Ocean Engineering 180 (2019) 211–222

Contents lists available at ScienceDirect

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

Analysis of waterjet-hull interaction and its impact on the propulsion T

performance of a four-waterjet-propelled ship
Jie Gonga,b, Chun-yu Guoa,∗, Chao Wanga, Tie-cheng Wua, Ke-wei Songa
a
College of Shipbuilding Engineering, Harbin Engineering University, Harbin, 150001, China
b
Department of Mechanical Engineering, National University of Singapore, Singapore, 117575, Singapore

A R T I C LE I N FO A B S T R A C T

Keywords: Waterjet-hull interaction has different effects on the propulsion performance of the outer and inner waterjets in
Waterjet-hull interaction ships propelled by four waterjets. The dead rise angle of the hull lines near the stern leads to different inflow
Propulsion characteristics in the capture areas of individual waterjets. To better understand these effects, self-propulsion
Thrust deduction tests were performed to investigate the thrust deduction fraction and energy/momentum velocity coefficients,
Flow filed characteristics
and numerical simulations of unsteady multiphase flows were conducted using a discretised impeller/stator with
dynamic overset grids. Results obtained indicate that the flow rates and gross thrust of inner waterjets were
greater than those of the outer waterjets under the same rotational speeds, and that these differences gradually
decreased with increases in the advanced speed. The hull had a stronger negative effect on the inflow of the outer
waterjet because its energy/momentum velocity coefficients were smaller than those of the inner waterjets.
Additionally, flow field analysis of the capture area verified the quality of the computations, and internal flow
field analysis provided additional validation of observed results. It was found that the inlet-duct design and the
presence of stabilizer fins caused different interaction behaviours, resulting in the propulsion performance
discrepancies observed in four-waterjet systems.

1. Introduction research and recommendations were adopted by the ITTC, and con-
tinuous improvements have been added to perfect the procedures. Coop
Waterjets provide higher propulsive efficiency, less vibration and (1995) investigated the effects of interaction between a waterjet fea-
better manoeuvrability compared to conventional propellers, and are turing a flush-type intake and a planing hull in both model-scale and
thus extensively applied in high-speed ships in operating at speeds full-scale. The thrust deduction characteristics and interaction forces
excess of 30 knots (Molland et al., 2017). With the increasing tonnage were measured and reported, while an analytical propulsion model was
and size of high-speed ships, waterjet propulsion must provide higher constructed to incorporate the waterjet momentum forces. Moreover,
output power to meet speed demands. To this end, ships utilizing wa- the mechanisms possibly contributing to the overall interaction effect
terjet propulsion systems generally adopt multiple waterjets (two to were discussed. Van Terwisga (1996) developed a complete study on
four) to provide sufficient thrust at a high efficiency in limited in- waterjet-hull interaction in his doctoral thesis, and utilised analytical,
stallation spaces. Consequently, the resulting interaction between numerical, and experimental methods. He explained the difference
multiple-waterjet propulsion and hull systems becomes exceedingly between gross thrust and net thrust and specified the negative thrust
complicated, and its complete understanding requires in-depth in- deduction fraction of a waterjet-propelled ship. Bulten (2006) con-
vestigations to be performed. ducted a waterjet test in a cavitation tunnel and performed detailed
The International Towing Tank Conference (ITTC) set up the investigations both experimentally and numerically. He measured the
Waterjet Specialist Committee in 1993, and more recently has sum- velocity distribution of an impeller plane with a three-hole Pitot tube
marised several recommended procedures for marine vehicles with and revealed non-uniform characteristics of the inflow pattern. Rispin
waterjets (ITTC 25th, 2008; ITTC 26th, 2011). Waterjet-hull interaction (2007) studied the effect of model scale on the boundary layer thickness
is the core concern of scholars focused on self-propulsion analysis of a demi-hull with two waterjets. He found that the velocity dis-
(Coop, 1995; Van Terwisga, 1996; Eslamdoost, 2014). Some of their tributions of the capture area and the thrust deduction fraction were

∗
Corresponding author.
E-mail addresses: gongjie13@hrbeu.edu.cn (J. Gong), guochunyu_heu@outlook.com (C.-y. Guo), wangchao0104@hrbeu.edu.cn (C. Wang),
wutiechengship@126.com (T.-c. Wu), songkewei1125@126.com (K.-w. Song).

