10 1016@j Oceaneng 2019 106696

Ocean Engineering xxx (xxxx) xxx
Contents lists available at ScienceDirect
Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng
Numerical and experimental study on hydrodynamic performance of ships

advancing through different canals
Khaled Elsherbiny a, b, *, Momchil Terziev b, Tahsin Tezdogan b, Atilla Incecik b, Mohamed Kotb a
a
Department of Marine Engineering, College of Engineering and Technology, Arab Academy for Science, Technology and Maritime Transport, Abo Kir, Alexandria, P.O.
Box 1029, Egypt
b
Department of Naval Architecture, Ocean and Marine Engineering, Henry Dyer Building, University of Strathclyde, 100 Montrose Street, Glasgow, G4 0LZ, UK
A B S T R A C T
In international shipping, there are several waterways that are widely viewed as bottlenecks. Among these is the Suez Canal, where recent expansions have taken
place. Although the Suez Canal has a high importance in international shipping, little research has been carried out in maximising the number of ships capable of
traversing for a set period of time. The present study aims to examine hydrodynamic phenomena of ships advancing through the Suez Canal in the allowed speed
range to determine the relative effects of the canal depth and/or width restrictions on the overall ship sailing performance. A rectangular canal is also included as a
reference to gauge the effects of varying canal cross-section. The present study combines experimental, numerical, analytical and empirical methods for a holistic
approach in calm water. As a case-study, the KCS hullform is adopted, and analysed experimentally, via Computational Fluid Dynamics, using the slender body
theory, and empirical formulae. The results reveal strong coupling between the canal’s cross section and all examined parameters.
1. Introduction Hekkenberg, 2015). Empirical formulae can quickly estimate the squat
according to the ship dimensions, coefficients, speed, and underwater
When a ship enters shallow waters, it has been observed that the topology. These formulae are typically obtained from a series of model
distance between the keel of the ship and the seabed decreases as the tests. Alternatively, analytical methods have been developed by re
speed increases, and on occasion, the ship has been known to strike the searchers that make use of the assumptions inherent in potential flow
bottom. This phenomenon is known as ‘ship squat’ (Constantine, 1960). theory such as slender body theory (Tuck, 1966). Namely, the flow is
Because of the Bernoulli effect the free water surface around the ship inviscid, irrotational and incompressible. Naturally, experimental in
drops. There is a vertical motion downwards and trim resulting in a vestigations provide the most accurate results, and should be performed
decrease in the under-keel clearance (UKC). Thus all vessels squat as if possible (el Moctar et al., 2012; Zeraatgar et al., 2011). However, the
they move, even in deep water, although the magnitude of the squat is availability of testing facilities, time required to perform the experiment,
usually small. This phenomenon is a function of the shape of the hull and and the cost are highly prohibitive.
the forward speed through the water (Millward, 1996). Finally, numerical, or Computational Fluid Dynamics (CFD) methods
Ship squat is increased further when a vessel enters confined chan have been more recently developed and can be easily used for predicting
nels or rivers, since the velocity of water must increase due to the greater ship squat. Although this approach has its own limitations, such as the
degree of restriction. Furthermore, a significant increase in resistance computational cost, and knowledge required to perform a simulation, it
arises due to this hydrodynamic interaction. A drop in speed in the re is rapidly gaining popularity (Stern et al., 2013).
gion of 30% can be expected upon entering shallow waters, which and The present study combines all methods mentioned above. To elab
may rise up to 60% if the ship is advancing through a confined channel orate, empirical formulae applicable in canal case-studies are used. The
such as a river or a canal (Barrass and Derrett, 2012). sinkage and trim of the ship are also analysed via the slender body
Much research has been devoted to accurately predict ship squat, theory as it is one of the most widely applied methods in this context.
leading to a plethora of methods and approaches to the problem. Some CFD simulations are also performed to replicate the experimental results
of these are empirical methods, which are easy to use, but are reliable and demonstrate the versatility of the numerical method. The analysis is
only for an early design stage (Barrass and Derrett, 2012; Rotteveel and performed on the KCS hullform, for which experimental data in shallow
* Corresponding author. Department of Marine Engineering, College of Engineering and Technology, Arab Academy for Science, Technology and Maritime
Transport, Abo Kir, Alexandria, P.O. Box 1029, Egypt.
E-mail address: [email protected] (K. Elsherbiny).
https://doi.org/10.1016/j.oceaneng.2019.106696
Received 3 August 2019; Received in revised form 4 November 2019; Accepted 6 November 2019
0029-8018/© 2019 Elsevier Ltd. All rights reserved.
Please cite this article as: Khaled Elsherbiny, Ocean Engineering, https://doi.org/10.1016/j.oceaneng.2019.106696
K. Elsherbiny et al. Ocean Engineering xxx (xxxx) xxx
water is available for the specific case-studies examined herein Kreitner (1934) calculated ship squat using a one-dimensional hydraulic
(Elsherbiny et al., 2019b, 2019a). theory. He showed that the equation for the flow velocity in a canal
This study is organised as follows. Section 2 presents the background ceases to provide reasonable solutions as the critical speed is
of the methods used as part of this research. Section 3 contains the approached. Constantine (1960) studied the different behaviour of ship
specific details relating to the implementation of each method. Section 4 squat for various ship speeds (subcritical, critical and supercritical), as
is devoted to the obtained results and relevant discussion. Finally, the well as the ratio of midship section to the cross section of the fairway. He
conclusion and recommendations for future work are given in Section 5. determined that laterally restricted waterways have substantial effect on
the dynamic squat of a vessel. Inui (1954) investigated the effect of
2. Background shallow water and restricted water on the wave-making resistance of
ships. He concluded that higher degrees of discontinuity in the
There are four main parameters influencing ship squat. These are the wave-resistance of a ship are caused by an increase in the restrictions of
blockage factor (K), the block coefficient (CB), the ship’s velocity (V), a waterway. This showed that the resistance itself is not a continuous
draft (T) and water depth (h). The blockage factor can be defined as the function of depth Froude number in the case of restricted shallow wa
ratio of the submerged midship cross-sectional area and the underwater ters. Tuck (1966) reproduced Michell’s linearised slender body theory
area of the canal or channel (Fig. 1). This dimensionless parameter is and explored the scenario where a ship is travelling in shallow water of
utilised in calculating ship squat by empirical formulations, and is constant, unrestricted width. Tuck (1966) solved for the hydrodynamic
shown in Eq. (1): forces in shallow water to calculate the sinkage and trim for sub-critical
speed (Fnh < 1) and super-critical speed (Fnh > 1) by using matched
B � T � CM
K¼ (1) asymptotic expansions. He validated the results with model-scale ex
W �h
periments which showed good agreement for depth Froude number
where B is the ship’s breadth, T is the ship’s draft, W is the canal’s smaller than 0.7. However, the theory fails as depth Froude number
breadth, CM is the midship area coefficient and h is the water depth of approaches 1 because the formulations used become singular. Trim is
the water. the leading factor in the supercritical range, while sinkage is predomi
Michell (1898) developed a thin-body method to predict the wave nant in the subcritical range (Tuck, 1966). The method predicts zero
resistance of a ship moving in shallow water. This method, henceforth wave-making resistance in the subcritical range. Tuck (1967a) analysed
referred to as the slender body theory, is based on fundamental the effect of restricted channel width in addition to depth on ship
assumption that the ship’s beam is small compared to its length. As a behaviour. Beck et al. (1975) expanded on the Tuck’s work to account
consequence, the waves generated are also of small amplitude, which for vessels in dredged canals with an unrestricted shallow water region
allows the linearisation of the free water surface. Later, Joukovski of constant depth extending infinitely on either side of the dredged
(1903) derived a similar formulation of the problem independently. The section of the channel.
changes in a ship’s wave pattern have been studied by Havelock (1924) Gourlay (2008a) obtained a general Fourier transform method to
for a point pressure impulse travelling over a free water surface. Have calculate the sinkage and trim of a ship advancing in unrestricted
lock examined the wave patterns in shallow water by taking into account shallow waters, canals and stepped channels as well as channels of
the speed of the vessel, and the depth of water, which led to the intro arbitrary cross-section. He focussed on the subcritical range of motion.
duction of the depth Froude number (Fnh ), Eq. (2). Gourlay (2008b) extended his modification of the Slender-body theory
to calculate the sinkage and trim of a fast displacement catamaran
V
Fnh ¼ pffiffiffiffiffi (2) propagating through horizontally unrestricted shallow water, which was
gh valid for all speed regimes. Then, Gourlay (2008b) showed that trim,
resistance and sinkage are affected by a change in the spacing between
where V is the speed of the vessel (m/s), g is the acceleration due to
the catamaran hulls. In addition Gourlay (2009) used a theoretical
gravity (m/s2) and h is the water depth.
method based on the linear superposition of slender-body shallow-water
The depth Froude number can be thought of as the ratio of the ship’s
flow solutions to predict the sinkage and trim of two moving ships as
speed to the maximum wave velocity in shallow water of depth h. The
they pass each other, either from opposite directions, or one ship over
well-known Kelvin wave pattern resulting from moving objects in water
taking the other.
can be observed for Fnh < 0.57 (Tezdogan et al., 2015). As the ship’s
Alderf et al. (2008) developed a new method for the numerical
velocity increases, the angle between the wave pattern and the ship’s
modelling of dynamic squat by using a finite element method. They also
centreline theoretically approaches 90� until Fnh becomes 1 (Tunaley,
illustrated the effect of sea floor topology on a ship sailing at critical
2014), which is called the critical speed. When the depth Froude number
speed. Their model can give results for the dynamic responses of a ship
is greater than one, the angle of the wave pattern begins to decrease
in highly restricted canals on any seafloor shape. Alderf et al. (2008)
again. The terms subcritical and supercritical speed are used for vessels
developed their approach to validate the stability model as an extension
propagating at Fnh < 1 and Fnh > 1, respectively. Of greater practical
of the method proposed by Janssen and Schijf (1953), who predicted the
interest is the former scenario, namely when the depth Froude number is
unstable squat positions for ships. Yao et al. (2011) produced and tested
smaller than 1 (Beck et al., 1975).
their theory for a Series 60 hull (CB ¼ 0.6). They calculated the shallow
Many researchers have investigated ship squat in restricted water.
water effects in terms of sinkage, trim, resistance and wave patterns for
sub- and supercritical speeds on a ship by discretising the hull by a panel
method. They distributed free and wall surface panels onto which
Rankine sources of constant strength are mapped. Yao et al. (2011)
validated their results with experimental data which showed good
agreement.
Calisal and Alidadi (2011) performed a potential flow-based study to
predict the squat of the Wigley hull. A slender-body theory approach
was utilised to convert the three-dimensional problem into a series of
2-D cross sections distributed from the bow to the stern at equal in
tervals. They applied a boundary element method sequentially to each
cross section, which consisted of a solution to the summation of the
Fig. 1. Ship in a canal. disturbance and free-stream potentials. Thus, they obtained the velocity
2
potential function expressing the problem at hand. To predict the ship 3.1. Empirical methods
squat, the pressure integration approach was used. This consists of
integrating the pressure, acting on the hull to predict the hydrodynamic Empirical methods are typically derived based on a regression
forces. To validate their theory, Calisal and Alidadi (2011) compared technique, employed on a dataset. A common problem with this
their predictions with experimental data. Interestingly, the comparison approach is that upon extrapolating variables beyond the range con
revealed that the predicted data does not deviate significantly from the tained within the dataset introduces errors. The abovementioned dataset
experimentally predicted values at higher speeds. Instead, agreeable is usually experimental (Duffy, 2008). In the realm of ship hydrody
comparison was established for a range of depth Froude numbers from namics, the number of parameters, one must account for are high,
0.2 to 0.4. rendering the problem difficult. A superimposed issue is that a slight
Lataire et al. (2012) conducted an experimental study for a model alteration in any of the modelled (or otherwise) independent parameters
scale KVLCC2 to predict the squat for a wide range of water depths and induces a substantially different case-study than what the empirical
widths of a canal with rectangular cross section. Ji et al. (2012) per method is suited for. For example, introducing a slope in the bathymetry
formed numerical simulations by solving the 3-D Navier-Stokes equa changes the flow physics, versus a rectangular canal.
tions along with the standard k-ε turbulence model. They simulated the In the present context, the hullform is also of critical importance.
wave patterns, induced by moving convoys composed of one or two Slight variations in the wetted area of the ship can have dramatic con
barges in restricted waterways in order to predict the relationship be sequences in terms of the behaviour and performance of a ship (Tez
tween geometric and kinematic parameters, as well as the amplitude of dogan et al., 2016). Coincidently, this is also the main motivation of hull
ship-generated waves, and the water plane drawdown. Sergent et al. form optimisation studies (Zhang et al., 2018a, 2018b, 2017). A change
(2015) estimated the unstable equilibrium position of a ship during in depth or width also invalidates any analysis that was not performed
heave motions as a function of canal and ship parameters by using a new using a similar set-up. This is the main problem of empirical methods:
mathematical expression from a 2D analytical model. Tezdogan et al. their predictions may be excellent for the case-study used in their
(2016) predicted the squat and resistance of a model scale container ship inception, but this is far from the case when applied elsewhere. This
advancing in a canal using a numerical method based on nonlinear drawback stems from the range and number of parameters required to
unsteady RANS simulations. describe the flow around a ship. A simple length to beam, draught, depth
Terziev et al. (2018) presented a numerical study to investigate the etc. ratio is insufficient to provide information on the actual flow
sinkage, trim and resistance of ships by using a scale model of the DTC characteristics. For our purposes, the Suez Canal is modelled as shown in
container ship advancing through restricted shallow water in four Fig. 2, and explained in the following section (Section 4.2).
channels with different cross sectional area and ship speeds. They used As part of this research, we have employed several empirical
Computational Fluid Dynamics (CFD), the slender body theory and formulae derived from experimental databases. Naturally, only those
various empirical methods to calculate the trim and squat of the DTC applicable to canal case-studies are examined. For a complete descrip
advancing through different channel geometries. tion of the mathematical background, the reader is referred to Briggs
Elsherbiny et al. (2019) conducted a series of experiments on a model (2009, 2006), Briggs et al. (2013, 2009b), and Terziev et al. (2018). A
of the KRISO Container Ship (KCS). The KCS performance was examined description of the mathematical background of the empirical formulae is
by measuring its sinkage, trim and total resistance. A wide range of given in the Appendix.
water depth to ship draft ratios at various ship speeds were investigated.
Additionally, the blockage effect was studied by varying the canal width. 3.2. Slender body theory
Also, the measured model resistance data was used to determine a form
factor value for the KCS at various water depth to ship draft ratios. The The slender body theory is one of the most successful approaches to
estimation of experimental uncertainty was conducted for all tests. computing shallow water trim and sinkage. This is particularly true in
Later, Elsherbiny et al. (2019a) presented a series of model tests the low speed range, because viscous effects play a secondary role
measuring the resistance, sinkage and trim variations with speed, water (Dand, 1967). The aforementioned theory, utilising a velocity potential
depth and loading conditions under different trim angles at 1:75 scale. function, renders it unable to compute viscous contributions. As such,
This was done to examine the range of ship trim for safe and efficient low speeds and moderate depths are well-suited for predictions using the
sailing in restricted water in both depth and width, and to detect the best slender body theory. In the present study, an in-house code which solves
trim angle for ships sailing in restricted waters to reduce resistance and the equations of Tuck (1966), Tuck (1967), and Beck et al. (1975) is
therefore fuel consumption. used. The debut of the code was presented in Terziev et al. (2019a) for
Based on the above literature, the vast majority of studies focus on dredged channels and canals. The method avoids integration of highly
rectangular cross section canals. However, to the best of our knowledge, oscillatory integrals, representing the Fourier transform of the ship’s
the Suez Canal has not been incorporated and assessed in any study cross-sectional area and beam (Gourlay, 2014; Tuck, 1967b) by using
using empirical, analytical, numerical and experimental techniques. The the convolution form of the governing equations (Tuck, 1967a, 1966).
present paper will attempt to fill this gap by modelling the Suez Canal This is done because, as stated by Gourlay (2014), the convolution form
and a rectangular canal for reference. The KCS hull form is modelled to of the equations is favourable for practical applications. The input data
advance through the aforementioned canals. To predict the wave for KCS, as modelled for the slender body theory is shown in Fig. 3 using
resistance and form factor of the ship, CFD simulations are run in both a scale factor of 1:75. The principal characteristics of the ship are shown
multiphase and double body conditions. The details relating to each in Section 5.1.
method are presented in the following section sequentially. The theory begins by supposing that the flow is two dimensional, and
satisfies Laplace’s equation:
3. Empirical, analytical and numerical methods
∂2 ϕ ∂2 ϕ
ð1 Fnh Þ þ ¼0 (3)
As referred to in Section 1, experimentation has several drawbacks. ∂x2 ∂y2
In an attempt to circumvent the use of expensive and time consuming
where ϕ is used to denote the velocity potential. For the cases examined
physical model tests, researchers have developed a wide variety of tools.
here (subcritical speeds), the solutions of the above equation are ellip
Naturally, each of these is associated with a set of assumptions and
tical. The boundary conditions are a critical component of the theory.
limitations. There assumptions and limitations are discussed in the
The hull is described by
following sections.
3
Fig. 2. Case-studies: a – Suez Canal case-study, b – rectangular canal.
Fig. 3. KCS ship model input for slender body calculations.
∂ðx þ ϕÞ 8 0 qffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
¼0 (4) > 2
∂n >
> Bh∞ F∞ 1C
>
> arctan@ qffiffiffiffiffiffiffiffiffiffiffiffiffi A for F∞ < 1
>
>
where ∂=∂n is the derivative in the normal direction. This also holds for >
> h0 1 F20
<
the seabed. To satisfy mass conservation, Tuck (1966) arrived at: θ¼ 0 qffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 (8)
>
>
>
> 2
∂ϕ V ’ >
> Bh∞ 1 F∞ C
¼� S ðxÞ at y¼0 i:e at the hull (5) >
>
> i � sgnðkÞarctan@ qffiffiffiffiffiffiffiffiffiffiffiffiffi A for F∞ > 1
∂x 2h0 : h0 1 F20
where S(x) is the hull cross-sectional area at position x, and prime is used pffiffiffiffiffiffi
to denote the derivative dS/dx, h0 is the interior water region depth, and where i ¼ 1 and sgnðkÞ is the signum of the Fourier transform vari
V is the ship speed. The remaining boundary conditions are that the able. k:
velocity potential must vanish at an infinite distance from the ship. To The only change between the open shallow water, dredged channel
account for a change in depth, which characterises a dredged channel, and canal cases are expresed by the θ parameter and the ratio h0 =h∞ .
Beck et al. (1975) split the flow into interior and exterior regions based When θ ¼ 0, the above relationships reduce to the canal case-studies.
on the depth: Whereas when h0 =h∞ ¼ 1, the relationships reduce to their the open
� water variants (Beck et al., 1975).
hðyÞ
ho ; jyj < w=2
(6) The next step is to calculate the force and moment coefficients, which
h∞ ; jyj > w=2 are predicted as shown in Eq. (9) and Eq. (10), respectively:
R
Where w is the width of the interior region. Splitting the domain thus bðxÞfs’ ðξÞkðx ξÞdξdx
Cf ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffi R (9)
creates two different flow regimes. The interior is characterised by the 2wL 1 F 20 bðxÞdx
pffiffiffiffiffiffiffi
depth h0 and resulting interior depth Froude number F0 ¼ V= h0 g,
pffiffiffiffiffiffiffiffi R
and exterior depth Froude number F∞ ¼ V= h∞ g. xbðxÞf s’ ðξÞkðx ξÞdξdx
Cm ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffi R (10)
The solution proceeds by defining the functions k(x) and θ as follows:
2wL 1 F20 bðxÞx2 dx
2 3 0 1
πx 7 B 2θw C
6
kðxÞ ¼ 4coth qffiffiffiffiffiffiffiffiffiffiffiffiffi 1 5exp@ qffiffiffiffiffiffiffiffiffiffiffiffiffiA (7) Where fs’ ðξÞkðx ξÞdξdx is the convolution mentioned previously, fis
w 1 F20 w 1 F20 used to denote the Cauchy or principle value integral and ξ is the
convolution variable.
Once these are obtained, the solution requires the definition of two
shape parameters, α and β (Eq. (11) and Eq. (12)), which are used to
4
predict the sinkage (CS ) and trim (Cϑ ) coefficients, as demonstrated in Table 1
Eq. (13) and Eq. (14), respectively. Cell numbers.
R Case-study Simulation type Number of cells
xbðxÞdx
α¼ R (11) Rectangular canal Multiphase 1446076
L bðxÞx2 dx
Double body 1055015
R Suez Canal Multiphase 1954292
L bðxÞxdx
β¼R (12) Double body 1038586
bðxÞx2 dx
Cf αCm
CS ¼ (13)
1 αβ
Cm βCf
Cϑ ¼ (14)
1 αβ
Once the above parameters have been obtained, the sinkage (s) and
trim (t) can be calculated:
LCS F 2
s ¼ qffiffiffiffiffiffiffiffiffi0ffiffiffiffi ½m� (15)
1 F 20 Fig. 4. Top view of the generated mesh for the rectangular canal (top half) and
Suez Canal (bottom half) – not to scale.
Cϑ F 2
t ¼ qffiffiffiffiffiffiffiffi0ffiffiffiffiffi ½radians� (16)
1 F20
In order to predict the parameters of interest in the modelled Suez

