See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/361581245

Numerical Simulations of the Hydrodynamic Performance of the Propeller

with Wake Equalizing Duct behind the Ship

Article  in  Scientia Iranica · June 2022

DOI: 10.24200/SCI.2022.58440.5723

CITATIONS READS

0 98

3 authors:

Saman Rezaei Milad Bamdadinejad

Sharif University of Technology Amirkabir University of Technology
3 PUBLICATIONS   4 CITATIONS    7 PUBLICATIONS   15 CITATIONS   

SEE PROFILE SEE PROFILE

Hassan Ghassemi
Amirkabir University of Technology
301 PUBLICATIONS   2,003 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Investigation of buoy on OC4-DeepCwind semi-submersible platform mooring line systems View project

...Correcting and enriching vessel's noon report data using statistical and data mining methods View project

All content following this page was uploaded by Hassan Ghassemi on 06 May 2023.

The user has requested enhancement of the downloaded file.

Scientia Iranica B (2022) 29(5), 2332{2348

Sharif University of Technology

Scientia Iranica
Transactions B: Mechanical Engineering
http://scientiairanica.sharif.edu

Numerical simulations of the hydrodynamic

performance of the propeller with wake equalizing duct
behind the ship
S. Rezaeia , M. Bamdadinejadb; , and H. Ghassemib
a. Department of Mechanical Engineering, Sharif University of Technology, Tehran, Iran.
b. Department of Maritime Engineering, Amirkabir University of Technology, Tehran, Iran.
Received 28 May 2021; received in revised form 16 February 2022; accepted 27 June 2022

KEYWORDS Abstract. The equalizing wake ow into the propeller behind the ship is important
DTMB4119 propeller; from the hydrodynamic performance viewpoint. In this study, numerical simulations of the
KRISO Container DTMB4119 propeller with two symmetric and asymmetric duct types behind the KRISO
Ship (KCS); Container Ship (KCS) are performed using Computational Fluid Dynamics (CFD). In order
Pre-swirl stator; to improve the wake equaling ow, a combined duct and stators' congurations are installed
Symmetric and before the propeller in the stern of the ship and its hydrodynamic performance is studied
asymmetric ducts; using CFD. A duct with the NACA4415 section and two types of stator congurations are
Propeller selected. The STAR-CCM+ software using the nite volume discretization method was
performance; used to solve the governing equations of the uid ow. For simulating the turbulence model,
Wake equalizing duct. the standard k-! model was used and the solution method was validated in comparison
with the available experimental data. Output parameters including thrust coecient
and torque coecient in the open-water condition and behind the ship are presented
and discussed. The propeller performance after mounting the asymmetric and symmetric
ducts are improved by 4.8% and 6.57%, respectively. Therefore, it is concluded that the
symmetric duct is more aected by the propeller performance and, hence, fuel consumption
is reduced considerably.
© 2022 Sharif University of Technology. All rights reserved.

1. Introduction tors. Thus, decision-making for each device is required

to go through the normal process of technical feasibility
Today, ship owners around the world are looking aspects and economic cost-eectiveness analysis for the
for solutions to three major challenges: diminishing specic ship under consideration.
costs, enhancing eciency, and reducing impact on the One of the ecient devices functions via a com-
environment. Several augmentation devices for energy bination of Wake Equalizing Duct (WED) and Pre-
saving can be considered. It should be noted that the Swirl Stator (PSS), as shown in Figure 1. The stator
applicability of such devices and technologies depends blades create a pre-swirl, giving the propeller a more
on ship type, ship size, operation prole, and other fac- favorable angle of attack. The duct increases ow
velocity towards the propeller and creates a forward-
*. Corresponding author. directed force with its wing-shape section. These fea-
E-mail address: miladbamdadi@aut.ac.ir (M. tures facilitate possible energy saving up to 8% for
Bamdadinejad) full-form slower ships, such as tankers and bulkers.
The improvement of the propulsion eciency of these
doi: 10.24200/sci.2022.58440.5723 types of ships will contribute largely to the reduction
 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348 2333

Figure 1. (a) Wake equalizing duct mounted on the ship's stern, (b) Asymmetric duct.

of polluting emissions and to the saving of fossil fuels in (RBCF) device optimization and Computational Fluid
shipping [1]. According to the statistical studies on the Dynamics (CFD) analysis. In order to solve the RANS
fuel consumption of various vessels since the year 1980, equations, parametric descriptions of RBCF properties
even a 1% reduction in the fuel consumption of an oil and an optimization algorithm were applied in Open
tanker that consumes 60{100 tons/day (when moving) FOAM software. The results demonstrated that the
and imposes a daily cost of about $35{50 thousand eciency of the model scale increased by about one
on the owner (increasing exponentially based on the percent, which could reach a signicant amount of four
voyage length/time) is very important [2]. Figure 1(a) percent in the unrened propeller design [17]. Nowruzi
shows a WED mounted on the ship's stern [3], and and Naja (2019) investigated the eects of three
Figure 1(b) shows an asymmetric duct [4]. dierent pre-swirl ducts on the propulsion performance
Generally, two approaches are employed to re- of a Series-60 ship by experimental and numerical
duce engine power. One approach is to design an methods [18]. Tacar et al. (2020) used CFD analysis
optimum hull that reduces the resistance, and the and experiments of a new model with a larger model of
second approach is to improve the propulsion eciency. a container ship to investigate the Gate Rudder system
To improve the thrust and eciency, many types of and the eect of Gate Rudder on ship performance [19].
Energy-Saving Devices (ESDs) are employed in three Obwogi et al. (2021) studied the eect of rudder bulb
zones of the ship stern; see Chapter 13 of the referenced diameter, thrust n, span, chord length, and angle of
book authored by Carlton [5]. attack on propulsion eciency using CFD [20].
Many numerical and experimental works were BEM and CFD simulation methods are widely
carried out to analyze and design the ESD [6{11]. used as important tools to improve the eciency
Anderson (1988) applied the VLM to obtain the hy- and performance of the propeller design. Ghassemi
drodynamic behavior of the propeller performance [12] et al. published many types of research by numerical
using the Vortex Lattice Method (VLM) and dening methods (BEM and CFD) to simulate the propeller
a propeller blade surface replaced by the vortex dis- performance with dierent congurations under dier-
tribution. Later, many researchers have worked more ent operating conditions [21{27].
precisely on these methods and used the Boundary WED is one of the most common ESDs used
Element Method (BEM) separately to study the ow to improve the propulsion performance of a ship,
behavior in steady-state ducted propellers [13,14]. Balt- propeller-excited vibrations, and viscous resistance
hazar et al. (2012) evaluated the torque and thrust of a forces. Many studies have been done on WED for the
ducted propeller in an open water mode using the panel past three decades, most of which have attempted to
method where the propeller hydrofoil is modeled as a increase propulsion eciency [28]. A commercial code
well or spring. In this method, conditions considered called \Comet" was developed by Ok and Hamburg-
as the problem default are closer to those governing Harburg (2005) using the RANSE method to study the
the propeller performance in open water and the hull ow around a WED [29]. Korkut (2006) conducted
and wake eects on the propeller eciency are stud- a study on energy saving in powering characteristics
ied [15]. Lee et al. (2016) employed a new structural of cargo ships using the concept of partial WED [30].
safety assessment method for ESD. In this method, Celik (2007) explored the eect of WED on the propul-
Morison equation was solved for a velocity at a certain sion performance of a chemical tanker using the RANS
probability level using two neural network methods as numerical method for various WED arrangements at
well as time domain simulation and Gamble tting dierent speeds of many ships [31]. Heinke and Hellwig-
method [16]. Gaggero et al. (2018) evaluated propulsion Rieck (2011) investigated the eect of the Reynolds
eciency improvements using Propeller Boss Cap Fin number on the ow around the appendages and on
2334 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348

