 7.90E-01 9.11E-01
1.10 0.5 4.90E-03 3.15E-02 3.64E-02
1.20 0.5 6.44E-03 2.22E-02 2.86E-02 scale tests. Even for model scale tests, predicting viscous resistance
1.50 0.4 6.76E-03 1.17E-02 1.84E-02 generally requires double body simulation, which is expensive. Thus, the
1.50 0.5 5.76E-03 1.22E-02 1.79E-02 application of empirical formulation for predicting frictional (thus,
2.00 0.3 7.13E-03 7.20E-03 1.43E-02
2.00 0.5 5.49E-03 7.11E-03 1.26E-02
viscous) resistance for both model and full scale is quite common.
Draft 4: 0.0870 m Several researchers have already tried to address the issue related to the
1.10 0.1 1.31E-01 7.61E-01 8.92E-01 scale effect in deep and shallow waters (Min and Kang, 2010; Zeng et al.,
1.10 0.5 5.29E-03 3.04E-02 3.57E-02 2018, 2020; Terziev et al., 2019; Niklas and Pruszko, 2019; Zhang et al.,
1.20 0.5 6.40E-03 2.13E-02 2.77E-02
2020) and it is still an ongoing study.
1.50 0.5 6.04E-03 1.27E-02 1.88E-02
2.00 0.3 7.15E-03 7.71E-03 1.49E-02 When it comes to RANS simulations, one particular reason behind
2.00 0.4 6.17E-03 7.39E-03 1.36E-02 running such a sophisticated, resource-consuming solution is to predict
2.00 0.5 5.46E-03 7.13E-03 1.26E-02 the viscous resistance of the vessel with good accuracy. The solvers also
2.00 0.7 4.59E-03 1.01E-02 1.46E-02 include multi-equation turbulence models to better realize the turbu­
lence effect in both pressure and viscous resistance. However, while
3.4. Full-scale simulation results converting the data to full scale, following the common practice, the
viscous resistance predicted by the RANS model is simply discarded and
As explained in Section 3.1.1, for the conversion of model scale data an empirical formula is used to estimate the viscous resistance. The
to full scale, considering a similarity law, it is assumed that the wave- practice begs the question that what is the point of running the RANS
making resistance coefficient for both model and full-scale vessel re­ simulation in the model scale, to begin with.
mains the same, whereas, the viscous resistance coefficient is calculated However, running full-scale simulation ensuring the required level of
following the empirical formulation from ITTC-57. This process for mesh resolution (y + value) is highly resource consuming. Thus, for ship
scaling model scale results to full scale has been in use for several de­ behavioral investigation, running all the simulations in full scale might
cades and has been adopted in numerical studies from the experimental not be feasible. As such, for comparative study, model scale simulations
field. In the experimental study, it is simply impossible to perform full- are quite efficient. However, for power prediction, the extreme cases
identified in the model scale should be run again in full scale to properly

10
H. Islam et al. Ocean Engineering 229 (2021) 108621

Table 11
Calm water propulsion power requirement for the 64TEU inland water vessel (full scale) in restricted water, converted results from the model scale.
Depth to draft ratio Velocity Viscus drag Pressure or wave drag Total drag Required Effective Required Installed
(h/d) (knots) coefficient, Cv coefficient, Cp/Cw coefficient, Ct Power (KW) Power (KW)