Canal, the sloped shape is transformed into a rectangular cross-section
with equal depth and blockage ratio (Gourlay, 2008a). This process is
depicted in Fig. 2a for the Suez Canal. The equivalent canal is charac
terised by a depth of 0.32 m, and width equal to 2.89 m. Then, the
application of the abovementioned equations for a canal of constant
width and depth are straightforward. This study will serve as a valida
tion of the in-house code. For reference, the rectangular canal is
modelled as depicted in Fig. 2b.
3.3. Reynolds-averaged Navier-Stokes
Perhaps the most rapidly developing field in ship hydrodynamics

involves the use of Reynolds-averaged Navier-Stokes (RANS) methods.
These solve for the 3D fully nonlinear viscous flow around the ship. The
rapid development of supercomputers, coupled with the RANS method’s
ability to resolve complex flow phenomena with good accuracy has
Fig. 5. Distribution of yþ values in multiphase and double body modes for the
meant that many researchers have adopted the RANS method in their
Suez Canal at Fnh ¼ 0.2439.
toolkit.
In this study, the rectangular canal depicted in Fig. 2 is modelled.
Naturally, the speeds examined replicate the model-test parameters. The depicted mesh is achieved by imposing concentric volumetric
This implies that all ship dimensions and parameters follow those, refinements in the vicinity of the ship. In particular, the location where
prescribed during the experimental investigation. The commercial the Kelvin wake is expected has been refined considerably in the rect
available RANS solver, Star-CCMþ, version 13.02.011 is used in this angular canal case. For the Suez Canal, the refinements are concentrated
study. Star-CCMþ is a finite-volume based solver, which uses the inte near the lateral extents of the canal. As will be shown in Section 4, the
gral form of the governing equations and divides the computational manner in which the waves interact with the bottom is important. The
domain into a finite number of adjoining cells. mesh density required in these areas is increased when compared to
For the present investigation, hexahedral cells of minimal skewness other locations. In any case, the mesh in the Suez Canal is sufficiently
are used. This is known to provide superior predictions in ship hydro refined to capture the waves everywhere in the domain. Although this
dynamics when compared to tetrahedral cells (Jones and Clarke, 2010). implies that the cell numbers have increased dramatically, as shown in
The mesh is generated via the automatic facilities of Star-CCMþ. To Table 1, the heightened computational effort is considered justified.
accomplish this, the trimmed cell mesher is used to construct all cells In terms of numerical set-up, the recommendations of Terziev et al.
outside the immediate vicinity of the ship. The near-wall cells are pre (2019b) are followed. Namely, a 2nd order convection scheme is
scribed via the prism layer meshes, which is set to ensure a yþ<1 over adopted and the k – ω turbulence model is used, which showed reliable
the wetted area of the ship. In doing so, high local wall-based Reynolds predictions over a range of similar case-studies (Terziev et al., 2018;
numbers, which require the use of wall-functions are avoided. This is Wilcox, 2006). The k – ω turbulence model exhibited good, consistent
done in view of the fact that wall functions are incompatible with sep predictions over a range of case-studies, similar to those examined here.
aration and complex phenomena, such as stagnation and recirculation The wall time, required per iteration was also found to be the lowest of
(Durbin and Pettersson Reif, 2011). The resulting cell numbers can be all two equation turbulence models, making it a highly attractive choice.
consulted in Table 1. The generated mesh is depicted in Fig. 4 for both Additionally, the k – ω turbulence model does not require any modifi
case-studies, while the resulting yþ values distribution on the hull are cations if applied near solid boundaries, when attempting to resolve
shown in Fig. 5. boundary layer phenomena, which govern resistance. The present study
5
will also serve to validate the assertion that the k – ω turbulence model
provides good predictions over a greater range of case-studies. The
expectation is that the resistance will be predicted with a small, negative
error based on previous experience (Terziev et al., 2019b). The temporal
term of the Navier-Stokes equations is discretised via a 1st order accu
rate scheme, with a time-step of t ¼ 0.0035 � L/V, following Tezdogan
et al. (2016). In the present simulations, the mesh remains constant for
all speeds, while the time step is varied according to the aforementioned
formula.
To model ship squat, the Dynamic Fluid Body Interaction (DFBI)
module is used. This computes the normal (pressure) forces and
tangential (shear or frictional) forces on the ship hull and adjusts its
position to achieve equilibrium. In the examined case-studies, only
motions in the vertical plane (y – z) are allowed. To dampen the initial Fig. 7. Sinkage for the rectangular canal.
shock, resulting from the initiation of the simulation, the ship is con
strained during the first 10 s, which is imposed to allow the flow to suggested by Hughes (1954), which splits the resistance as shown in Eq.
develop before the ship is allowed to move. Once this time limit has been (17).
overcome, the solver gradually applies forces and moments on the hull
during an additional 10 s. CT ¼ CF � ð1 þ kÞ þ CW (17)
The domain dimensions follow the recommendations of the ITTC
(2017). While the domain bottom, set as a velocity inlet, and side (slip where CT is the total resistance coefficient, CF represents the frictional
wall) are prescribed to match the experimental set-up, the domain top is component, (1 þ k) is the abovementioned form factor, and CW is the
placed at a distance of 1.5 � L from the undisturbed water surface level. wave resistance coefficient. CW and (1 þ k) are assumed constant with
The inlet is also positioned 1.5 � L upstream of the forward perpendic scale, while CF is predicted via a friction line (Molland et al., 2017).
ular, where a velocity inlet condition is imposed. The outlet is located Since CFD cannot be used to predict all of these components via a
2.5 � L ship lengths downstream of the aft perpendicular, and is set to multiphase simulation, one may replace the free-surface with a sym
maintain the hydrostatic pressure. To calculate the hydrostatic pressure, metry plane (Kinaci et al., 2016). Essentially, this is equivalent to
model current velocities, and capture free-surface deformations, the removing CW from Eq. (3), since there are no longer any waves present
Volume of Fluid method is utilised (Hirt and Nichols, 1981). Domain in the simulation (Farkas et al., 2017). Doing this renders the VOF
dimensions and boundary conditions are summarised in Fig. 6 for the method inapplicable. For this reason, in double body simulations the
rectangular canal (top) and Suez Canal (bottom) (see Fig. 7). velocity is prescribed at the inlet as a constant, while the outlet main
If one seeks to demonstrate changes in the components of ship tains 0 Pa pressure.
resistance, a multiphase simulation is insufficient. As mentioned previ To ensure the flow is dissimilar only due to the absence of waves, the
ously, the RANS solver computes forces as normal and tangential. These ship’s vertical position is adjusted, according to the result obtained by
translate into frictional and pressure resistance coefficients upon divi the multiphase simulation (Terziev et al., 2019c). Performing double
sion by 0.5ρSV2, where ρ ¼ 997.561 kg/m3 is the fresh water density, S is body simulations allows for the calculation of the wave resistance sim
the ship’s wetted area, and V is the ship speed. On the other hand, the ply by subtracting the CT achieved in multiphase and double body re
resistance extrapolation procedure, endorsed by the ITTC (2017b) de gimes (Min and Kang, 2010). Furthermore, the form factor is simply the
composes the total differently. Namely, the form factor approach is used, double body total resistance, divided by the double body frictional
component. These will enable the prediction of changes in (1 þ k) and
Fig. 6. Domain dimensions and boundary conditions in all multiphase simulations. Top: rectangular canal, bottom: Suez Canal (not to scale).
6
CW with speed as well as bathymetry. For the typical cell numbers, predicting the expected changes in all factors discussed above.
achieved for the double body simulations, Table 1 can be consulted. The results and relevant discussions are given in Section 4, while the
The double body approach is primarily adopted in this study to following Section provides an overview of the errors and uncertainties
determine the wave resistance of the KCS under different conditions. induced as a result of the RANS method.
Utilising the resistance decomposition shown in Eq. (17), it is possible to
circumvent the complexity of wave resistance estimation. To elaborate, 3.4. Numerical verification
wave resistance estimation in shallow water is a particularly difficult
problem. While there are many theories capable of providing an esti As referred to in the previous section, RANS solvers require a time
mate of the sinkage and trim of the ship, resistance in the subcritical step and grid size to discretise the governing equations temporally and
range has proved elusive (Beck et al., 1975; Tuck, 1967a, 1966). That is, spatially, respectively. The set of partial differential equations, modelled
although deep-water wave resistance can be estimated with the same by the solver (the RANS equations) are thought to represent the physics
theory (Tuck and Lazauskas, 2008). Three-dimensional potential flow of the problem with sufficient accuracy (Lesieur, 2008). However, this
theories are required to predict ship shallow water wave resistance, applies to their continuum form, which are solvable and can be used for
which are characterised by a substantial increase in implementation relatively simple flows. In any case, analytical solutions to the
difficulty. For example, Yuan’s (2018) method can be used in this Navier-Stokes equations in three dimensions are rare, and cannot be
respect. However, in the absence of commercially available software, derived for a problem as complex as multiphase (or double body) flow
where the above theory has been implemented, one cannot apply it in about a ship hull. For this reason, verification procedures are devised to
practice routinely. estimate the error, resulting from either mode of discretisation, and the
The problems associated with wave resistance in shallow water are corresponding uncertainty by extrapolating the solution to a 0 time step
further magnified by two factors. Firstly, shallow water flows are highly or grid size (Roy, 2005). In other words, the solution estimated as if it
three-dimensional. As demonstrated recently by Terziev et al. (2019b), were possible to model the continuum form of the partial differential
the boundary layer of the ship is predicted to come in contact with the equations.
seabed in very shallow water cases. This brings about the second diffi For the present case, the Grid Convergence Index (GCI) method is
culty, referred to previously, namely, non-linear effects. The presence of used, which was devised by Roache (1998) as a uniform method to
such phenomena suggests that the vast majority of shallow water the report numerical uncertainty. This method uses generalised Richardson
ories are inapplicable, even if they provide non-zero predictions. For Extrapolation (Richardson, 1911), and provides a 95% confidence in the
instance, the extension of the slender body theory to account for computed uncertainty (Roache, 1997). To elaborate, upon estimating
ship-ship interactions of Gourlay (2009) can be supplemented by addi the uncertainty, one can have sufficient grounds to maintain that the
tional non-linear terms, as shown by Yeung (1978). In fact, the supple exact solution lies within the bracket, calculated as the uncertainty.
menting terms, introduced in the latter reference have been shown to be Therefore, the error is defined as a quantity characterised by a magni
of similar magnitude as the original terms, formulated by Tuck and tude and sign, whereas the uncertainty simple provides an interval,
Newman (1976). within which the error must be located 95 out of 100 times.
Non-linearity and three-dimensional effects are not a distinctive The first step in the verification procedure is to define a refinement
problem when it comes in CFD predictions. The RANS equations feature ratio (r), which is used to magnify the grid size, or time step. Here, the
both of the above characteristics. Therefore, CFD can provide insight recommendation of the ITTC (2008) are adopted in this respect, namely
pffiffiffi
into the phenomenology of the underlying physics without user inter r ¼ 2. The refinement ratio is used to generate a triplet of solutions,
vention, or knowledge of separate linear/non-linear and 2D/3D effects. which are used to predict the numerical error (Celik et al., 2008). Here,
These are captured automatically in the computational model. In this it is useful to define the modes of convergence or divergence. These are
context, the prediction of the form factor, which is typically used to characterised by the convergence ratio, R, whose value defines four
account for 3D effects is trivial in the presence of data from multiphase possibilities:
and double body simulations. The prediction of the form factor is of
importance because of its central role in resistance extrapolation. 1. Monotonic convergence: 0 < Rκ < 1
In the likely event where the expected routes the ship will spend the 2. Oscillatory convergence: Rκ < 0 [ jRκ j < 1
majority of its operational lifetime are known, it is of critical importance 3. Monotonic divergence, Rκ > 1
to predict the full-scale resistance under these conditions. The manner in 4. Undefined error or uncertainty
which this is typically done is via the resistance decomposition, shown in
Eq. (17). Naturally, the use of this relationship requires adequate Where the subscript refers to the κth input parameter. In the present
knowledge of each component on the right-hand side of the equation. As context, R is defined as the ratio of the difference between medium and
explained previously, wave resistance is associated with a distinct set of fine solutions ε21 ¼ (ϕ2 – ϕ1), and the difference between coarse and
challenges. The form factor is also an elusive parameter, which is not medium solutions ε32 ¼ (ϕ3 – ϕ2). Once these are known, the observed
sufficiently understood in shallow waters. Frictional resistance in order of accuracy is estimated as shown in Eq. (18):
shallow water is also subject to some debate, because as will be shown in
Section 4, the ITTC line cannot capture shallow water effects. Zeng et al. lnðεκ23 =εκ21 Þ
pκ ¼ (18)
(2018) provided a basis for a correction based on the depth over draught lnðrκ Þ
ratio (h/T), but their formulation is applicable for flat ship bottoms. This The next step is to predict the extrapolated value, formulated in Eq.
is certainly not the case in the vast majority of cases due to the occur (19):
rence of ship squat, and is therefore not applied here. �� p �
In the presence of continual debate regarding each component of the ϕext ¼ rpκ � ϕ1 ϕ2 rκ 1 (19)
resistance decomposition shown in Eq. (17), and indeed the relationship Then, the approximate relative error and extrapolated relative error
itself, it is thought prudent to investigate the RANS solver’s performance are calculated as shown in Eq. (20) and Eq. (21), respectively.
in predicting these parameters. The availability of experimental data to � �
compare our numerical predictions against is used to establish confi �ϕ1 ϕ2 �
e21 ¼ � � (20)
dence in the numerical model where appropriate. To the best of our � ϕ �
a
1
knowledge, form factor and wave resistance studies in shallow waters
are few, especially with changing channel cross-section. Thus, the pre
sent investigation would serve to supplement knowledge in the field by
7
� �
�ϕext21 ϕ2 � Table 3
e21
ext ¼ �
� ϕ 21 �
� (21) Temporal uncertainty in the rectangular canal, Fh ¼ 0.469.
ext
Sinkage (m) Trim (� ) Total resistance (N)
Once the quantities estimated by Eq. (18) – Eq. (21) are known, the
pffiffiffi pffiffiffi pffiffiffi
uncertainty can be calculated as expressed in Eq. (22): r(-) 2 2 2
� � ϕ1 0.0052 0.0675 2.8896
�1:25 � e21 �
GCI 21 � p
fine ¼ �
a �
(22) ϕ2 0.00517 0.0674 2.8883
rκ 1 �
ϕ3 0.005169 0.0670 2.8823
The computed data for this case is shown in Table 2, where grid R(-) 0.41748 0.50193 0.21667
discretisation uncertainty is reported. Table 3 presents estimates of the p(-) 2.5205 1.9889 4.4129
temporal discretisation-induced uncertainty. The procedure has been ϕ21
ext
0.0051831 0.067718 2.8909
carried out for sinkage, trim and total resistance. Since the total is simply e21
a ð%Þ
0.00019 0.263 0.045
the sum of the pressure and shear, it is not thought necessary to extend e21
ext ð%Þ
0.059678 0.0111 0.0009
the reported results for the remaining parameters. The uncertainty GCI21
fine ð%Þ
0.077 0.2926 0.0775
analysis presented in this section was performed for the rectangular
canal for Fh ¼ 0.469. This is used as a representative case, providing
guidelines in terms of uncertainty for the remaining case-studies.
It is interesting to note that sinkage and trim exhibit super conver
gence with mesh refinement, but are close to the theoretical order of
accuracy (pt ¼2) when subjected to temporal refinement (Roy, 2005).
The opposite is true in terms of resistance.
To ensure that the solution has converged sufficiently, the residuals
are monitored, requiring a minimum drop of at three orders of magni
tude, following the recommendations of the ITTC (2011). To assess
iterative errors, the non-intrusive, a posteriori method of Roy and
Blottner (2001) is used (Phillips, 2012). The results from this analysis
indicate that absolute errors are in the range of 10 5 – 10 6. To achieve
the here reported levels of iterative convergence, the solution is allowed
to evolve for a minimum of 200 s physical time. The iterative errors must
be several orders of magnitude smaller than the discretisation error to
enable the applicability of the GCI (or any other) discretisation uncer
tainty estimator technique (Eca and Hoekstra, 2014). For the present Fig. 8. Trim for the rectangular canal.
purposes, this condition is thought to have been satisfied.
4. Results and discussion
In this Section, the results for both case-studies are presented. To