the propeller in a typical container ship by WED and discusses how it works. The objective is to determine
vortex generator ns [32]. A model was developed to the hydrodynamic performance of this propeller by
improve the performance of a PSS and was used to installing new duct and PSS, while no one worked on
estimate power savings. In this study, the circulation this special type of propeller.
distribution method was employed based on the char- To evaluate this propeller type and the proposed
acteristics of the propellers [33]. Using a numerically combination of the duct and PSS, two types of symmet-
variable method, Lee et al. (2019) designed an asym- ric and asymmetric PSS including ducts at dierent
metric stator using an auxiliary function by displaying velocities have been investigated. The hydrodynamic
the lifting lines of a vortex propeller and a stator [34]. performance of the propeller (thrust and torque coef-
Han (2019) estimated the energy saving eciency of a cients) under four cases (open-water, without duct,
PSS using the numerical method based on the lifting- symmetric duct, and asymmetric duct) at dierent
surface method as well as the RANS equations, and the velocities are presented and discussed.
viscous ow around the hull in three systems: propeller The remainder of the paper is organized as fol-
in open water, the towing resistance experiment, and lows. Section 2 describes the governing equations and
the self-propulsion test [35]. In another study, the eect dierent turbulence models. Section 3 presents nu-
of angle-axis PSS was investigated using a suitable merical implementation, computational domains, and
design method for each blade or location for the radius boundary conditions. Section 4 presents and discusses
with respect to the hydrodynamic pitch angle in order the numerical results of the propeller performance
to improve the propulsive eciency in non-uniform under dierent upstream wake equalizing devices at
ow elds of the stern [36]. Furcas et al. (2020) pro- various speeds. Finally, Section 5 draws the conclu-
posed a Simulation-Based Design Optimization method sions.
(SBDO) to design an ESD based on the concept of
WED [37]. Nadery and Ghassemi (2020) carried out 2. Governing equations
the hydrodynamic performance of the KP505 propeller
behind the KRISO Container Ship (KCS) with and Here, the ow simulation and propeller modeling are
without PSS and duct. It was concluded that good the main concerns, and the uid ow is studied in
design increased eciency by 1.67% and a bad design the control volume; therefore, a conversion tool that
might reduce eciency by 3.25% [38]. Koushan et al. can provide its \general transfer theorem" is required.
(2020) numerically and experimentally investigated the Governing equations use two basic hydrodynamic equa-
eect of a PSS on propulsion eciency in both model- tions: conservation of mass (the continuity equation)
scale and full-scale modes [39]. Recently, Nadery et al. and conservation of momentum (Navier-Stokes equa-
(2021) numerically evaluated the impact of four ESDs tions).
including PSS, WED, and Pre-Swirl Ducts called PSD- 2.1. Continuity equation for the turbulent ow
1 and PSD-2 to improve the propeller performance of In its general form, the continuity equation is expressed
the KCS container ship [40]. Another study examined as follows:
the eect of a new PSS conguration connected to the
KP505 propeller on the propulsion performance of the @
+ div(U ) = 0; (1)
KCS container ship [41]. Su et al. (2021) performed @t
numerical and experimental analyses on a 25-meter ore where  and U are the density and velocity vectors
carrier model to investigate the operation of ESDs, of the uid, respectively. This equation is valid for
including a PSS and a rudder bulb [42]. Guo et al. the instantaneous values of the turbulent ow. The
(2021) used a bulk carrier-scale model to study the temporal averaging of Eq. (1), replacing momentary
ow mechanism of a ship and the working principles quantities with average temporal values plus the uc-
of energy-saving appendages [43]. Qin et al. (2021) nu- tuating values, and using Reynolds averaging rules will
merically measured the ow eect of a pre-swirl pump yield Eq. (2) as follows:
jet propulsor based on improved delayed detached eddy
simulation [44]. Mikkelsen et al. (2022) examined the @ @
@  0 0
+ ui +  ui = 0; (2)
nominal wake elds in ve dierent heading conditions @t @xi @xi
for the KCS container ship in regular waves with a where  and 0 are the average density and the average
wavelength equal to the length of the ship [45]. oscillation density; and ui and u0i are the average ve-
Following a detailed review of the literature, the locity and the average oscillating velocity, respectively.
most published works in this eld revolve around For an incompressible ow, since 0 = 0, Eq. (2) will
the propeller KP505 and the KCS ship hull, which be as follows:
were investigated and presented by the third author
of the present paper. However, this study provides a @ui
= 0: (3)
new propeller of DTMB4119 matching the KCS and @xi
 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348 2335

2.2. Momentum equation for the turbulent the scale length of large vortices in the turbulent ow
ow eld. Thus, we have:
Momentum equations for an incompressible viscous
ow are as follows: k2
   t = C  ; (9)
@ui @u @P @ 2 ui "
 + uj i = Bi + ; (4)
@t @j @xi @xi @xj where t is called turbulent viscosity and C is equal
where Bi , P , and  are the body force, uid hydro- to 0.09. " and k are found by semi-experimental
static pressure, and uid viscosity, respectively. This Eqs. (10) and (11). In the following relation, " is the
equation is valid for both steady and turbulent ows; in turbulent Schmidt number and equal to 1.3, k is the
the latter, dependent variables (velocity, pressure, etc.) Prandtl number and equal to 1, G (shear production
are all time dependent. Expressing Eq. (4) in terms of term) is the turbulence kinetic energy produced by the
average temporal quantities, i.e., placing ui = u0i + ui average ow-turbulent ow eld interaction, and B is
and Pi = Pi0 +Pi (ui , u0i , and ui are called instantaneous the production-buoyancy loss due to the ow uctu-
velocity, oscillating velocity, and average velocity and ating density eld. C1 , C2 , and C3 are called model
Pi , Pi0 , and Pi are called instantaneous hydrostatic constants, which are 1.44, 1.92, and 0:33, respectively.
pressure, oscillating hydrostatic pressure, and average  
@k 
hydrostatic pressure, respectively) in Eq. (4), and  + uj kj =  + t kj +G+B "; (10)
simplifying and re-applying the temporal averaging will @t k j
yield Eq. (5) as follows:  
@"  "
@ui @u @u0
!
@Pi @ 2 ui  + uj "j =  + t "j + C1 G
 + uj i + u0j i = Bi + ; (5) @t " j k
@t @xj @xj @xi @xi @xj
" "2
where Bi and Pi are average body force and average + C1 (1 C3 ) B C2  : (11)
k k
uid hydrostatic pressure, respectively. Term three
on0 the left side is usually expressed dierently. Since 2.4. k ! SST model
@uj The SST is, in fact, an optimized k ! standard
@xj =0 is a continuity equation of incompressible ows, model, which is itself a modied form of the Wilcox
0j
adding and subtracting u0i @u @xj to and from the sides model at low Reynolds, compressibility, and shear ow
of Eq. (5) will result in the turbulent ow momentum dispersion eects. To use both k ! and k " equations,
equation as follows: an integration function is introduced which equals 1 in
   