Draft 1: 1.5 m
1.10 1.13 4.70E-03 3.09E-02 3.56E-02 3.28 7.54
1.10 5.70 3.71E-03 3.20E-02 3.57E-02 423.24 972.43
1.20 5.70 3.57E-03 2.53E-02 2.89E-02 342.89 787.82
1.50 5.70 3.30E-03 1.51E-02 1.84E-02 218.68 502.44
2.00 5.70 3.06E-03 8.38E-03 1.14E-02 135.69 311.75
Draft 2: 2.0 m
1.10 1.13 4.52E-03 2.81E-02 3.26E-02 3.37 7.74
1.10 5.70 3.57E-03 3.28E-02 3.63E-02 482.32 1108.16
1.20 5.70 3.43E-03 2.34E-02 2.68E-02 355.93 817.77
1.50 5.70 3.15E-03 1.31E-02 1.63E-02 215.81 495.83
2.00 5.70 2.92E-03 7.93E-03 1.08E-02 144.03 330.92
Draft 3: 2.6 m
1.10 1.13 4.69E-03 7.90E-01 7.94E-01 90.12 207.06
1.10 5.70 3.71E-03 3.15E-02 3.52E-02 513.10 1178.89
1.20 5.70 3.57E-03 2.22E-02 2.58E-02 375.53 862.80
1.50 4.57 3.39E-03 1.17E-02 1.51E-02 113.25 260.19
1.50 5.70 3.29E-03 1.22E-02 1.55E-02 225.24 517.50
2.00 3.44 3.28E-03 7.20E-03 1.05E-02 33.58 77.15
2.00 5.70 3.06E-03 7.11E-03 1.02E-02 148.11 340.28
Draft 4: 3.0 m
1.10 1.13 4.80E-03 7.61E-01 7.65E-01 91.64 210.55
1.10 5.70 3.80E-03 3.04E-02 3.42E-02 525.64 1207.71
1.20 5.70 3.66E-03 2.13E-02 2.49E-02 383.68 881.54
1.50 5.70 3.38E-03 1.27E-02 1.61E-02 247.79 569.31
2.00 3.44 3.37E-03 7.71E-03 1.11E-02 37.50 86.15
2.00 4.57 3.24E-03 7.39E-03 1.06E-02 84.32 193.73
2.00 5.70 3.14E-03 7.13E-03 1.03E-02 157.95 362.91
2.00 7.97 3.01E-03 1.01E-02 1.31E-02 547.03 1256.83

simulations are performed for the vessel design speed (10 knots) and the
design draft (draft 3). The results are shown in Table 12.
The grid dependency analysis shows that with increasing resolution,
the solver predicts lower power. A further uncertainty study on the drag
components, following the factor of safety (Celic et al., 2008) based
method in ITTC-2017 guidelines, showed that the corrected uncertainty
for the total drag coefficient is around 0.4%, whereas, for viscous and
pressure drag, it is 0.67% and 4.3%, respectively. Thus, although the
mesh resolutions used are relatively low (considering full-scale simula­
tion), overall uncertainty in resistance prediction remains low.
Furthermore, since a decrease in mesh resolution increases the resis­
tance prediction, the application of a relatively low mesh resolution for
Fig. 16. Power prediction for the 64TEU inland vessel while passing through further studies should not be an issue. Just to be certain, another
restricted waters at varying depths and drafts, at 5.70 knots velocity. simulation was run using a 30 million cell resolution with a y + value of
around 200, which produced a power prediction of 770 KW. As such, the
predict the required power. moderate mesh resolution of 3.39 million was chosen for subsequent
In general, following the results from this particular study, ITTC-57 simulations.
over predicts the viscous resistance coefficient in model scale for all As mentioned before, simulations are repeated in full scale only for
cases comparing to CFD results. Thus, it may be assumed that ITTC-57 the maximum resistance cases observed in model scale simulations. In
over predicts the values for full scale as well. As such, to avoid over total, eight cases are repeated, calm water simulations in open water for
prediction of required power for the 64TEU inland container vessel, draft 3 and 4, at design speed (10.1 knots); static drift simulation at a 15
model simulations that predicted maximum resistance are repeated in
◦
drift angle at design speed for draft 3 and 4; and restricted water
full scale. However, all simulations performed on the full scale are static simulations for draft 3 and 4, at 5.6 knots speed, h/D = 1.1, and channel
simulations, thus heave and pitch motion are restricted. width of 60 m and 45 m. Initially restricted water simulations for draft 3
Initially, a grid dependency analysis is performed to investigate how and 4, at 5.6 knots speed, and h/D = 1.1, for a single way width of 30 m,
resistance prediction changes with changing mesh resolution. The show very high power requirements, implying that the vessel would

Table 12
Grid dependency analysis for full-scale simulation of 64TEU vessel at design speed and draft.
Mesh Total resistance Viscous Pressure Total drag Viscus drag Pressure drag Required Required
resolution Force, F_tot (N) resistance, F_vis resistance, F_Pr coefficient, Ct coefficient, Cv coefficient, Cp Effective Power Installed Power
(million) (N) (N) (W) (KW)