facilitate comparison and discussion, the data from all methods are
presented jointly for each case-study.
4.1. Sinkage and trim
As mentioned previously, sinkage and trim are of great practical

importance in restricted waters (Ferguson, 1977). While it is well-known
that smaller canal cross-sectional area causes greater ship squat, the
manner in which this occurs requires further investigation. The results
shown jointly in Fig. 8 and Fig. 9 demonstrate that the present CFD
model agrees well with experimental observations in the rectangular
canal. Moreover, the assertion that our CFD model will have a tendency Fig. 9. Sinkage for the Suez Canal.
to provide a small negative error is validated for the entire speed range
for sinkage in this case-study. In terms of trim, the CFD model has also
Table 2 performed well, predicting values within a reasonable margin. The
Spatial uncertainty in the rectangular canal, Fh ¼ 0.469. slender body theory is shown to perform better for trim than for sinkage,
as measured at the ship centre of gravity. In any case, the results
Sinkage (m) Trim (� ) Total resistance (N)
pffiffiffi pffiffiffi pffiffiffi
compare well for low speeds, which is where the slender body theory’s
r(-) 2 2 2 strengths lie (Tuck and Taylor, 1970). The apparent disagreement in
ϕ1 0.0052 0.0675 2.8896 sinkage for very low speeds is exhibited due to the difficulties in
ϕ2 0.0054 0.0682 2.9559 measuring displacements smaller than 2 mm experimentally (Elsherbiny
ϕ3 0.0104 0.0756 3.0935 et al., 2019b).
R(-) 0.04 0.096573 0.4818 For the rectangular canal, the examined speeds range from 0.1 to
p(-) 9.2877 6.7445 2.1068 0.57 depth Froude numbers. Although the high speed range is unlikely
ϕ21
ext
0.005 0.066831 2.8233 to occur in practice, it is interesting to note that the disagreements be
e21
a ð%Þ
0.0192 0.009 0.021 tween the CFD and EFD curves in sinkage increase with velocity. On the
e21
ext ð%Þ
0.08 0.0204 0.0469 other hand, the differences in trim remain relatively constant. This im
GCI21 7.5321 1.9867 2.463 plies that sinkage is more difficult to predict that trim using the present
fine ð%Þ
8
CFD model. It is important to keep in mind the experimental uncertainty