areas near the wall (to activate k ! model) and zero in
@ui @u @P @ @u those far from the wall (to activate k " model); here, a
 + uj i = Bi +  i u0i u0j : (6) negative point is the possibility of instability and poor
@t @xj @xi @xi @xj
convergence due to model-to-model switching.
The only dierence between Eq. (6) and momentum
equation with instantaneous quantities lies in the addi-   
@ (k) @ @ t @k
tion of the last term, i.e., u0i u0j , which is idiomatically @t
+ (U k) =
@xj j @xj
+
k @xj
called the turbulence tension or Reynolds's tension.
2.3. Standard k " model + Pk  0 k! + Pkb ; (12)
Due to its simplicity in understanding and program-   
ming, this model is very popular. It is used in the @ (!) @ @  @!
+ (Uj !) = + t
Boussinesq relationship and expresses the turbulent @t @xj @xj ! @xj
eld using two variables: the turbulent kinetic energy
(k) (Eq. (7)) and the rate of dissipation of turbulent !
+  Pk !2 + P!b : (13)
kinetic energy (") (Eq. (8)): k
1 In Eqs. (12) and (13), U is the velocity vector,  the
k = u0i u0j ; (7) density, and P the turbulence production rate. Table 1
2
 
provides a list of the constants used in the equations.
 0 0
"= u u : (8) Table 1. The stability used in the turbulence model.
 i;j i;j
   0  k !
Dimensional analyses have shown that the turbulent
viscosity is directly proportional to the velocity and 5.9 0.075 0.09 2 2
2336 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348

3. Numerical implementation
In the CFD, the meshing of the computational domain
is of great importance. It seems that more than 50%
of the entire endeavor is dedicated to dening geom-
etry and meshing production. CFD is known as an
important tool for the design of industrial products
and processes related to uids engineering because the
development of strong hardware in the 1990s led to
a considerable growth in this eld and the numerical Figure 2. Dierent view of DTMB4119 propeller.
method entered the computation eld on a larger scale.
In numerical solutions, dierent methods namely
the Finite Dierence Method (FDM), Finite Volume
Method (FVM), Finite Element Method (FEM), and
BEM are used to solve the equations. Since this study
uses the FVM, it is described here with its solution
algorithm which involves the following four steps:
(a) Integrating the equations governing the uid ow
on the control volume;
(b) Discretizing and converting the integral equations Figure 3. Propeller with the duct in the stern.
into a set of algebraic equations;
modeling sensitivity, it is quite necessary to design
(c) Selecting the solution method of the basic equa- carefully and pay attention to the geometry structure
tions governing the ow (Navier-Stokes and conti- of the propeller under consideration. This research used
nuity equations); the DTMB4119 propeller comprising three blades. Due
(d) Solving the set of algebraic equations. to its high accuracy and capability, the Rhino software
was used to generate the propeller geometry and make
The STAR-CCM+ software was used to numerically a smooth and fair surface. Figure 2 shows dierent
simulate all parts of this study. Since the ows around views of the propeller being studied and Table 2 lists its
the ship had high Reynolds numbers, the turbulent ow specications. In addition, the propeller with the de-
method was selected using the k " turbulence model signed duct on the stern of the ship is shown in Figure 3.
for ship modeling. The y+ values were evaluated and
selected based on the turbulence model. Free surface 3.1.2. The KRISO Container Ship (KCS)
boundaries were calculated using the Volume Of Fluid The KCS is a modern container ship with a bulbous
(VOF) method. Also, the standard k " turbulence bow used to explore conceptual data in ow physics
model and numerical method of the rotating ow of as well as validate CFD models [46,47]. The KCS
the propeller Multiple Reference Frame (MRF) were container ship model was used to evaluate the propeller
used for propeller modeling. Here, the next steps are: performance in the wake eld. Figure 4 and Table 3
show the geometry and features of the model under
1. Generating a problem geometry; study; the main ship was scaled by 0.03165 coe-
2. Creating a solution domain, producing a mesh or cient. DTMB4119 and KP505 as two standard marine
discretizing the domain; propellers are used for the validation of numerical
3. Explaining software settings for simulation pur- modeling in several types of research, although the
poses; KP505 is dened as the original propeller used for KCS
4. Describing the general trend of the problem solu- Table 2. Principal parameters of DTMB4119 propeller.
tion.
Propeller diameter (m) 0.3048
3.1. Test case: Model of ship hull and Expanded Area Ratio (EAR) 0.607
propeller Number of blades 3
3.1.1. DTMB4119 propeller Pitch ratio Variable
To calculate the propeller hydrodynamic performance Rake (deg) 0
from either an empirical or theoretical standpoint, Skew (deg) 0
it is essential to have a detailed understanding of Blade section NACA66 a = 0:8
propeller geometry and the corresponding denitions
used. Depending on the type of analysis and the Rotation direction Right
 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348 2337

3.2. Computational domain

3.2.1. Computational domain of the ship model
To ensure a viable problem-solving capability and high
accuracy and to simulate the ow around the ship
model, the STAR-CCM+ software was selected for the
present calculations. Considering the studies done on
the modeling of the ship motion, a rectangular cube
was selected for the computational domain where the
ship motion simulation was performed, instead of mov-
ing the ship, based on the assumption that water owed
from one side of the domain as the input and hit the
ship front. This is, in fact, a simulation method of the
ship motion without using highly altered meshing that
speeds up the process and reduces the computational
cost while maintaining accuracy. The ship is a 2DOF
object dened in the heave and pitch directions and
is modeled using a two-phase model. Since the ship
structure is symmetrical, only half of it was simulated.
Figure 5 shows the computational domain dimensions
Figure 4. Dierent view of the KCS model. and plan in terms of the ship length.
Table 3. Principal parameters of the KCS model. 3.2.2. Computational domain of the propeller model
Here, two types of domains for the propeller simulation
Hull parameter Value were used, i.e., cubical and cylindrical, with equal
Length between perpendicular, LPP (m) 7.2786 distances (Figures 6 and 7). After checking the
Length of Water Line, LWL (m) 7.357 numerical results with experimental data, the cubical
domain was selected due to fewer errors (see Section 4).
Breadth, B (m) 1.019 In simulations, domains are both xed and rotating and
Depth, D (m) 0.6013 modeling of the rotational motion of the propeller is
Draft, T (m) 0.3418 done by the rotating disk (rotating cylinder) method;
in addition, that of the water ow is based on the xed
Wetted area surface (m2 ) 9.4379 domain which involves the rotating domain.
Midship section coef. [-] 0.9849
Block coef. [-] 0.6505
3.3. Meshing of the computational domain
Many encountered problems in the real world involve
complex geometries for which creating a structured
and modeling the wake pattern DTMB4119 was not block mesh of hexagonal or quadrilateral cells is time-
applied to KCS hull. Thus, an attempt is made here intensive. However, the computational time and,
to examine its performance for KCS in our research as hence, the cost can be reduced using unstructured
an innovative act. meshes with triangular-quadrilateral cells; if the geom-

Figure 5. Schematic of the computational domain and boundary conditions of the ship.
2338 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348

Figure 6. Rectangular cubic computational domain and boundary conditions of the propeller with the rotating domain.