9.35 − 64966.7 − 28455.3 − 36511.4 4.17E-03 1.82E-03 2.34E-03 337.83 776.2

3.39 − 66635.1 − 28761.8 − 37873.3 4.27E-03 1.84E-03 2.43E-03 346.50 796.1
1.73 − 67043.0 − 28354 − 38689 4.30E-03 1.82E-03 2.48E-03 348.62 801.0

11
H. Islam et al. Ocean Engineering 229 (2021) 108621

have to substantially reduce its speed while traveling through such 3.4.1. Final recommendation
narrow and shallow channel. This is mainly because the bow waves Overall, the simulation results suggest that an installed capacity of
generated by the hull in shallow and narrow channel reflects the hull 1200 KW or 1600HP should prove sufficient for the proposed vessel to
and creates added resistance. Further investigation for such cases using safely and comfortably maneuver through the restricted inland waters of
dynamic conditions (heave and pitch-free motion) would provide better the intended sites in China. The vessel should be able to operate
information in this regard. For the present study, the single way (30 m) smoothly at its design speed in a relatively open channel and during
simulation results are not included in the comparison. The simulation small maneuvering motion. However, it would have to reduce the speed
results are shown in Table 13 and comparison among power prediction while passing through restricted channels, and during sharp maneuvers.
for the cases in full scale vs power predicted by the model scale is shown The power requirement would also depend on the average current ve­
in Fig. 17. locity in the channel. As can be realized from the simulation results, if
The results show that for restricted channel cases, the model scale the proposed channel has a current higher than 4 knots in very shallow
simulations (60 m equivalent width, and h/D = 1.1) predict that pres­ and narrow areas, the required installed power would be higher.
sure drag will decrease with an increasing draft, whereas, the viscous
drag formulation by Millward (1989) suggests that viscous drag would 4. Conclusion
increase with ship draft. However, the full-scale restricted channel
simulation predicts that for draft 4, viscous drag also decreases. The A substantial number of simulations are performed for a proposed
full-scale results also contradict model scale results, which is interesting. inland container vessel to determine its required propulsion power.
Further investigation is required to understand the issue in detail. Initially, model scale dynamic simulations are performed in calm open
However, overall, the difference between the two cases remains small, waters in straight and drift conditions, next static simulations are per­
and as can be seen from Fig. 15, the deciding factor here for power formed for restricted channel conditions. Full-scale power predictions
would be not just the depth but also the width of the channel. If the are made for all the results. Finally, following the assessment from
vessel faces two-way traffic and is forced to travel close to one of the model scale results, full-scale simulations are performed for the identi­
banks, instead of taking the middle path, the total drag significantly fied maximum resistance cases. Following the results, the required
increases. Thus, with lower installed power, the vessel would propel at a installation power of 1200 KW is recommended for the vessel.
substantially reduced speed in narrow channels. The calm water simulations showed that while the vessel’s design
The overall results from model scale and full-scale simulations sug­ speed is 10 knots, the vessel should perform relatively better at 7.8
gest that, in general, the model scale simulations slightly over predict knots, in terms of efficiency. As expected, the vessel shows the highest
the pressure drag, and the ITTC formulation for viscous drag signifi­ resistance in case of the highest speed and highest draft. Linear de­
cantly over predict the viscous drag in full scale. As such, the model scale rivatives calculated from the static drift results indicate that the vessel
converted results over predict the overall power requirement by around has relatively less lateral stability, however, the yaw stability is high.
20% for calm water simulations, by around 10% for drift cases, and by The lateral stability gets slightly compromised with an increasing draft.
around 20% for restricted channel cases. This over the installation of The static drift simulation results also indicate that the consideration of
power would increase the initial cost of the vessel, would passively combined sinkage and trim effect of the vessel during maneuvering
promote operations in overcapacity, and produce higher emissions. might be important during navigating through shallow waters. The
Flow field visualization for full-scale simulations are shown in power prediction shows that the drift angle has a limited influence on
Figs. 18–21. All figures are plotted using the same pressure range values. propulsion power requirement and only registers notable variation for
Fig. 18 shows pressure distribution on the free surface and the hull form relatively high draft cases. Finally, the restricted water simulations show
during calm water simulation at draft 3 and 4, at speed 10.1 knots. that both channel depth and width significantly influences the resistance
Fig. 19 shows the same for a drift angle of 15◦ . Fig. 20 shows restricted encountered by the vessel, and propulsion power requirement at shallow
water simulation cases for 60 m channel width with 5.6 knots velocity, water conditions is significantly larger compared to open channel cases.
whereas, Fig. 21 shows so for 45 m channel width with the vessel being For the same installed power, the vessel would face a sharp reduction in
placed at aside. Comparing to Figs. 18, Figs. 20 and 21 show a sub­ speed while maneuvering through shallow restricted channels,
stantial increase in drag force due to channel restriction effects, while comparing to open waters. Thus, consideration of channel current is
Fig. 19 shows a significant increase in drag at the side of the bow and crucial while predicting the propulsion power requirement.
stern due to drifting motion. Finally, the study observes that the old methods developed to convert
model scale tests to full-scale results are outdated (as stated in several
other previous researchers). Considering resource requirements, model