and its effect on the predicted values. It is important to observe that in
the region of practical interest (depth Froude numbers smaller than 0.3),
the two sets of data agree well.
For the Suez Canal, EFD and CFD simulations up to an equivalent of 9
knots in full-scale, which translates to depth Froude numbers between
0.1 and 0.33. This choice is made because the maximum operational
speed allowed in the Suez Canal is 7 knots (Suez Canal Authority, 2019).
While it is interesting to examine higher speeds for the present case, they
cannot occur in practice due to legal restrictions. The current legal re
strictions limit the allowable speed, which restricts the number of ves
sels per year to approximately 18 000. Therefore, the present
investigation focuses on practically relevant operational speeds in the
Suez Canal. Additionally, Elsherbiny et al. (2019b) demonstrated Fig. 11. CFD and EFD comparison of sinkage at the ship’s centre of gravity for
experimentally that a change in h/T values within the abovementioned all case-studies.
depth Froude number range does not have a significant impact. For this
reason, the present study focuses on h/T ¼ 2.2 (full-scale depth ¼ 24 m) rectangular canal case-study, our CFD model underpredicts the sinkage.
as a representative case, which translates into the actual full-scale depth However, the behaviour of the numerically predicted values for this
of the Suez Canal. parameter are overpredicted in the case of the Suez Canal. This pattern
Figs. 9 and 10 jointly present the sinkage and trim values obtained makes the predictions for the Suez Canal match the measured sinkage in
for the Suez Canal via CFD, EFD and the slender body theory. The figures the rectangular canal. Conversely, the measured sinkage values in the
indicate that the present CFD model, whose physical modelling char rectangular canal seem to lie close the Suez Canal, especially in the high
acteristics have been carried from the previous case-study, performs speed range. The source of these disagreements is not known. However,
adequately in the case of trim. Sinkage once again proves more difficult one may speculate that a combination of uncertainty, both experimental
to accurately estimate. However, the differences between CFD and EFD and numerical, superimposed onto the choices pertaining to the nu
are smaller than those exhibited by the slender body curves. This may merics of the CFD model are the source of the disagreements observed
stem from our assumption, that the sloped canal is equivalent to a above. RANS solutions contain many sources of error, stemming from
rectangular canal with equal blockage. More research in this direction is sources as diverse as boundary conditions, levels of inlet turbulence,
left as future piece of work. iterative and discretisation errors, convection scheme, etc. (Eca et al.,
The slope and shape of the slender body curves are governed by 2013; Roy, 2005; Xing and Stern, 2010). Separating each of these
several parameters. These include the depth and width of the canal, as components and assessing their impact (when possible) on different
well as the shape of the vessel. Additionally, as the depth Froude number aspects of the numerical solution is an ongoing field of research (Eca
approaches unity, a singularity is predicted (Gourlay and Tuck, 2001). In et al., 2017).
other words, the slope of the slender body prediction will increase until The trim predictions, shown in Fig. 12 exhibit a different pattern
the critical depth Froude number is reached, where it will attain an than the sinkage (Fig. 11). Although an agreement is observed in the low
infinite value. This is not predicted experimentally or numerically speed range, the trend is not continued past Fh ¼ 0:235 for the Suez
(Elsherbiny et al., 2019b; Terziev et al., 2018), which restricts the Canal. The EFD data for trim show the oscillatory patterns discussed
applicability of the slender body theory to low speeds. previously, whereas the CFD predictions follow a smooth path. The two
Fig. 11 presents a comparison of the measured sinkage values versus sets of data in the case of the rectangular canal show better agreement in
the numerical predictions for all case-studies. An apparent pattern, terms of trend, but not in the exact location along the y – axis. Instead,
emerging from this figure is that CFD predicts gradual changes in the CFD method shows a systematic underprediction. However, due to
sinkage with the increase of speed. The experimental results show os the sharp decline in the EFD curve, the two sets of data almost coincide
cillations at certain speeds in the displacement of the centre of gravity of for the highest speed examined. That is, although this particular point on
the ship, specifically at Fh ¼ 0:303 for the Suez Canal. Coincidently, this the graph (Fh ¼ 0:57) is unlikely to occur in practice. The sources of
dip in the measured sinkage matches the numerically predicted sinkage error, discussed in the case of sinkage, carry forward to trim as well. In
almost exactly. However, once the dip in measured sinkage has been fact, it is well known that the two parameters (jointly forming ship
overcome, the apparent difference between the two sets of data begins to squat) are intrinsically linked (Shivachev et al., 2017).
increase anew. Fig. 13 depicts a comparison of different empirical models for both
An additional observation, made from Fig. 11 is that in the case-studies. In the figure, how the different models compare to each
other for the same case have been plotted, as well as across case-studies.
These reveal that formulae, containing higher powers of the speed, tend
Fig. 10. Trim for the Suez Canal. Fig. 12. CFD and EFD comparison of ship trim.
9
Fig. 13. Empirical methods comparison for both case-studies.
to massively overpredict values in the high speed range. For instance, and their respective decompositions into constituent components. The
Ro€misch (1989) contains more terms involving higher powers of the discussion begins with the total resistance comparison to establish
speed than the remaining empirical formulae examined in Fig. 13 relative differences between the predicted and measured values.
(Briggs et al., 2009a). The empirical formulae used can be found in the Over the entire speed range examined for the rectangular canal, the
Appendix. CFD model compares well with the EFD data, as shown in Fig. 14. The
Out of all empirical models compared in Fig. 13, only the one due to calculated values show a small underprediction over the majority of the
Ankudinov (whose mathematical basis is explained in detail Briggs speeds, validating the assertion relating to turbulence modelling. The k –
(2009) and Briggs and Daggett (2009)), is capable of predicting whether ω turbulence model is therefore proven as a good choice for examina
a ship will squat by stern or by bow. Unfortunately, both CFD and EFD tions focusing on towed calm shallow water predictions. The resistance
methods show that the ship will squat by bow, rather than by stern, values are shown to grow in a quadratic manner with speed, as expected.
which is the prediction made by the abovementioned method. This When comparing these to the Suez Canal, it is evident that the CFD
highlights the need for accurate tools even in the early design stage. This model does not have a clear tendency to over- or underpredict the EFD
particular formulation predicts virtually identical squat for both data, shown in Fig. 15. This is in all likelihood due to the complex nature
case-studies, which make it difficult to distinguish the two curves in of the problem.
Fig. 13. The CFD method requires the accurate resolution of the free surface
to predict the pressure component of resistance. In the rectangular canal
4.2. Resistance this is straightforward. However, the Suez Canal’s bathymetry is char
acterised by a slope, which terminates with an intersection of the water
This section examines the resistance values computed via CFD, EFD surface. As the waves, shed from the bow approach the bank, they slow
10
within waves vanishes at a rate, proportional to the water depth, and

wave heights increase accordingly. This is clearly observed in Fig. 16,
especially near the banks. No wave breaking is identified in the present
case near the lateral extremities of the domain. However, one may
expect such phenomena to occur and be visible if the ship speed were to
be increased.
It is important to keep in mind that shallow water waves propagate at
pffiffiffiffiffi
a speed of gh. This explains the observed curve in the shape of the
waves as the bank is approached. The propagation speed diminishes
with distance from the ship and the waves fall behind (towards the
outlet) at a greater rate than the remaining wave system. The rectan
gular canal free surface disturbance for the same depth Froude number is
shown in Fig. 17 for comparison.
The generated wave patterns induce a wave resistance onto the ship.
Fig. 14. EFD and CFD of total resistance for the rectangular canal. As explained in Section 4.3, in the present study the wave resistance
coefficient is calculated by subtracting the multiphase total from the
double body total resistance in CFD. In the rectangular canal, the pre
dicted values show a smooth variation, characterised by an initial
decrease, which recovers with increasing depth Froude number (shown
in Fig. 18). This pattern repeats in all likelihood as one progresses
through the depth Froude number range, causing the typical oscillatory
pattern of wave resistance to emerge (Tuck and Lazauskas, 2008). The
Suez Canal wave resistance on the other hand exhibits a sharp rise,
following which, the pattern of the rectangular canal seems to be
reproduced. This is not surprising, considering the wave disturbance
observed in Fig. 17. To elaborate, wave effects become more significant
as speed is increased at a greater rate than in the rectangular canal.
The wave resistance coefficients presented in Fig. 18 show similar
behaviour to those experimentally estimated by Elsherbiny et al.
(2019b). However, our predictions are higher than those reported by the
abovementioned authors. This can be explained by the choice of friction
Fig. 15. EFD and CFD of total resistance for the Suez Canal. line, which during the experimental stage (in Elsherbiny et al. (2019b))
was the ITTC line (ITTC, 2017b). The impact of the choice of skin fric
tion coefficient is illustrated in Fig. 19. Namely, depending on which
down, grow in height and change direction (Lamb, 1932). The manner
method to estimate the skin friction of the ship is chosen, the wave
in which they interact with the bathymetry is not limited to the above. A
resistance may vary substantially, especially during the experimental
part of the waves will also reflect back onto the ship upon reaching the
stage, where there is no other choice but to calculate CF via a friction
beginning of slope, interacting with the hull. Now, the speeds examined
line.
for the Suez Canal are low, rendering the waves largely inconsequential
Another interesting property of the results presented herein is that in
near Fnh ¼ 0:1. However, as shown in Fig. 16, even for a depth Froude
the rectangular canal, CW does not show signs of decreasing even when
number as low as 0.33, the generated disturbance is substantial. The
the depth Froude number is as low as 0.1. The Suez Canal results on the
interaction between the ship and the waves, shed from its bow and stern
other hand exhibit the aforementioned decline. Thus, the zero wave
are significant, making this a highly unsteady case.
resistance assumption at low speeds may not be valid, depending on the
In Fig. 16, the projection of the beginning of the slope onto the free
underwater topology.
surface (white lines) have been included to enable a better under
In Fig. 19, the double body and multiphase frictional resistance co
standing of the underlying phenomena. In particular, it seems that the
efficients are presented graphically against Reynolds number. Several
seabed immediately after the beginning of the slope does not interact
differences between the case-studies become immediately apparent.
strongly with the waves. This is in line with theoretical predictions,
Firstly, contrary to the deep water cases of Terziev et al. (2019c), the
which dictate that the effect on the wave pattern depends on the tran
frictional resistance coefficient in double body mode is higher, rather
sition in depth. The large waves, appearing with increasing distance
than lower than the multiphase case for the same Reynolds number in
from the ship centreline seem to decompose into smaller components
the rectangular canal.
with larger wave heights. According to Newman (1965), the energy
Fig. 16. Generated wave patterns in the Suez Canal for Fr ¼ 0:33. Mirrored about the central plane.
11
Fig. 17. Free surface disturbance, generated in the rectangular canal at.Fr ¼ 0:33
effects decrease in importance with increasing speed. It is also noted that

the prediction made herein is virtually the same as in Elsherbiny et al.
(2019b) for h/T ¼ 2.2. In fact, it seems that once a speed threshold is
passed, the form factor ceases to vary and approaches the experimen
tally determined value.
The form factor curve for the rectangular case-study shows a smooth
behaviour, which is taken as an indication of the success of the present
CFD model. The fact that the Suez Canal prediction seems to follow a
similar path is taken as a good sign in terms of accuracy as well. The
apparent disagreement between the two curves for the lowest speed
suggests that there may be difficulty in predicting the underlying physics
accurately. This is in line with expectations, which dictate that at a very
low speed, the waves generated by the ship will be very short. These are
difficult to capture if the mesh density is not high in the vicinity of where
the abovementioned waves are expected. Although the generated mesh
was progressively refined as the bank is approached, this may not have
been sufficient for the lowest speeds. However, since the experimental
method faces equal, if not greater challenges for the lowest speeds, it is
considered that the results presented herein exhibit sufficient levels of
accuracy.
Fig. 18. Calculated wave resistance coefficients for both case-studies The fact that the experimentally determined form factor is close to
using CFD.
the low speed predictions of the present CFD model is encouraging.
However, it also brings into question the experimental method for
The frictional resistance coefficients predicted for the Suez Canal determination of the form factor, as applied to canal case-studies. This is
exhibit a surprising feature. Namely, the double body predictions are typically assessed at very low speeds, where the wave resistance is
either close to, or below the multiphase skin friction predictions. This thought to be negligible. Our findings, specifically Fig. 17, demonstrate
implies that underwater bathymetry has a substantial effect on CF . To that wave resistance is likely the cause in the experimental method’s
further this argument, one may refer to the observed deviation between inability to predict the change in (1 þ k). As discussed previously, wave
the frictional predictions for the Suez and rectangular case-studies. Here, resistance does not decay in the rectangular canal as one might expect.
it is evident that a lateral restriction creates an increase in frictional The consequence of this is expressed in a contamination of the procedure
resistance. Thus, the extension of the ITTC line for shallow waters pro used to experimentally determine the form factor. The results of this
posed by Zeng et al. (2018), which corrects only the ship bottom, may study suggest that CFD should be used to supplement the experimental
not be applicable in the present case. In addition, the ship bottom must determination of (1 þ k).
be maintained flat as part of the abovementioned authors’ friction line,
which was not done in this study. The difference between CF for all cases 5. Conclusion and future work recommendations
shows signs of monotonic decline with increasing Reynolds number.
This may suggest that free surface effects become negligible at higher This study presented an experimental and numerical assessment of
speeds or scale factors. However, one must remember that in the resis the effects of bathymetry on ship squat and resistance. To demonstrate
tance calculation process, the underwater area and speed are used to the practical importance of the work, the Suez Canal was modelled and
non-dimensionalise the resistance coefficients. Thus, an apparent compared to a rectangular canal. Emphasis was placed on low and
decline in coefficient form could translate into an increase in dimen moderate speeds in the Suez Canal, following the legal restrictions
sional resistance, depending on the case. imposed on ships in the abovementioned waterway. Specifically, the
Fig. 20 shows that the form factor behaves in a manner, opposite to maximum allowed speed is 7 knots.
the wave resistance for the Suez Canal. That is to say, (1 þ k) attains a In this paper, the numerical results showed best agreement with the
high value of approximately 1.245 for a depth Froude number of 0.1. experimental data. This is likely the case due to the ability of CFD to
This is followed by a sharp decline, which brings the Suez Canal form model complex phenomena without the need to individually prescribe
factor in line with the rectangular canal. In fact, the difference between each term in suitable manner, as is the case for potential flow. With the
the two is vanishingly small for the highest depth Froude number rapid growth in available computational power, studies such as the one
examined in the present study. This suggests that, if one were to accept presented here will become frequent. More importantly, the RANS
the present resistance decomposition methodology, three dimensional method’s ability to provide good predictions when compared to
12
Fig. 19. Frictional resistance coefficients for all case-studies (Eca and Hoekstra, 2008; Gadd, 1967; Hughes, 1954; ITTC, 2017b; Katsui et al., 2005; Korkmaz et al.,
2019; Lazauskas, 2009; Prandtl, 1925; Schlichting, 1979; Schoenherr, 1932; Schultz-Grunow, 1941; Telfer, 1927; Wang et al., 2015; White, 2006).
course of an experiment. From a numerical point of view, these case-