Figure 7. Cylindrical computational domain and boundary conditions of the propeller with the rotating domain.

Figure 8. The meshing of the computational domain.

etry is simple, both ideas will take equal time. Another Table 4. Specications of the boundary layer.
point worth noting in meshing and modeling is that Number of layers 12 layers
the mesh and ow regime should match each other. First layer thickness 0.1 mm
Since convergence takes place better and faster for
hexagonal-quadrilateral meshes, a great deal of eort The overall thickness of the boundary layer 3 mm
was made for all areas to be created with these meshes
to save both solution time and cost and reach more meshes should be small enough to sense the eects of
accurate solutions in the end. the free surface. Therefore, about 6,000,000 meshes
Here, the emphasis is on the free water surface were used for meshing the ship hull. Specications of
meshing because of the ship's two-phase computational the important hull boundary layer are listed in Table 4,
domain. To increase the computational accuracy in and the computational domain meshing is shown in
the front and rear parts of the ship greatly aecting Figure 8.
the output results, a separate control volume was Since the propeller is fully submerged and not
established to make meshes ner. The hull periphery considered as a free surface, the computational domain
 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348 2339

Figure 10. The y+ contour on the propeller.

Table 5. Comparison of resistance coecient at the

model speed of VM = 2:196 m/s (F n = 0:26).
Figure 9. Computerization of the propeller's
computational domain. Resistance Calculated Exp. Error
coef. (%)
is considered in a single phase. To increase the accu- 103 Ct 3.415 3.554 4.1%
racy, ner meshes were used in the rotating cylindrical
sub-domain where it is close to the propeller blades, Table 6. Numerical characteristics used in the software.
and to reduce the computational costs, meshes used
far from the propeller were relatively larger (Figure 9). Characteristics Propeller Ship modeling
Upon applying meshing and wall functions, the modeling
rst node distance from the wall surface called y+ The solver Open water Behind the ship
should be considered. If the latter is too large, the Time step 0.0005 0.04
rst node may lie beyond the boundary layer, causing
the functions to be applied too far from the wall, and Number of loops
| 10
if it is too small, this node will lie in the steady zone in each time step
of the boundary layer where the related functions are Turbulent model k ! SST k "
invalid. The rst wall thickness of the boundary layer
(
y) can be obtained through the approximation of
Eq. (14) as follows: conditions, subdomains, boundary conditions, and so
p on are the issues that require specic techniques to be

y = 74Ly+ ReL 13=14 ; (14) determined in the software setting. Setting bound-
where L is the ship length and ReL is the Reynolds ary conditions that can accurately re ect the real
number based on the ship's length. conditions is important in acquiring accurate results.
However, the use of wall functions will reduce the Boundary conditions are a set of features and condi-
separation estimation and accuracy in vortex ows. To tions that need to be fully dened on the boundaries
use the SST and k ! turbulence models, we should of the solution domain so that the ow is simulated.
have y+<300, and to benet from low-value Reynolds In the solver section, the maximum iterations are
models, y+ <2 is required which is not quite easy due determined for solving the governing equations in each
to the massive computational volume in most industrial time step and the remaining is used to control the
applications. Figures 10 and 11 show the contour y+ on solution of the equations. The number of iterations
the propeller and the ship hull, respectively. Table 5 and the acceptable remainder are important factors in
shows the comparison of the calculated resistance the solution time. To simulate the ship motion, the
coecient (Ct ) of the bare hull with the experimental unsteady method is employed to work out a solution,
value at a speed of 2.196 m/s (F n = 0:26). This and in the case of the propeller, the steady method
speed is the service speed for the KCS which takes is adopted. Table 6 shows the characteristics of the
24 knots. It can be seen that there is close agreement applied numerical model.
between the calculated and experimental values and
the corresponding relative error is about 4%. 3.5. Meshes of the solution strategy
The general simulation process and modeling strategy
3.4. Computational settings are shown in Figure 12. The simulation takes ve
Specications of the ship, propeller, domain, solver general steps to complete:
2340 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348

Figure 11. The y+ contour on the ship hull.

Figure 12. The general simulation process and modeling strategy.

(a) Modeling the ship. In this case, upon initiating the problem was solved by dening the center of
the software and dening the ship geometry, a gravity and ship velocity and setting the meshes
square box was plotted as the solution domain. and time steps. To obtain the problem outputs,
Since the ship hull was symmetrical, half of it a circular plate was dened and the ow velocity
was simulated to speed up the calculations and results were recorded on it;
 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348 2341