Table 13
Propulsion power prediction for the 64TEU inland container vessel using full-scale simulations for different operating conditions.
Cases/Draft Velocity Total resistance Viscous Total drag Viscus drag Pressure drag Required Required
(knots) Force, F_tot (N) resistance, F_vis coefficient, Ct coefficient, Cv coefficient, Cp Effective Power Installed Power
(N) (KW) (KW)

Open/D3 10.1 − 66635.1 − 28761.8 4.27E-03 1.84E-03 2.43E-03 346.50 796.1

Open/D4 10.1 − 78270.4 − 30110.7 4.76E-03 1.83E-03 2.93E-03 407.01 935.1
Drift_15deg/ 10.1 − 95314.1 − 30495.2 6.11E-03 1.96E-03 4.16E-03 495.63 1138.75
D3
Drift_15deg/ 10.1 − 124095 − 32623.9 7.54E-03 1.98E-03 5.56E-03 645.30 1482.61
D4
Rest._60 m/ 5.6 − 1.5 E+05 − 15077.6 3.02E-02 3.04E-03 2.72E-02 440.25 1011.5
D3
Rest._60 m/ 5.6 − 1.3 E+05 − 15779.4 2.54E-02 3.01E-03 2.24E-02 391.01 898.4
D4
Rest._45 m/ 5.6 − 2.5 E+05 − 19401 5.03E-02 3.91E-03 4.64E-02 733.75 1685.8
D3
Rest._45 m/ 5.6 − 2.40 E+05 − 20177 4.58E-02 3.85E-03 4.20E-02 705.33 1620.6
D3

12
H. Islam et al. Ocean Engineering 229 (2021) 108621

Fig. 17. Relative comparison among pre­

dicted propulsion power for the 64TEU
inland container vessel form full-scale sim­
ulations and model scale simulations at
different operating conditions (The first two
cases are for draft 3 and 4, at 10.1 knots
speed, in open water; the second two cases
are for draft 3 and 4, at 10.1 knots speed, at
a 15◦ drift angle in open water; the next two
are for draft 3 and 4, at 5.6 knots speed, with
a depth to draft ratio of 1.1 and channel
width of 60 m; the last two are for draft 3
and 4, at 5.6 knots speed, with a depth to
draft ratio of 1.1 and channel width of 45 m,
vessel placed at a side).

Fig. 18. Pressure distribution at the free surface and side of the 64TEU vessel hull at 10.1 knots speed, open water, at draft 3 (left), and draft 4 (right).

Fig. 19. Pressure distribution at the free surface and side of the 64TEU vessel hull at 10.1 knots speed, 15-degree drift, at draft 3 (left), and draft 4 (right).

Fig. 20. Pressure distribution at the free surface and side of the 64TEU vessel hull at 5.6 knots speed, restricted channel (60 m width, h/D = 1.1), at draft 3 (left), and
draft 4 (right).