studies are also challenging. The ship produces short waves, whose
resolution requires a much finer mesh than was adopted here.
In terms of form factor, the research, reported above indicates a
sharp decline with increasing speed for both case-studies. Our pre
dictions suggest that as the speed is increased, the difference in (1 þ k),
calculated for either case vanishes after Fh ¼ 0:25. In the present cases,
the form factor’s multiplicative relationship with the frictional resis
tance coefficient should not be ignored. It is also unlikely that three
dimensional effects are captured adequately by a linear resistance
decomposition. To support this claim, changes in CF between the Suez
Canal, rectangular canal, and their double body variants were presented.
Surprisingly, the difference between multiphase and double body fric
tional resistance coefficients was shown to be greater for the rectangular
canal than the Suez Canal.
The combination of insurmountable experimental and numerical
limitations suggests that it may be more efficient to shift one’s attention
away from extremely low speeds. These low speeds are used to predict
the ship resistance without the influence of wave resistance. In this
study, it was demonstrated that this may not be a good approach because
Fig. 20. Predicted form factor for all case-studies using CFD. wave resistance plays an important role even at very low speeds (near
Fnh ¼ 0.1) Instead, the ability of modern computational tools should be
experiments reinforces the case for simulation-based design. exploited to a greater extent, as demonstrated in this study. The nu
The present investigation focused on changes in form factor, wave merical method can be used to manipulate the physics of the problem to
resistance and friction across case-studies. The results reveal that the provide the desired component of ship resistance. Whether it is physi
Prohaska test, which requires a vanishingly small wave resistance is cally meaningful to split the total into linear components is a disputable
valid for the Suez Canal, but may not be true for the rectangular canal. In matter. In the absence of better extrapolation techniques, one has little
fact, CW vanishes with a reduction in depth Froude numbers below about choice but to accept the above limitations. However, this is only the case
0.2. Surprisingly, the rectangular canal’s wave resistance showed no if one continues to view the problem of ship resistance prediction and
signs of decay with low speeds. This suggests that a speed in the vicinity extrapolation to full-scale as a problem to be approached in a purely
of 0.05 depth Froude number may be required to guarantee that wave experimental way. It is hoped that our study has demonstrated the
effects are asymptotically negligible. Unfortunately, such case-studies strong applicability of CFD for problems of practical importance.
are difficult to perform, due to limitations in the sensors used in the The present work could be extended in several ways. The ship’s
13
surface was assumed smooth to reduce the complexity of the physics. Methodology, Software, Writing- Reviewing and Editing
However, this is rarely, if ever, the case in reality. The variations in skin Tahsin Tezdogan, Atilla Incecik, Mohamed Kotb: Supervision,
friction and wave resistance will undoubtedly change upon the inclusion Writing- Reviewing and Editing
of such effects. For a sample of the intricacies of the subject, the reader is
referred to the work of Demirel et al. (2017, 2014), and Song et al. Declaration of competing interest
(2019).
Self-propulsion is also an important topic. The presence of a rotating The authors declare that they have no known competing financial
propeller at the stern of the ship causes a complex, unsteady three interests or personal relationships that could have appeared to influence
dimensional pressure field. This affects all aspects of the ship’s hydro the work reported in this paper.
dynamics (Wang et al., 2017). The shallow water cases examined herein
amplify the abovementioned effects in a manner that is not well un Acknowledgements
derstood. These problems should be addressed in subsequent studies.
Results were obtained using the ARCHIE-WeSt High Performance
Author contributions Computer (www.archie-west.ac.uk) based at the University of Strath
clyde. The experimental results were obtained at the Kelvin Hydrody
Khaled Elsherbiny, Momchil Terziev: Conceptualization, namics Lab at the University of Strathclyde.
Appendix
Parameter Symbol Unit Formula Notes
Bow Squat (Barrass, 1981) Sb m 2=3

Cb S2 V2:08
(7)
30
Velocity return factor S2 AS
Ac As
Blockage factor S As
Ac
Channel cross-sectional area Ac m2
Ship cross sectional area As m2
� �
Bow squat (Eryuzlu and Hausser, 1978) Sb m 1 0:27 1:8
0:113B Fd (8)
h=T
� �2:289 � � 2:972
Bow squat (Eryuzlu et al., 1994) Sb m h 2
V h
0:298 pffiffiffiffiffi Kb (9)
T gT T
Correction factor Kb 3:1 For unrestricted channels, use Weff
pffiffiffiffiffiffiffiffiffiffifor W=B < 9:61 ¼ 1for W=B � 9:61
W=B
Effective width Weff m CM B ¼ ½7:7 þ 45ð1 CWP Þ2 �(10)
Bow squat (Hooft, 1974) Sb m r Fn2 ris the volumetric displacement
1:96 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi(11)
L 1 Fn2
Bow squat (ICORELS, 1980) Sb m r Fn2
2:4 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
L 1 Fn2
� �
Bow squat (Millward, 1990, Millward, 1992) Sb m 1 F2
0:001L 61:7Cb 0:6 qffiffiffiffiffiffiffiffidffiffiffiffiffiffiffiffiffiffiffiffi(12)
L=B 1 0:9F2d
Bow squat (R€omisch, 1989) Sb m CV CF KT T(13)
Stern squat (R€
omisch, 1989) Ss m CV KT T(14)
� � �� 4 ��
Correction factor CV V 2 V For ship speed
8 0:5 þ 0:0625 (15)
VCR VCR
Relevant water depth hmT m hm
h ðh hm Þ
m pffiffiffiffiffiffihffiffiffiffi
Wave celerity CmT ghmT Based on the relevant water depth
s
Mean water depth hm m Ac
WTop
Channel width WTop m W þ 2nh At the water surface
� �
Correction factor for the ship’s shape CF 10Cb 2 ¼ 1for stern squat
L=B
pffiffiffiffiffiffiffiffi
Correction factor KT 0:155 h=T For squat at critical speeds
Wave celerity C m pffiffiffiffiffi
gh Based on the depth h
s
Inverse bank slope n Specified as an integer
Ship critical speed VCR m VCR ¼ CKch Uconfigurations
s VCR ¼ CKC Rconfigurations
� � � ��
hm hm Cconfigurations
VCR ¼ CmT Kch 1 þ Kc
h h
� �
Correction factor Kc 1
0:2306log þ 0:0447
S
Wave celerity CmT m pffiffiffiffiffiffiffiffiffiffi
ghmT Based on the relevant water depth
s hmT
Bow/Stern squat (Ankudinov et al., 1996) SMax m LðSmid �0:5TrimÞ(16) þ 0:5Trim: Bow Bow/Stern squat depends on trim (±). For Fd � 0:6
0:5Trim : Stern
Midship sinkage (Ankudinov et al., 1996) Smid m ð1 þ KSP ÞPHu PFd Pþh=T PCh1 (17) For Fd � 0:6
(continued on next page)
14
(continued )
Parameter Symbol Unit Formula Notes
Factor KSP 0:15 Ships with single propellers