(b) Modeling the propeller in open water condi- other areas, because the physical phenomena aect it
tions. In this case, mesh settings were performed considerably and a steady and high-quality free surface
for the steady solution after dening the propeller is produced. In this study, modeling of the propeller
geometry in the software and determining the performance is done based on dierent meshing for
solution domain (input, output, and wall). The which the thrust coecient is a comparison factor.
thrust and torque coecients and the eciency In this study, three types of unstructured meshes
(according to the last recorded results) were ex- were used to calculate the values of thrust, torque,
tracted following the convergence of the solution and ow eld around the propeller. In this method,
results; a trimmed mesh with the surface growth rate of 1.3
(c) Modeling the propeller behind the ship. In and fast volume growth rate was used. Therefore,
this case, besides observing the solution and the results were obtained using a smaller number of
geometry conditions, the data recorded on the meshes. For grid renement, the convergence analysis
circular plate were added to the simulation as the method was employed for the results of mesh indepen-
input (input ow) and in doing so, the eect of dence. The outcome was obtained from four mesh sizes
the ship hull axial wake was considered in the (coarse, medium, ne, and ner) for the propeller.
propeller performance. The recorded input in this Based on the Grid Convergence Index method
step is saved to an Excel le. The data include (GCI) proposed by Celik et al. in 2008 [48], the mesh
the velocity components (i; j; k) at a known local generation analysis was validated. In this method, the
point; apparent orders (p) and q(p) are dened using Eqs. (15)
and (16) as follows:
(d) Wake measurement before duct installation
location. In this step, the hull wake is measured p=
1
jln j"32 ="21 j + q(p)j ; (15)
again and stored before the point intended for the ln(r21 )
duct installation;  p 
(e) Modeling the symmetrical and asymmetri- r21 s
q(p) = ln p ; (16)
cal duct sections behind the propeller. The r32 s
duct is designed in the Rhino software based on the where s is calculated by s = 1. sign("32 ="21 ).
wake model recorded in step (b) and placed in the Here, the grid renement factors for four dierent
front part of the propeller. The problem is then meshes, i.e., (1) ner, (2) ne, (3) medium, and (4)
solved to extract the thrust and torque coecients coarse, are r21 = h2 =h1 , r32 = h3 =h2 , and r43 = h4 =h3
(Figure 12). (hi is the basic mesh size). Moreover, for the intended
parameter of ' (in the current study KT ), "i+1;i is
4. Results and discussion calculated by "21 = '2 '1 , "32 = '3 '2 , and "43 =
'4 '3 . In this case, the extrapolated value is dened
Before presenting and analyzing the results obtained as Eq. (17):
from the numerical solution of the ow on the selected
ship and propeller geometries, it is necessary to check p '
r21 1 '2
the solution accuracy and study the issues related to '21
ext = p : (17)
r21 1
mesh independency and then, to verify the accuracy
of the results after ensuring their independence from Finally, the relative approximate error (Eq. (18)),
the meshing size. A very important issue of numerical extrapolated relative error (Eq. (19)), and ne-grid
solutions is to validate the obtained results by compar- convergence index (Eq. (20)) were dened as follows:
ing them with the experimental, numerical, or semi-  
analytical results.  '1 '2 
e21
a =

; (18)
'1
4.1. Mesh generation analysis  
Mesh size and mesh number are important parameters  '12
 ext '1  ;

in the numerical analyses; when meshes increase, the '21
ext =  '12  (19)
ext
computational time increases too and the existing
hardware fails to conduct the analyses. Since ner 1:25e21
meshes experience increased computational time and GCI21
ne = p
a : (20)
r21 1
cost and larger ones cause computational errors, their
size and number should be selected such that the As shown in Table 7, the error rate is very low given
analyses may have the least error and the existing that the mesh number is 1.1e6. Thus, this mesh
hardware can perform the calculations. Meshes on the number will be enough for the propeller calculations.
ship hull are considered to be much ner than those on In Table 8, the values of these parameters are computed
2342 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348

Table 7. Mesh independency and mesh size eect.

Number The percentage error rate Basic Mesh size
of mesh KT of thrust compared to mesh size increase
(N) the previous case (m) coecient
750,000 0.153548792 | 0.022 |
800,000 0.152624435 0.65% 0.016 1.062
1,100,000 0.151853246 0.65% 0.012 1.27
1,410,000 0.151264583 0.38% 0.011 1.21

Table 8. Discretization error for propeller thrust (T ) based on grid convergence method.
N1 (ner) 1,410,000 N3 (medium) 800,000
N2 (ne) 1,100,000 N4 (coarse) 750,000
h1 0.011 h3 0.016
h2 0.012 h4 0.022
r21 1.09 r32 1.33
r43 1.375 '1 0.151264583
'2 0.151853246 '3 0.152624435
'4 0.153548792 "21 0.000588663
"32 0.000771189 "43 0.000924357
q {1.389653346 p 12.991455817
'21
ext 0.151746354 e21
a 0.389161156%
e21
ext 0.236572241% GCI21
ne 0.2357337073%

Table 9. Comparison of the open-water characteristics of the DTMB4119 propeller (cubical domain).
J KT (Num) KT (Exp) 10KQ (Num) 10KQ (Exp)  (Num)  (Exp)
0.5 0.301 0.281 0.495 0.463 0.484 0.483
0.7 0.218 0.207 0.389 0.363 0.624 0.635
0.833 0.16 0.155 0.301 0.28 0.704 0.734
0.9 0.125 0.123 0.263 0.243 0.681 0.725
1.1 0.037 0.037 0.122 0.112 0.537 0.578
Ave. error (%) 3.49% 7.18% 3.97%

for the considered variable of KT . As shown in Table 8, should be noted that the numerical model used is the
Ni is the total number of cells; the thrust coecient is same as the experimental model. In these tables, J , KT ,
calculated; and the maximum uncertainty is obtained KQ , and  are the advance coecient, thrust coe-
as follows: GCI21
ne = 0:2357337073%. cient, torque coecient, and eciency of the propeller,
The selected time step is calculated such that respectively. These parameters are dened as follows:
it can rotate between 0.5 and 2 degrees in each step T Q
according to the recommendations made in Interna- KT = 2 4 ; KQ = 2 5 ; (21)
tional Towing Tank Conference (ITTC) [49]. Here, the n D n D
time step for the propeller is 0.0005, which allows the J KT VA
propeller to rotate at 1.9 degrees. = ; J= ; (22)
2 K Q nD
4.2. Validation of the propeller results where n, D, T , and Q are rotational speed, diameter,
The open-water characteristics of the propeller of thrust, and torque of the propeller, respectively. In
DTMB4119 are compared with the experimental Eq. (22), VA is the advance velocity.
data [50,51] in two computational domains (cubical and Tables 9 and 10 compare the open-water char-
cylindrical domains), as shown in Tables 9 and 10. It acteristics of the DTMB4119 propeller in the cubical
 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348 2343

Table 10. Comparison of the open-water characteristics of the DTMB4119 propeller (cylindrical domain).
J KT (Num) KT (Exp) 10KQ (Num) 10KQ (Exp)  (Num)  (Exp)
0.5 0.308 0.281 0.515 0.463 0.483 0.483
0.7 0.214 0.207 0.389 0.363 0.635 0.635
0.833 0.159 0.155 0.305 0.28 0.734 0.734
0.9 0.124 0.123 0.263 0.243 0.725 0.725
1.1 0.039 0.037 0.121 0.112 0.578 0.578
Ave. error (%) 4.1% 8% 4.28%

Table 11. Predicted values of wake led behind the KCS.