13
H. Islam et al. Ocean Engineering 229 (2021) 108621

Fig. 21. Pressure distribution at the free surface and side of the 64TEU vessel hull at 5.6 knots speed, restricted channel (45 m width, h/D = 1.1), at draft 3 (left), and
draft 4 (right).

scale simulations can serve the purpose of getting insights regarding Hinostroza, M.A., Xu, H.T., Guedes Soares, C., 2019. Cooperative operation of
autonomous surface vehicles for maintaining formation in complex marine
different configurations and operating conditions. However, for power
environment. Ocean Eng. 183, 132–154.
prediction, full-scale simulation for particular cases is essential. As such, Hoydonck, W.V., Toxopeus, S.L., Eloot, K., Bhawsinka, K., Queutey, P., Visonneau, M.,
with the availability of greater computational power, both academia and 2018. Bank effects for KVLCC2. J. Mar. Sci. Technol. 24, 174–199.
industry should target full-scale simulations instead of model scale ones, Islam, H., Akimoto, H., 2018. Simulation dependency on degrees of freedom in RaNS
solvers for predicting ship resistance. In: Maritime Transportation and Harvesting of
at least for selected cases, to get a better estimate for resistance Sea Resources: Guedes Soares, Teixeira. Taylor & Francis Group, London, UK, 2018.
prediction. Islam, H., Guedes Soares, C., 2018. Estimation of hydrodynamic derivatives of a
container ship using PMM simulation in OpenFOAM. Ocean Eng. 164, 414–425.
Islam, H., Guedes Soares, C., 2019. Uncertainty analysis in ship resistance prediction
CRediT authorship contribution statement using OpenFOAM. Ocean Eng. 191, 105805.
Islam, H., Kan, J., Liu, J., Wang, X., Guedes Soares, C., 2021. Investigation of the
H. Islam: Methodology, Formal analysis, Writing - original draft. C. hydrodynamic properties of an inland container vessel. In: Guedes Soares, C.,
Santos, T.A. (Eds.), Progress in Maritime Technology and Engineering. Taylor &
Guedes Soares: Writing - review & editing, Supervision. J. Liu: Writing Francis Group, London, UK, 2020.
- review & editing. X. Wang: Data curation, hull data generation. ITTC-, 1957. 8th International Towing Tank Conference. Madrid, Spain.
ITTC-, 2002. ITTC-recommended Procedures: Testing and Extrapolation Methods
Resistance Test, 7.5-02-02-01.
Declaration of competing interest ITTC-, 2011. ITTC- Recommended Procedures and Guidelines: Practical Guidelines for
Ship CFD Application, 7.5-03-02-03.
The authors declare that they have no known competing financial ITTC-, 2014. ITTC- Recommended Procedures and Guidelines: Practical Guidelines for
Ship Resistance CFD, 7.5-03-02-04.
interests or personal relationships that could have appeared to influence
Jasak, H., 2009. OpenFOAM: open source CFD in research and industry. International
the work reported in this paper. Journal of Naval Architecture and Ocean Engineering 1 (2), 89–94.
Kim, H., Akimoto, H., Islam, H., 2015. Estimation of the hydrodynamic derivatives by
RaNS simulation. Ocean Eng. 108, 129–139.
Acknowledgments
Kim, J., Park, I.-R., Kim, K.-S., Kim, Y.-C., Kim, Y.-S., Van, S.-H., 2013. Numerical towing
tank application to the prediction of added resistance performance of KVLCC2 in
The first author was funded by the project RESET (REliability and regular waves. In: Proceedings of the Twenty-Third (2013) International Offshore
Safety Engineering and Technology for large maritime engineering and Polar Engineering (ISOPE) Anchorage. Alaska, USA, June 30-July 5, 2013.
Kretschmann, L., Burmeister, H.C., Jah, C., 2017. Analyzing the economic benefit of
systems), which is partially financed by the European Union Horizon unmanned autonomous ships: an exploratory cost-comparison between an