0:13 Ships with twin propellers
� �
Hull parameter PHu BT
1:7CB 2 þ 0:004C2b
L
Forward speed parameter PFd F1:8þ0:4F d
d
Water depths effect parameter Pþh=T 0:35
01 þ
ðh=TÞ2
Channel effects parameter PCh1 1 Uconfigurations
pffiffiffiffiffi
1 þ 10Sh 1:5ð1 þ Sh Þ Sh Rand C configurations
� ��
Water depth factor Sh S hT Rand C configurations
CB
h=T h
Trim (Ankudinov et al., 1996) Trim �
1:7PHu PFd Ph=T KTr PCh2
Trim coefficient KTr CnTr
b ð0:15KSP þ KTP Þ ðKTB þ KTTr þ KTT1 Þ
Trim exponent nTR 2 þ 0:8PCh1 =Cb
Propeller parameter KTP 0:15 Single propellers
0:2 Twin propellers
Bulbous bow parameter KTB 0:1 Ships with bulbous bows
0 Ships without bulbous bows
Stern transom parameter KTTr 0:04 Ships with a stern transom
0 Ships without a stern transom
Initial trim factor KTT1 Tap Tfp
Tap þ Tfp
Trim correction parameter PCh2 1 Uconfigurations
1 5Sh Rand C configurations
��
Bow squat (Yoshimura, 1986) Sb m 1:5T BCb 15T BCb 3 2 U, R and C configurations g ¼ 9:81m=s
0:7 þ þ Ve
h L h L
g
Enhanced ship speed term Ve m V
s 1 S
The information, presented in this Appendix has been adopted form Terziev et al. (2018).
Khaled Elsherbiny, Momchil Terziev.
Conceptualization, Methodology, Software, Writing- Reviewing and Editing.
Tahsin Tezdogan, Atilla Incecik, Mohamed Kotb.
Supervision, Writing- Reviewing and Editing.
References Demirel, Y.K., Turan, O., Incecik, A., 2017. Predicting the effect of biofouling on ship
resistance using CFD. Appl. Ocean Res. 62, 100–118. https://doi.org/10.1016/j.
apor.2016.12.003.
Alderf, N., Leftançois, E., Sergent, P., Debaillon, P., 2008. Dynamic ship response
Duffy, J., 2008. Modelling of Ship-Bank Interaction and Ship Squat for Ship-Handling
integration for numerical prediction of squat in highly restricted waterways. Int. J.
Simulation by.
Numer. Methods Fluids 601–629. https://doi.org/10.1002/fld.
Durbin, P.A., Pettersson Reif, B.A., 2011. Statistical Theory and Modelling for Turbulent
Ankudinov, V., Daggett, L., Huval, C., Hewlett, C., 1996. Squat predictions for
Flow, second ed. Wiley.
maneuvering applications. In: Proceedings, International Conference on Marine
Eca, L., Hoekstra, M., 2014. A procedure for the estimation of the numerical uncertainty
Simulation and Ship Maneuverability (MARSIM ’96), September 9–13, Copenhagen,
of CFD calculations based on grid refinement studies. J. Comput. Phys. 262,
Denmark, pp. 467–495.
104–130. https://doi.org/10.1016/j.jcp.2014.01.006.
Barrass, C.B., 1981. Ship squat – A reply. The Naval Architect, pp. 268–272.
Eca, L., Hoekstra, M., 2008. The numerical friction line. J. Mar. Sci. Technol. 13,
Barrass, C.B., Derrett, D.R., 2012. Chapter 42 – ship squat in open water and in confined
328–345. https://doi.org/10.1007/s00773-008-0018-1.
channels. Sh. Stab. Masters Mates 367–388. https://doi.org/10.1016/B978-0-08-
Eca, L., Hoekstra, M., Beja Pedro, J.F., Falcao de Campos, J.A.C., 2013. On the
097093-6.00042-6.
characterization of grid density in grid refinement studies for discretization error
Beck, R.F., Newman, J.N., Tuck, E.O., 1975. Hydrodynamic forces on ships in dredged
estimation. Int. J. Numer. Methods Fluids 72, 119–134. https://doi.org/10.1002/
channels. J. Shellfish Res. 19, 166–171.
fld.3737.
Briggs, M.J., 2009. Ankudinov ship squat predictions – Part I : theory, parameters, and
Eca, L., Vaz, G., Hoekstra, M., 2017. Iterative Errors in Unsteady Flow Simulations : Are
FORTRAN programs. US Army Eng. Res. Dev. Cent. 1–15.
They Really Negligible ? NuTTS-2017 1–5. https://doi.org/10.1016/j.
Briggs, M.J., 2006. Ship squat predictions for ship/tow simulator (ERDC/CHL CHETN-I-
cej.2015.03.004.
72). Erdc/Chl Chetn-I-72 1–18.
el Moctar, O., Shigunov, V., Zorn, T., 2012. Duisburg test Case : post-panamax container
Briggs, M.J., Daggett, L., 2009. Ankudinov ship squat predictions – Part II : laboratory
ship for benchmarking. Ship Technol. Res. J. 59, 50–64. https://doi.org/10.1179/
and field comparisons and validations. US Army Eng. Res. Dev. Cent.
str.2012.59.3.004.
Briggs, M.J., Group, B., Debaillon, P., 2009. Comparisons of PIANC and Numerical Ship
Elsherbiny, K., Tezdogan, T., Kotb, M., Incecik, A., Day, A.H., 2019. An experimental
Squat Predictions for Rivers Elbe and Weser.
investigation of the trim effect on the behaviour of a containership in shallow water.
Briggs, M.J., Kopp, P.J., Ankudinov, V.K., Silver, A.L., 2013. Comparison of measured
In: Proceedings of the ASME 2019 38th International Conference on Ocean. Offshore
ship squat with numerical and empirical methods. J. Shellfish Res. 57, 73–85.
and Arctic Engineering OMAE2019, Glasgow, UK.
https://doi.org/10.5957/JOSR.57.1.120007.
Elsherbiny, K., Tezdogan, T., Kotb, M., Incecik, A., Day, S., 2019. Experimental analysis
Briggs, M.J., Vantorre, M., Uliczka, K., 2009. Prediction of squat for underkeel clearance.
of the squat of ships advancing through the New Suez Canal. Ocean. Eng. 178,
https://doi.org/10.1142/9789812819307_0026, 723-774.
331–344. https://doi.org/10.1016/j.oceaneng.2019.02.078.
Calisal, S., Alidadi, M., 2011. A Numerical Study on Squat of A Wigley Hull 1–5.
Eryuzlu, N.E., Hausser, R., 1978. Experimental investigation into some aspects of large
Celik, I.B., Ghia, U., Roache, P.J., Freitas, C., 2008. Procedure for estimation and
vessel navigation in restricted waterways. In: Proceedings of the Symposium of
reporting of uncertainty due to discretization in CFD applications. J. Fluids Eng. 130,
Aspects of Navigability of Constraint Waterways Including Harbor Entrances, vol 2,
078001 https://doi.org/10.1115/1.2960953.
pp. 1–15.
Constantine, T., 1960. On the movement of ships in restricted waterways. J. Fluid Mech.
Eryuzlu, N.E., Cao, Y.L., D’Agnolo, F., 1994. Underkeel requirements for large vessels in
9, 247. https://doi.org/10.1017/S0022112060001080.
shallow waterways. In: 28th International Navigation Congress. PIANC, Sevilla,
Dand, W.I., 1967. The Wavemaking Resistance of Ships: Vertical Force and Form
pp. 17–25. Paper S11–2.
Resistance of a Hull at Uniform Velocity. PhD Thesis. University of Glasgow.
Farkas, A., Degiuli, N., Marti�c, I., 2017. Numerical investigation into the interaction of
Demirel, Y.K., Khorasanchi, M., Turan, O., Incecik, A., Schultz, M.P., 2014. A CFD model
resistance components for a series 60 catamaran. Ocean. Eng. 146, 151–169. https://
for the frictional resistance prediction of antifouling coatings. Ocean. Eng. 89,
doi.org/10.1016/j.oceaneng.2017.09.043.
21–31. https://doi.org/10.1016/j.oceaneng.2014.07.017.
Ferguson, A.M., 1977. Factors Affecting the Components of Ship Resistance. PhD Thesis.
University o Glasgow, Department of Naval Architecture. University of Glasgow.
15
Gadd, G.E., 1967. A new turbulent friction formulation based on a reappraisal of Hughes’ stresses in a masonry dam. Philos. Trans. th R. Soc. London,Containing Pap. a Math.
results. Trans. RINA 109, 109–511. Phys. Character 210, 307–357.
Gourlay, T., 2014. ShallowFlow: a program to model ship hydrodynamics in shallow Roache, P.J., 1998. Validation and Verification in Computational Science and
water. OMAE 2014, 8. https://doi.org/10.1115/OMAE2014-23291. Engineering. Hermosa, Albuquerque, NM.
Gourlay, T., 2009. Sinkage and trim of two ships passing each other on parallel courses. Roache, P.J., 1997. Quantification of uncertainty in computational fluid dynamics. Annu.
Ocean. Eng. 36, 1119–1127. https://doi.org/10.1016/j.oceaneng.2009.06.003. Rev. Fluid Mech. 29, 123–160. https://doi.org/10.1146/annurev.fluid.29.1.123.
Gourlay, T., 2008. Slender-body methods for predicting ship squat. Ocean. Eng. 35, R€
omisch, K., 1989. Empfehlungen zur Bemessung von Hafeneinfahrten. Tech. Univ.
191–200. https://doi.org/10.1016/j.oceaneng.2007.09.001. Dresden.
Gourlay, T., 2008. Sinkage and trim of a fast displacement catamaran in shallow water. Rotteveel, E., Hekkenberg, R.G., 2015. The influence of shallow water and hull form
J. Shellfish Res. 52, 175–183. variations on Inland ship resistance, 2, 220–236.
Gourlay, T., Tuck, E.O., 2001. The maximum sinkage of a ship. J. Shellfish Res. 45, Roy, C., 2005. Review of code and solution verification procedures for computational
50–58. simulation. J. Comput. Phys. 205, 131–156. https://doi.org/10.1016/j.
Havelock, T., 1924. The propagation of groups of waves in dispersive media, with jcp.2004.10.036.
aplication to waves on water produced by a travelling disturbance 422–451. Roy, C., Blottner, F.G., 2001. Assessment of one- and two-equation turbulence models for
https://doi.org/10.1098/rspa.1933.0074. hypersonic transitional flows. J. Spacecr. Rocket. 38, 699–710. https://doi.org/
Hirt, C.W., Nichols, B.D., 1981. Volume of fluid (VOF) method for the dynamics of free 10.2514/2.3755.
boundaries. J. Comput. Phys. 39, 201–225. https://doi.org/10.1016/0021-9991(81) Schlichting, H., 1979. Boundary-Layer Theory, seventh ed. McGraw-Hill. https://doi.
90145-5. org/10.1007/978-3-662-52919-5.
Hooft, J.P., 1974. The behavior of a ship in head waves at restricted water depth. Schoenherr, K., 1932. Resistance of flat surfaces moving through a fluid. Trans. - Soc.
International Shipbuilding Progress 21 (244), 367–378. Nav. Archit. Mar. Eng. 40, 279–313.
Hughes, G., 1954. Friction and form resistance in turbulent flow and a proposed Schultz-Grunow, F., 1941. New Frictional Resistance Law for Smooth Plates.
formulation for use in model and ship correlation. Trans. Inst. Nav. Arch. 96. Sergent, P., Lefrançois, E., Mohamad, N., 2015. Virtual bottom for ships sailing in
Inui, T., 1954. Wave-making resistance in shallow sea and in restricted water. with restricted waterways (unsteady squat). Ocean. Eng. 110, 205–214. https://doi.org/