VS (kts) VM (m/s) Fn w
16.83 1.524 0.18 0.278
21.59 1.955 0.23 0.278
23.56 2.134 0.25 0.277
24.00 2.196 0.26 0.275
28.03 2.539 0.30 0.229
30.29 2.743 0.32 0.232

calculated by w = 1 VA =VS . Table 11 presents the

predicted values of the wake factor at dierent model
Figure 13. Comparison of the open-water characteristics speeds. In this table, F n, VS , and VM are Froude
of the DTMB4119 propeller (cubical domain). number, ship speed, and model speed, respectively.
The predicted values of wake factors are found to be
and cylindrical domains, respectively. The average around 0:22  0:28 in which the model speed is changed
relative errors of the thrust and torque coecients from 1.524 (F n = 0:18) to 2.743 m/s (F n = 0:32). All
and eciency in the cylindrical domain are 4.1%, 8%, numerical results are between the ranges simulated for
and 4.28%, respectively. The average relative errors the wake factor values for KCS. To be specic, they
of the thrust and torque coecients and eciency were in between the range 0:25 < w < 0:28 [52]. As
in the cubical domain are 3.49%, 7.18%, and 3.97%, an example, Figure 14 shows the axial and cross- ow
respectively. Thus, the results indicate that the cubical velocity contour behind the ship at a model speed of
domain functions relatively better than the cylindrical 2.134 m/s (F n = 0:25). The wake factor at this speed
domain. This domain is generalized to the propeller is 0.277 and advance velocity is obtained at 1.542 m/s.
performance in the wake eld including WED (duct
and stator). Figure 13 shows the comparison of the 4.4. Propeller performance under the wake
open-water characteristics of the DTMB4119 propeller eld
in the cubical domain. At the design speed (J = 0:833), After predicting the wake eld, applying it to the solu-
the results of the thrust and torque coecients are in tion domain, and checking the propeller performance,
good agreement with the experimental value. the wake eld is dened in the solution domain as
the input ow. Since the wake eld was recorded in
4.3. Wake eld prediction the previous section at a point before water entered
When the ship is moving, the wake ow behind the ship the propeller, it seems that the propeller is located
appears to be complicated ow which one can charac- behind the ship in this simulation and its performance
terize it as unsteady, non-uniform, and rotational due is investigated in the presence of the ship hull and
to boundary layer, shape of ship hull, and other factors. its resulting ow. The numerical results of propeller
Each ship has a typical wake eld. To predict the wake performance in the wake eld are shown in Table 12.
eld behind the ship, the ship hull should be modeled As given in the reference site [53], the service speed
in the computational domain. The velocity eld at the for the KCS takes 24 knots and with a scale model of
propeller position can be determined. The velocities 31.599, the advance velocity is 2.196 m/s. As shown in
behind the ship can be predicted using the STAR- this table, the thrust and torque coecients decrease
CCM+ software. By obtaining the advance velocity and eciency increases with an increase in the model
behind the KCS ship, the predicted wake factor (w) is speed. The eciency of the ship model at a service
2344 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348

Table 12. Propeller performance in the wake eld.

VM w VA J KT 10KQ 
(m/s) (m/s)
1.524 0.278 1.1 0.36 0.351 0.566 0.355
2.134 0.277 1.542 0.506 0.277 0.465 0.477
2.196 0.275 1.592 0.522 0.266 0.453 0.489
2.539 0.235 1.940 0.63 0.212 0.380 0.560
2.743 0.232 2.105 0.69 0.186 0.345 0.593

Figure 15. Overview of the symmetrical duct.

Figure 14. Axial and cross- ow velocity contours behind

the ship at the model speed of 2.134 m/s (F n = 0:25): (a)
Axial velocity, (b) along the y-axis, and (c) along the Figure 16. (a) Symmetric duct. (b) Asymmetric duct.
z -axis.
speed of 2.196 m/s is 0.489, which seems to be low for (symmetrical and asymmetrical) is selected. Figure 15
this type of DTMB4119 propeller. shows an overview of the duct and symmetrical duct
mounted at the stern of the ship. Figure 16 shows
4.5. Symmetric and asymmetric ducts two types of ducts and stators, which are symmetrical
The need to reduce fuel consumption and improve the and asymmetrical, respectively, and their eects on
environmental conditions led to the design of the de- the propeller performance are shown. The duct has
sired duct and stator to the ship, adapted from similar NACA4415 section type and the stator is NACA0012
installed cases in the industry. In the present paper, a section type. The symmetric type has 5 stators placed
duct with NACA4415 section and two types of stators inside the duct in symmetry to the vertical axis, while
 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348 2345

Table 13. Propeller performance in four cases (open-water, without duct, symmetrical duct, and asymmetrical duct).
Behind the ship
Asymmetric duct Symmetric duct Without duct Open-water Velocities
10KQ KT 10KQ KT 10KQ KT 10KQ KT VA w VM
(m/s) (m/s)
0.477 0.285 0.487 0.292 0.465 0.275 0.511 0.306 1.542 0.277 2.134
0.465 0.277 0.473 0.283 0.453 0.266 0.499 0.298 1.592 0.275 2.196
0.392 0.222 0.395 0.226 0.380 0.212 0.428 0.244 1.940 0.229 2.539

Figure 17. Comparison of the thrust coecient under

dierent conditions (open-water, without duct,
symmetrical duct, and asymmetrical duct) at three
advance velocities. Figure 18. Comparison of the torque coecient under
dierent conditions (open-water, without duct,
the asymmetric type has 4 stators, 3 of which are symmetrical duct, and asymmetrical duct) at three
arranged on the right side and one on the left side. advance velocities.

4.6. Hydrodynamic eect The purpose of this paper was to determine the hy-
As is evident, the duct provides a condition where the drodynamic performance of the DTMB4119 propeller
ow is in equilibrium with the propeller, which, in turn, by installing new ducts and stators. It appears that
improves the axial velocity. Here, the thrust and torque one author has previously explored this particular type
are presented in Table 13 in four cases (open-water, of propeller combined with the stator and duct. To
without duct, symmetrical duct, and asymmetrical validate the present numerical results, the available
duct) and three velocities. Since the advance velocity in experimental data were used for the propeller under the
open water is greater than that in wake ow, the thrust open-water condition and resistance coecient of the
and torque coecients are larger than those behind KCS. Two computational domains of the cubical and
the ship. On the other hand, three cases associated cylindrical were used. Thrust and torque coecients
with behind the ship (without duct, symmetric duct, of the propeller were determined at dierent advance
and asymmetric duct) show that the thrust and torque coecients. Then, the propeller was investigated under
for the symmetric duct are of larger values than the the wake ow of the KCS. After predicting the wake
two other cases (asymmetric duct and without duct). behind the ship, two types of ducts (symmetrical and
To ensure better display, these data are shown in asymmetrical) were mounted in front of the propeller.
Figures 17 and 18. Based on the results, the gain The following conclusions are drawn from the ndings
of the symmetric duct is 6.57%, while the gain of the of this study:
asymmetric duct is 4.8%.
ˆ At the model speed of 2.196 m/s (as a service speed
of the KCS), the resistance coecient pointed to
5. Conclusions the good agreement between the present numerical
This research studied the Wake Equalizing Duct result and experimental value;
(WED) and stator mounted in front of the DTMB4119 ˆ For propeller open-water calculations, two compu-
propeller behind the KRISO Container Ship (KCS). tational domains of the cubical and cylindrical were
2346 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348