2020 research and innovation program, under the Marie Skodowska- autonomous and a conventional bulk carrier. Research in Transportation Business &
Curie grant agreement No. 73088 RESET, which allowed his visit to Management 25, 76–86.
Larsson, L., Stern, F., Bertram, V., 2003. Benchmarking of computational fluid dynamics
the Wuhan University of Technology. This work contributes to the for ship flows: the Gothenburg 2000 workshop 2003. J. Ship Res. 47, 63–81.
Strategic Research Plan of the Centre for Marine Technology and Ocean Larsson, L., Stern, F., Visonneau, M., 2011. CFD in ship hydrodynamics- results of the
Engineering (CENTEC), which is financed by the Portuguese Foundation Gothenburg 2010 workshop. In: Eça, L., et al. (Eds.), MARINE 2011, IV International
Conference on Computational Methods in Marine Engineering. Computational
for Science and Technology (Fundação para a Ciência e Tecnologia - Methods in Applied Sciences, 2011.
FCT) under contract UIDB/UIDP/00134/2020. This work was also Lee, S., Hong, 2017. Study on the course stability of very large vessels in shallow water
supported by National Key R&D Program of China (2018YFB1601503, using CFD. Ocean Eng. 145, 395–405.
Liu, J., Hekkenberg, R., Rotteveel, E., Hopman, H., 2015. Literature review on evaluation
2018YFB1601505), National Natural Science Foundation of China
and prediction methods of inland vessel manoeuvrability. Ocean Eng. 106, 458–471.
(51709217). Liu, Jialun, 2017. Impacts of Rudder Configuration on Inland Vessel Manoeuvrability.
PhD Thesis. Delft University of Technology.
Marine Insight, 2019 [Online] Available at: https://www.marineinsight.com/naval-arch
References
itecture/power-requirement-ship-estimated/. Accessed 2020.
MASHCON, 2009-19 [Online] Avaliable at: http://www.shallowwater.ugent.be/EN/k
Celik, I.B., Ghia, U., Roache, P.J., Freitas, C.J., Coleman, H., Raad, P.E., 2008. Procedure c_conf_mashcon5.htm. Accessed 2020.
for estimation and reporting of uncertainty due to discretization in CFD applications. Millward, A., 1989. The effect of water depth on hull form factor. Int. Shipbuild. Prog. 36
J. Fluid Eng. 130 (7), 078001. (407), 283–302.
EU Emission Report, 2020. 2019 Annual Report on CO2 Emissions from Maritime Min, K.-S., Kang, S.-H., 2010. Study on the form factor and full-scale ship resistance
Transport. SWD(2020) 82 Final. Brussels. prediction method. J. Mar. Sci. Technol. 15, 108–118.
Hajivand, A., Mousavizadegan, S.H., 2015a. Virtual maneuvering test in CFD media in Molland, A.F., Turnock, S.R., Hudson, D.A., 2011. Ship Resistance and Propulsion -
presence of free surface. International Journal of Naval Architecture and Ocean Practical Estimation of Ship Propulsive Power. Cambridge University Press,
Engineering 7, 540–558. Cambridge, USA.
Hajivand, A., Mousavizadegan, S.H., 2015b. Virtual simulation of maneuvering captive Moreira, L., Fossen, T.I., Guedes Soares, C., 2007. Path following control system for a
tests for a surface vessel. International Journal of Naval Architecture and Ocean tanker ship model. Ocean Eng. 34, 2074–2085.
Engineering 7, 848–872. Niklas, K., Pruszko, H., 2019. Full-scale CFD simulations for the determination of ship
Hino, T., Kenkyūjo, K.G., 2005. The Proceedings of CFD Workshop, Tokyo, 2005. resistance as a rational, alternative method to towing tank experiments. Ocean Eng.
National Maritime Research Institute, Japan, 2005. 190, 106435.
Hino, T., Stern, F., Larsson, L., Visonneau, M., Hirata, N., Kim, J., 2021. Numerical Ship Perera, L.P., Carvalho, J.P., Guedes Soares, C., 2012. Intelligent ocean navigation and
Hydrodynamics - an Assessment of the Tokyo 2015 Workshop. Springer fuzzy-bayesian decision-action formulation. IEEE J. Ocean. Eng. 37 (2), 204–219.
International Publishing.