Special Reference to its Discontinuities 1, 1–10. 10.1016/j.oceaneng.2015.10.017.
ITTC, 2017. Quality System Manual Recommended Procedures and Guidelines Procedure Shivachev, E., Khorasanchi, M., Day, A.H., 2017. Trim influence on KRISO container ship
Uncertainty Analysis in CFD Verification and Validatio. (KCS); an experimental and numerical study. In: Proceedings of the ASME 2017 36th
ITTC, 2017. Recommended procedures 1978 ITTC performance prediction method, 4th International Conference on Ocean, Offshore and Arctic Engineering, pp. 1–7.
revision, 7.5 – 02 03 – 01.4. Int Towing Tank Conf. https://doi.org/10.1115/OMAE2017-61860.
ITTC, 2011. Recommended procedures and guidelines: practical guidelines for ship CFD. Song, S., Demirel, Y.K., Atlar, M., 2019. An investigation into the effect of biofouling on
26th Int. Towing Tank Conf. the ship hydrodynamic characteristics using CFD. Ocean. Eng. 175, 122–137.
ITTC, 2008. Uncertainty analysis in CFD verification and validation methodology and https://doi.org/10.1016/j.oceaneng.2019.01.056.
procedures. 25th ITTC 2008, Resist. Comm. 12. Stern, F., Yang, J., Wang, Z., Sadat-Hosseini, H., Mousaviraad, M., 2013. Computational
ICORELS (INTERNATIONAL COMMISSION FOR THE RECEPTION OF LARGE SHIPS), ship hydrodynamics: nowadays and way forward. Int. Shipbuild. Prog. 60, 3–105.
1980. Report of Working Group IV. PIANC Bulletin, 35, suppl, 39 pp. https://doi.org/10.3233/ISP-130090.
Janssen, P.P., Schijf, J., 1953. He relation between the form of cross section the cross Suez Canal Authority, 2019. Suez Canal Rules of Navigation. www.suezcanal.gov.eg/En
section, the method of revetment and the distribution of the water velocities in a glish/Navigation/Pages/RulesOfNavigation.aspx.
waterway. In: PIANC 18th Congress 1953, Rome, Section SI C1. PIANC. Telfer, E.V., 1927. Ship resistance similarity. Trans. R. Inst. Nav. Archit. 69, 174–190.
Ji, S.C., Ouahsine, A., Smaoui, H., Sergent, P., 2012. 3-D numerical simulation of convoy- Terziev, M., Tezdogan, T., Incecik, A., 2019. Influence of Mixed Flow on Ship
generated waves in a restricted waterway. J. Hydrodyn. 24, 420–429. https://doi. Hydrodynamics in Dredged Channels. OMAE2019, Glasgow, UK, pp. 1–10.
org/10.1016/S1001-6058(11)60263-1. Terziev, M., Tezdogan, T., Incecik, A., 2019. Application of eddy-viscosity turbulence
Jones, D.A., Clarke, D.B., 2010. Fluent Code Simulation of Flow Around a Naval Hull: the models to problems in ship hydrodynamics. Ships Offshore Struct. 1–24. https://doi.
DTMB 5415. Def. Sci. Technol. Organ. Victoria (Australia). Marit. platforms Div. org/10.1080/17445302.2019.1661625.
Joukovski, N., 1903. On the wave of translation. Complet. Work. Terziev, M., Tezdogan, T., Incecik, A., 2019. In press: a geosim analysis of ship resistance
Katsui, T., Asai, H., Himeno, Y., Tahara, Y., 2005. The proposal of a new friction line. In: decomposition and scale effects with the aid of CFD. Appl. Ocean Res.
Fifth Osaka Colloquium on Advanced CFD Applications to Ship Flow and Hull Form Terziev, M., Tezdogan, T., Oguz, E., Gourlay, T., Demirel, Y.K., Incecik, A., 2018.
Design, Osaka, Japan. Numerical investigation of the behaviour and performance of ships advancing
Kinaci, O.K., Sukas, O.F., Bal, S., 2016. Prediction of wave resistance by a Reynolds- through restricted shallow waters. J. Fluids Struct. 76, 185–215. https://doi.org/
averaged Navier-Stokes equation-based computational fluid dynamics approach. 10.1016/j.jfluidstructs.2017.10.003.
Proc. Inst. Mech. Eng. M J. Eng. Marit. Environ. 230, 531–548. https://doi.org/ Tezdogan, T., Demirel, Y.K., Kellett, P., Khorasanchi, M., Incecik, A., Turan, O., 2015.
10.1177/1475090215599180. Full-scale unsteady RANS CFD simulations of ship behaviour and performance in
Korkmaz, K.B., Werner, S., Bensow, R.E., 2019. Numerical friction lines for CFD based head seas due to slow steaming. Ocean. Eng. 97, 186–206. https://doi.org/10.1016/
form factor determination. In: VIII International Conference on Computational j.oceaneng.2015.01.011.
Methods in Marine Engineering MARINE 2019. Tezdogan, T., Incecik, A., Turan, O., 2016. A numerical investigation of the squat and
Kreitner, J., 1934. The Resistance of Ships in Confined Waters. Werft-Rederei-Hafen. resistance of ships advancing through a canal using CFD. J. Mar. Sci. Technol. 21,
Lamb, H., 1932. Hydrodynamics. Cambridge Univ. Press 6th revise, pp. 262–264. 86–101. https://doi.org/10.1007/s00773-015-0334-1.
https://doi.org/10.1017/CBO9781107415324.004. Tuck, E.O., 1967. Sinkage and trim in shallow water of finite width. Schiffstechnik 14,
Lataire, E., Vantorre, M., Delefortrie, G., 2012. A prediction method for squat in 92–94.
restricted and unrestricted rectangular fairways. Ocean. Eng. 55, 71–80. https://doi. Tuck, E.O., 1967. A Simple “Filon-Trapezoidal” Rule 121, 1–12.
org/10.1016/j.oceaneng.2012.07.009. Tuck, E.O., 1966. Shallow-water flows past slender bodies. J. Fluid Mech. 26, 81–95.
Lazauskas, L.V., 2009. Resistance, Wave-Making and Wave-Decay of Thin Ships, with https://doi.org/10.1017/S0022112066001101.
Emphasis on the Effects of Viscosity. Tuck, E.O., Lazauskas, L.V., 2008. Drag on a ship and Michell’s integral. In: Int. Congress
Lesieur, M., 2008. Turbulence in fluids. In: Physics of Fluids, fourth ed. Springer. https:// of Theoretical and Applied Mechanics, Adelaide, South Australia, August.
doi.org/10.1063/1.1711203. Tuck, E.O., Newman, J.N., 1976. Hydrodynamic interactions between ships. In:
Michell, J.H., 1898. The wave-resistance of a ship. London, Edinburgh, dublin philos. Symposium on Naval Hydrodynamics, 10th, Cambridge. Mass, June 24-28.
Mag. J. Sci. 45, 106–123. Tuck, E.O., Taylor, J.P., 1970. Shallow wave problems in ship hydrodynamics. In: 8th
Millward, A., 1990. A preliminary design method for the prediction of squat in shallow Symposium Naval Hydrodynamics, pp. 627–659.
water. Mar. Technol. 27 (1), 10–19. Tunaley, J.K.E., 2014. Ship wakes in shallow waters. LRDC Rep 6–9.
Millward, A., 1992. A comparison of the theoretical and empirical prediction of squat in Wang, J., Zou, L., Wan, D., 2017. CFD simulations of free running ship under course
shallow water. Int. Shipbuild. Prog. 39417, 69–78. keeping control. Ocean. Eng. 141, 450–464. https://doi.org/10.1016/j.
Millward, A., 1996. A review of the prediction of squat in shallow water. J. Navig. 77–88. oceaneng.2017.06.052.
https://doi.org/10.1017/S0373463300013126. Wang, Z.Z., Xiong, Y., Shi, L.P., Liu, Z.H., 2015. A numerical flat plate friction line and its
Min, K.-S., Kang, S.-H., 2010. Study on the form factor and full-scale ship resistance. application. J. Hydrodyn. 27, 383–393. https://doi.org/10.1016/S1001-6058(15)
J. Mar. Sci. Technol. 15, 108–118. https://doi.org/10.1007/s00773-009-0077-y. 60496-6.
Molland, A.F., Turnock, S.R., Hudson, D.A., 2017. Model-ship extrapolation. In: Ship White, F., 2006. Viscous Fluid Flow, third ed. McGraw-Hill.
Resistance and Propulsion: Practical Estimation of Ship Propulsive Power. Wilcox, D.C., 2006. Turbulence Modeling for CFD, third ed. Transportation Research
Cambridge University Press, Cambridge, pp. 70–85. https://doi.org/10.1017/ Record. https://doi.org/10.1016/j.aqpro.2013.07.003. DCW Industries.
9781316494196.006. Xing, T., Stern, F., 2010. Factors of safety for Richardson extrapolation. J. Fluids Eng.
Newman, J.N., 1965. Propagation of water waves over an infinite step. J. Fluid Mech. 23, 132, 061403 https://doi.org/10.1115/1.4001771.
399–415. https://doi.org/10.1017/S0022112065001453. Yao, J.X., Zou, Z.J., Wang, H.M., 2011. Numerical study on bank effects for a ship sailing
Phillips, T., 2012. Extrapolation-based Discretization Error and Uncertainty Estimation in shallow channel. J. Shanghai Jiao Tong Univ. (Sci.) 16, 91–96. https://doi.org/
in Computational Fluid Dynamics. 10.1007/s12204-011-1100-0.
Prandtl, L., 1925. Report on the investigation of developed turbulence, Translation of “ Yeung, R.W., 1978. On the interactions of slender ships in shallow water. J. Fluid Mech.
Bericht über Untersuchungen zur ausgebildeten Turbulenz. Z. Angew. Math. Mech. 5 85, 143–159. https://doi.org/10.1017/S0022112078000567.
(2). April 1925. Yoshimura, Y., 1986. Mathematical model for the manoeuvring ship motion in shallow
Richardson, L.F., 1911. The approximate arithmetical solution by finite differences of water. Journal of the Kansai Society of Naval Architects, Japan 61 (200), 41–51.
physical problems involving differential equations, with an application to the
16
Yuan, Z.M., 2018. Ship hydrodynamics in confined waterways. J. Shellfish Res. 1–14 Zhang, S., Tezdogan, T., Zhang, B., Xu, L., Lai, Y., 2018. Hull form optimisation in waves
https://doi.org/10.5957/JOSR.04170020. based on CFD technique. Ships Offshore Struct. 13, 149–164. https://doi.org/
Zeng, Q., Thill, C., Hekkenberg, R., Rotteveel, E., 2018. A modification of the ITTC57 10.1080/17445302.2017.1347231.
correlation line for shallow water. J. Mar. Sci. Technol. 0, 0. https://doi.org/ Zhang, S., Zhang, B., Tezdogan, T., Xu, L., Lai, Y., 2018. Computational fluid dynamics-
10.1007/s00773-018-0578-7. based hull form optimization using approximation method. Eng. Appl. Comput. Fluid
Zeraatgar, H., Vakilabadi, K.A., Yousefnejad, R., 2011. Parametric analysis of ship squat Mech. 12, 74–88. https://doi.org/10.1080/19942060.2017.1343751.
in shallow water by model test. Brodogradnja 62, 37–43. Zhang, S., Zhang, B., Tezdogan, T., Xu, L., Lai, Y., 2017. Research on bulbous bow
optimization based on the improved PSO algorithm. China Ocean Eng. 31, 487–494.
https://doi.org/10.1007/s13344-017-0055-9.
17