used. It was concluded that the cubical domain B Production buoyancy loss
caused less error in the numerical solution than the D Depth of ship
cylindrical domain. The average relative errors for Fn Froude number
the thrust and torque coecients and eciency were
3.49%, 7.18%, and 3.97% in the cubical domain, G Turbulence kinetic energy
respectively; GCI Grid Convergence Index
ˆ The eciency was reduced when the propeller op- J Advance coecient
erated behind the ship compared to the open-water L Ship length
condition. At the service speed of 2.196 m/s, the LPP Length between Perpendiculars
eciencies were 0.497 and 0.489 in the open-water LWL Length of Waterline
condition and behind the ship, respectively; P Hydrostatic pressure
ˆ Improvement rates for the propeller performance T Draft of ship
after mounting the asymmetric and symmetric ducts
were 4.8% and 6.57%, respectively. U Velocity vector
w Wake factor
Nomenclature k Kinetic energy
" Rate of dissipation of turbulent kinetic
Bi Average body force energy
Pi Average uid hydrostatic pressure  Eciency
ui Average velocity  Density vector
u0i Average oscillating velocity u0i u0j Reynolds's tension
0 Average oscillating density ' Intended parameter
hi Basic mesh sizes
Bi Body force References
Ct Resistance coecient 1. Ziarati, R., Bhuiyan, Z., De Melo, G., et al. \MariEMS
KQ Torque coecient train the trainee (MariTTT) courses on energy ecient
ship operation", Technical Report (2017).
KT Thrust coecient
2. Choi, Y.B. \2008.11 DSME ship energy saving devices:
Ni The total number of cells Hydrodynamics R&D team", In The House of Wisdom
Pi Instantaneous hydrostatic pressure and Innovation, Technical report (2008).
P Turbulence production rate 3. Skipsrevyen Company \The green solution below the
ReL Reynolds number based on the ship's waterline", https://www.skipsrevyen.no/article/the-
length green-solution-below-the-waterline (Accessed 15
VM Model speed February 2018).
VS Ship speed 4. Wartsila Encyclopedia of Marine Technology. Power-
Saving Devices, https://www.wartsila.com/encyclo-
eija Relative approximate error pedia/term/power- saving-devices (Accessed 14 Au-
eijext Extrapolated relative error gust 2019).
rij Grid renement factor 5. Carlton, J., Marine Propellers and Propulsion, 3th
Edn., Butterworth Heinemann, J.S. (2012).
ui Instantaneous velocity 6. Park, S., Oh, G., Rhee, S.H., et al. \Full scale
u0i Oscillating velocity wake prediction of an energy saving device by using
t Turbulent viscosity computational uid dynamics", Ocean Engineering,
 Average density 101, pp. 254{263 (2015).
k Prandtl number 7. Kim, J.H., Choi, J.E., Choi, B.J., et al. \Development
of energy-saving devices for a full slow-speed ship
" Turbulent Schmidt number through improving propulsion performance", Inter-
'ijext Extrapolated value national Journal of Naval Architecture and Ocean
Engineering, 7(2), pp. 390{398 (2015).

y First wall thickness of the boundary
layer 8. Mizzi, K., Demirel, Y.K., Banks, C., et al. \Design
optimisation of propeller boss cap ns for enhanced
 Fluid viscosity propeller performance", Applied Ocean Research, 62,
B Breath of KCS model pp. 210{222 (2017).
 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348 2347

9. Go, J.S., Yoon, H.S., and Jung, J.H. \Eects of a duct 23. Shamsi, R. and Ghassemi, H. \Time-accurate analysis
before a propeller on propulsion performance", Ocean of the viscous ow around puller podded drive using
Engineering, 136, pp. 54{66 (2017). sliding mesh method", Journal of Fluids Engineering,
10. Gaggero, S., Villa, D., Tani, G., et al. \Design 137(1), pp. 1{9 (2015).
of ducted propeller nozzles through a RANSE-based 24. Majdfar, S., Ghassemi, H., Forouzan, H., et al. \Hy-
optimization approach", Ocean Engineering, 145, pp. drodynamic prediction of the ducted propeller by CFD
444{463 (2017). solver", Journal of Marine Science and Technology,
11. Chang, X., Sun, S., Zhi, Y., et al. \Investigation of the 25(3) (2017).
eects of a fan-shaped Mewis duct before a propeller on 25. Esmailian, E., Ghassemi, H., and Zakerdoost, H. \Sys-
propulsion performance", Journal of Marine Science tematic probabilistic design methodology for simul-
and Technology, 24(1), pp. 46{59 (2019). taneously optimizing the ship hull-propeller system",
International Journal of Naval Architecture and Ocean
12. Andersen, S.V. \Numerical treatment of the design- Engineering, 9(3), pp. 246{255 (2017).
analysis problem of ship propellers using vortex lat-
tice methods", PhD Thesis, DTU, Lyngby, Denmark 26. Mahmoodi, K., Ghassemi, H., and Nowruzi, H. \Ob-
(1988). taining mathematical functions of the propeller thrust
and torque coecients uctuations at non-uniform
13. Baltazar, J., and Falcao de Campos, J.A.C. \On the wake ow including geometry eects", Mechanics &
modelling of the ow in ducted propellers with a Industry, 19(2), p. 205 (2018).
panel method", In Proceedings of the 1st International
Symposium on Marine Propulsors, 9, pp. 1{15 (2009). 27. Zakerdoost, H. and Ghassemi, H. \Hydrodynamic
optimization of ship's hull-propeller system under mul-
14. Cai, H.P., Su, Y.M., Li, X., et al. \Using the surface tiple operating conditions using MOEA/D", Journal
panel method to predict the steady performance of of Marine Science and Technology, 26(2), pp. 419{431
ducted propellers", Journal of Marine Science and (2020).
Application, 8(4), pp. 275{280 (2009).
28. Martinas, G. \Cavitation of a propeller and in uence
15. Baltazar, J., Falc~ao de Campos, J.A.C., and Boss- of a wake equalizing duct", TransNav: International
chers, J. \Open-water thrust and torque predictions Journal on Marine Navigation and Safety of Sea
of a ducted propeller system with a panel method", Transportation, 9(2), pp. 235{241 (2015).
International Journal of Rotating Machinery, 2012
(2012). 29. Ok, J.P. and Hamburg-Harburg, TU \Power savings
by wake equalizing ducts?", Ship Technology Research,
16. Lee, D., Jang, B.S., and Kim, H. \Development of 52(1), pp. 34{51 (2005).
procedure for structural safety assessment of energy 30. Korkut, E. \A case study for the eect of a ow
saving device subjected to nonlinear hydrodynamic improvement device (a partial wake equalizing duct)
load", Ocean Engineering, 116, pp. 165{183 (2016). on ship powering characteristics", Ocean Engineering,
17. Gaggero, S. \Design of PBCF energy saving devices us- 33(2), pp. 205{218 (2006).
ing optimization strategies: A step towards a complete 31. Celik, F. \A numerical study for eectiveness of a
viscous design approach", Ocean Engineering, 159, pp. wake equalizing duct", Ocean Engineering, 34(16), pp.
517{538 (2018). 2138{2145 (2007).
18. Nowruzi, H. and Naja, A. \An experimental and 32. Heinke, H.-J. and Hellwig-Rieck, K. \Investigation of
CFD study on the eects of dierent pre-swirl ducts scale eects on ships with a wake equilizing duct or
on propulsion performance of series 60 ship", Ocean with vortex generator ns", In Second International
Engineering, 173, pp. 491{509 (2019). Symposium on Marine Propulsors, Hamburg, Ger-
19. Tacar, Z., Sasaki, N., Atlar, M., et al. \An inves- many (2011).
tigation into eects of gate rudderr system on ship 33. Streckwall, H. and Xing-Kaeding, Y. \On the working
performance as a novel energy-saving and maneuvering principle of pre-swirl stators and on their application
device", Ocean Engineering, 218, p. 108250 (2020). benet and design targets", International Shipbuilding
20. Obwogi, E.O., Shen, H.L., and Su, Y.M. \The design Progress, 63(3{4), pp. 87{107 (2017).
and energy saving eect prediction of rudder-bulb-n 34. Lee, K.J., Lee, H.D., and Choi, S.H. \A design
device based on CFD and model test", Applied Ocean and validation study of a non-axisymmetric preswirl
Research, 114, p. 102814 (2021). stator", Ocean Engineering, 189, p. 106365 (2019).
21. Ghasemi, H. \The eect of wake ow and skew angle 35. Han, C. \Optimization and design research on marine
on the ship propeller performance", Scientia Iranica, pre-swirl stator-propeller by CFD", Master thesis,
16(2), pp. 149{158 (2009). Harbin Engineering University, Harbin, China (2019).
22. Ghassemi, H., Mardan, A., and Ardeshir, A. \Numer- 36. Lee, K.S., Kim, M.C., Shin, Y.J., et al. \Design of
ical analysis of hub eect on hydrodynamic perfor- asymmetric pre-swirl stator for KVLCC2 considering
mance of propellers with inclusion of PBCF to equal- angle of attack in non-uniform ow elds of the stern",
ize the induced velocity", Polish Maritime Research, Journal of the Society of Naval Architects of Korea,
19(2(73)), pp. 17{24 (2012). 56(4), pp. 352{360 (2019).
2348 S. Rezaei et al./Scientia Iranica, Transactions B: Mechanical Engineering 29 (2022) 2332{2348
View publication stats