14
H. Islam et al. Ocean Engineering 229 (2021) 108621

Perera, L.P., Ferrari, V., Santos, F.P., Hinostroza, M.A., Guedes Soares, C., 2015. Wu, B., Tian, H.B., Yan, X.P., Guedes Soares, C., 2020. A probabilistic consequence
Experimental evaluations on ship autonomous navigation and collision avoidance by estimation model for collision accidents in the downstream of Yangtze River using
intelligent guidance. IEEE J. Ocean. Eng. 40 (2), 374–387. Bayesian Networks. Journal of Risk and Reliability 234 (2), 422–436.
Sadat-Hosseini, H., Wu, P.-C., Carrica, P.M., Kim, H., Toda, Y., Stern, F., 2013. CFD Xu, H.T., Hinostroza, M.A., Wang, Z., Guedes Soares, C., 2020. Experimental
verification and validation of added resistance and motions of KVLCC2 with fixed investigation of shallow water effect on vessel steering model using system
and free surge in short and long head Waves. Ocean Eng. 59, 240–273. identification method. Ocean Eng. 199, 106940.
Santos, T.A., Guedes Soares, C., 2018. Economic feasibility of an autonomous container Yao, J., Jin, W., Song, Y., 2016. RANS simulation of the flow around a tanker in forced
ship. In: Guedes Soares, C., Teixeira, A.P. (Eds.), Maritime Transportation and motion. Ocean Eng. 127, 236–245.
Harvesting of Sea Resources. Taylor & Francis Group, London. UK, pp. 861–870. Zhang, J.F., Teixeira, A.P., Guedes Soares, C., Yan, X.P., Liu, K.H., 2016. Maritime
Shipping and World Trade, 2020 [Online] Available at: https://www.ics-shipping.org/ transportation risk assessment of Tianjin port with bayesian belief networks. Risk
shipping-facts/shipping-and-world-trade. Accessed 2020. Anal. 36 (6), 1171–1187.
SIMMAN, 2008. [Online] Available at: [Online] Available at: http://www.simman2008. Zhang, J.F., Zhang, D., Yan, X.P., Haugen, S., Guedes Soares, C., 2015. A distributed anti-
dk/2008. Accessed 2020. collision decision making formulation in multi-ship encounter situations under
SIMMAN, 2014 [Online] Available at: https://simman2014.dk/2014. Accessed 2020. COLREGs. Ocean Eng. 105, 336–348.
Tokyo, 2015 [Online]. Available at: https://t2015.nmri.go.jp/ Accessed: August 2020. Zhang, Y.-Y., Zou, Z.-J., Yao, J.-X., 2020. RANS simulation of the flow around a ship
Terziev, M., Tezdogan, T., Oguz, E., Gourlay, T., Demirel, Y.K., Incecik, A., 2018. advancing in shallow water. In: The Proceedings of the 39th International Conference
Numerical investigation of the behaviour and performance of ships advancing on Ocean, Offshore and Arctic Engineering. OMAE2020. August 3-7, 2020, Virtual,
through restricted shallow waters. J. Fluid Struct. 76, 185–215. Online.
Terziev, M., Tezdogan, T., Incecik, A., 2019. A geosim analysis of ship resistance Zeng, Q., Thill, C., Hekkenberg, R., Rotteveel, E., 2018. A modification of the ITTC57
decomposition and scale effects with the aid of CFD. Appl. Ocean Res. 92, 101930. correlation line for shallow water. J. Mar. Sci. Technol. 24, 642–657.
Toxopeus, S.L., Simonsen, C.D., Guilmineau, E., Visonneau, M., Xing, T., Stern, F., 2013. Zeng, Q., Hekkenberg, R., Thill, C., 2019. On the viscous resistance of ships sailing in
Investigation of water depth and basin wall effects on KVLCC2 in manoeuvring shallow water. Ocean Eng. 190, 106434.
motion using viscous-flow calculations. J. Mar. Sci. Technol. 18, 471–496. Zeng, Q., Hekkenberg, R., Thill, C., Hopman, H., 2020. Scale effects on the wave-making
Wang, H.M., Xie, Y., Liu, J.M., Zou, Z.J., He, W., 2011. Experimental and numerical resistance of ships sailing in shallow water. Ocean Eng. 212, 107654.
researches on the viscosity hydrodynamic in hydrodynamic forces acting on a Zhou, X.Q., Sutulo, S., Guedes Soares, C., 2016a. A paving algorithm for dynamic
KVLCC2 model in oblique motion. In: Proceedings of the International Conference in generation of quadrilateral meshes for online numerical simulations of ship
Remote Sensing. Environment and Transportation Engineering (RSETE), manoeuvring in shallow water. Ocean Eng. 122, 10–21.
pp. 328–331. Zhou, X.Q., Sutulo, S., Guedes Soares, C., 2016b. Ship-Ship hydrodynamic interaction in
Wneck, A.D., Sutulo, S., Guedes Soares, C., 2018. CFD analysis of ship-to-ship confined waters with complex boundaries by a Panelled Moving Patch Method. RINA
hydrodynamic interaction. J. Mar. Sci. Appl. 17 (1), 21–37. International Journal of Maritime Engineering 158. A-21 - A30.

15