10 1016@j Oceaneng 2019 106696

Uploaded by

Copyright:

Available Formats

10 1016@j Oceaneng 2019 106696

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

10 1016@j Oceaneng 2019 106696

Uploaded by

Copyright:

Available Formats

Ocean Engineering xxx (xxxx) xxx

Contents lists available at ScienceDirect

Numerical and experimental study on hydrodynamic performance of ships

Fig. 2. Case-studies: a – Suez Canal case-study, b – rectangular canal.

Fig. 3. KCS ship model input for slender body calculations.

In order to predict the parameters of interest in the modelled Suez

3.3. Reynolds-averaged Navier-Stokes

Perhaps the most rapidly developing field in ship hydrodynamics

4. Results and discussion

In this Section, the results for both case-studies are presented. To

4.1. Sinkage and trim

As mentioned previously, sinkage and trim are of great practical

CFD model. It is important to keep in mind the experimental uncertainty

Fig. 13. Empirical methods comparison for both case-studies.

within waves vanishes at a rate, proportional to the water depth, and

effects decrease in importance with increasing speed. It is also noted that

course of an experiment. From a numerical point of view, these case-

Parameter Symbol Unit Formula Notes

Bow Squat (Barrass, 1981) Sb m 2=3

(continued on next page)

Factor KSP 0:15 ​ Ships with single propellers

You might also like

Factor KSP 0:15 Ships with single propellers