37. Furcas, F., Vernengo, G., Villa, D., et al. \Design of 49. ITTC Proceedings \Practical guidelines for ship CFD
wake equalizing ducts using RANSE-based SBDO", applications ITTC - Recommended procedures and
Applied Ocean Research, 97, p. 102087 (2020). guidelines", Section 7.5-03-02-03, International Tow-
38. Nadery, A. and Ghassemi, H. \Numerical investigation ing Tank Conference (2014).
of the hydrodynamic performance of the propeller 50. Sezen, S., Dogrul, A., and Sakir, B.A.L. \An empirical
behind the ship with and without WED", Polish approach for propeller tip vortex cavitation noise",
Maritime Research (2020). Sigma Journal of Engineering and Natural Sciences,
39. Koushan, K., Krasilnikov, V., Nataletti, M., et al.
36(4), pp. 1127{1139 (2018).
\Experimental and numerical study of pre-swirl stators 51. Boumediene, K. and Belhenniche, S.E. \Numerical
PSS", Journal of Marine Science and Engineering, analysis of the turbulent ow around DTMB 4119 ma-
8(1), p. 47 (2020). rine propeller", Int. J. Mech. Aerosp. Ind. Mechatron.
Manuf. Eng., 10, pp. 347{351 (2016).
40. Nadery, A., Ghassemi, H., and Nowruzi, H. \Enhance-
ment of the ship propeller hydrodynamic performance 52. Kinaci, O.K., Gokce, M.K., Alkan, A.D., et al.
by dierent energy-saving devices mounted at the \On self-propulsion assessment of marine vehicles",
upstream zone", Journal of the Brazilian Society of Brodogradnja: Teorija i Praksa Brodogradnje i Po-
Mechanical Sciences and Engineering, 43(10), pp. 1{ morske Tehnike, 69(4), pp. 29{51 (2018).
25 (2021). 53. Workshop on verication and validation of ship
41. Nadery, A., Ghassemi, H., and Chybowski, L. \The maneuvering simulation methods (SIMMAN 2008),
eect of the PSS conguration on the hydrodynamic Copenhagen, Denmark, 14-16 April 2008, MO-
performance of the KP505 propeller behind the KCS", ERI Container Ship (KCS), http://www.simman2008.
Ocean Engineering, 234, p. 109310 (2021). dk/KCS/kcs geometry.htm (2012, accessed 10 March
2019).
42. Su, Y.M., Lin, J.F., Zhao, D.G., et al. \In uence
of a pre-swirl stator and rudder bulb system on the
propulsion performance of a large-scale ship model", Biography
Ocean Engineering, 218, p. 108189 (2020). Saman Rezaei received his MSs degree in Naval
43. Guo, C.Y., Xu, P., Han, Y., et al. \Working mechanism Architecture from Amirkabir University of Technology
of pre-swirl stator based on stereoscopic particle im- (Tehran Polytechnic). He is currently a PhD student of
age velocimetry", Ocean Engineering, 236, p. 109442 Maritime Engineering at the Department of Mechani-
(2021). cal Engineering at Sharif University of Technology. He
44. Qin, D., Huang, Q., Pan, G., et al. \Eect of the duct has researched on Marine Energy Device (ESD) and his
and the pre-swirl stator on the wake dynamics of a pre- current research interests include Marine Autonomous
swirl pumpjet propulsor", Ocean Engineering, 237, p. Surface Ship (MASS) collison avoidance methods base
109620 (2021). on COLREG.
45. Mikkelsen, H., Shao, Y., and Walther, J.H. \Numerical
study of nominal wake elds of a container ship in Milad Bamdadinejad received the BSc degree in
oblique regular waves", Applied Ocean Research, p. Civil-Surveying Engineering from the Department of
102968 (2022). Geodesy and Geomatics Engineering at Tafresh Univer-
46. Carrica, P.M., Fu, H., and Stern, F. \Computations sity in 2014 and the MSc degree in Coastal Engineer-
of self-propulsion free to sink and trim and of motions ing from the Department of Maritime Engineering at
in head waves of the KRISO Container Ship (KCS) Amirkabir University of Technology (Tehran Polytech-
model", Applied Ocean Research, 33(4), pp. 309{320 nic) in 2018. His research interests include application
(2011). of remote sensing and GIS in coastal engineering, ma-
47. Feng, D., Ye, B., Zhang, Z., et al. \Numerical sim- rine renewable energy, and coastal sediment transport
ulation of the ship resistance of KCS in dierent modeling.
water depths for model-scale and full-scale", Journal of
Marine Science and Engineering, 8(10), p. 745 (2020). Hassan Ghassemi is a Professor at Amirkabir Uni-
48. Celik, I.B., Ghia, U., Roache, P.J., et al. \Procedure versity of Technology (AUT). His research interests in-
for estimation and reporting of uncertainty due to clude ocean renewable energy, marine propulsor design,
discretization in CFD applications", Journal of Fluids dynamics of oshore structures, high-speed craft, and
Engineering-Transactions of the ASME, 130(1), pp. hydrodynamic numerical methods such as BEM and
1{4 (2008). CFD.