https://doi.org/10.1016/j.oceaneng.2019.04.002
Received 19 July 2018; Received in revised form 23 February 2019; Accepted 1 April 2019
Available online 12 April 2019
0029-8018/ © 2019 Elsevier Ltd. All rights reserved.
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

influenced by the scale. Jessup et al. (2008) conducted model tests and
a detailed power analysis of a Joint High-Speed Sealift (JHSS), which
was propelled by four axial-flow waterjets. The experiments included
detailed Laser Doppler Velocimetry (LDV) measurements at different
stations. LDV provided a superior presentation detailing the flow non-
uniformities at each station. Gong et al. (2017) applied Particle Image
Velocimetry (PIV) techniques to measure the velocity distribution of the
Fig. 1. Geometric model in the self-propulsion test.
capture area in the mid-body of a trimaran with a pair of waterjet units.
They pointed out that the motion of the ship during the PIV test affected
the measured results of the capture area. Their research provided some Table 1
guidance for employing PIV applications in waterjet-propelled vessels. Main parameters of the geometric model.
Eslamdoost et al. (2018) studied the thrust deduction fraction of a Parameter Value
waterjet-propelled hull in a wide range of speeds, and clarified the
reasons for the large variation in the fraction. They concluded that the 1. Ship model
Scale factor λ 16
jet system thrust deduction fraction was the main reason for the ne- Length between perpendiculars Lpp (m) 6.670
gative thrust deduction. Length at designed water line Lwl (m) 6.670
Recent progress in Computational Fluid Dynamics (CFD) techniques Breadth (m) 0.758
has enabled faster and more effective approaches for predicting wa- Draught (m) 0.240
Displacement Volume ∇(m3) 0.549
terjet–hull interactions. Takai (2010) investigated the flow fields for a
Wetted surface Sw (m2) 5.453
bare hull and the self-propelled conditions of the JHSS, employing the 2. Waterjet model
unsteady Reynolds-Averaged Navier-Stokes (URANS) solver CFDSHIP- Number of waterjet 4
IOWA. Detailed CFD verification and validation analysis of the JHSS Number of impeller blades 6
were performed by Takai et al. (2011). Kandasamy et al. (2010) de- Number of stator blades 11
Nominal diameter of channel (m) 0.083
veloped an integral force/moment waterjet model and applied it by Diameter of impeller Di (m) 0.101
using CFDSHIP-IOWA to analyse the local flow and powering of ships. Diameter of nozzle (m) 0.060
Eslamdoost et al. (2014) presented a pressure jump method for mod- Width of inlet (m × m) 0.095
elling waterjet-hull interaction based on potential flow theory, and
identified the major parameters which contributed to waterjet-hull in-
teraction (Eslamdoost, 2014). Additionally, they investigated the dif- four waterjets. A medium-speed mono-hull ship with two sets of sta-
ference between the net thrust and gross thrust of a waterjet unit bilizer fins was chosen for the experiments. A deep-V hull form with a
through a RANS study (Eslamdoost et al., 2016) and summarised the transom stern was adopted in the ship design, and four flush-type wa-
mechanism of thrust deduction in a more recent study (Eslamdoost terjets were selected for the propulsion system (Fig. 1). To indicate the
et al., 2018). different positions of the four mixed-flow waterjets, they were defined
Many previous studies have contributed to the understanding of as inner waterjets (IW) and outer waterjets (OW) (see Fig. 1(c)). The
waterjet-hull interactions via experimental and numerical analyses. main parameters of the ship and waterjet models are presented in
However, the hydrodynamic performance of individual waterjets may Table 1.
be different since a dedicated inlet-duct design must be performed to In the self-propulsion test, the impeller rpm range could fully cover
facilitate accurate alignment of a ship's operational profile with the the full-scale performance across the model speed range (Fr = 0.167 to
optimum point of waterjet operation, especially in the case of ships 0.534, Re = 2.01 × 106 to 3.79 × 106) to achieve the self-propulsion
characterised by varying hull lines near the locations of waterjet in- point (SPP). Four waterjets maintain the same rotational speed during
stallation. The dead rise angle of the hull lines near the stern leads to the experiment. As the most critical quantity, the volume flow rate Qj
different inflow characteristics in the capture areas of different water- through the waterjet system was determined indirectly using differ-
jets. Also, initial immersed depth of the waterjets influences the effi- ential pressure transducers (DPTs). Two micro pulsating pressure
ciency and output power of the waterjets mounted on the hull. transducers (PT1 and PT2) were installed in different longitudinal po-
Waterjet-hull interaction has different effects on the propulsion per- sitions as a group, both perpendicular to the nozzle slope and 30° from
formance of individual waterjets in a multiple-waterjet system. Thus, the vertical direction (see Fig. 2(a)). The precision level of the pressure
the goal of the present study is to investigate how waterjet-hull inter- transducer was 0.25% and the probe diameter was 5.0 mm. Two groups
action affects the propulsion performance of a self-propelled ship with of transducers, denoted as P1 through P4, were mounted on the IWs and
four waterjets. The inflow characteristics of the capture area and the OWs in order to measure the volume flow rate (Fig. 2 (b)).
gross thrust are analysed to quantitatively verify this complex interac- Calibration of the DPTs method was performed prior to com-
tion. mencement of the self-propulsion test and described in subsequent
This remainder of this study is organised as follows. The model tests Section 3.1. To confirm the repeatability and reliability of the DPTs
and CFD setup are described in Section 2. Section 3 describes a com- methods, the precision limit P(M) of the measured flow rates measured
parison of the experimental and numerical results, and contains a de- during multiple tests was analysed using the relation P(M) = KSdev/
tailed analysis comparing the inner waterjets and outer waterjets. Fi- (M1/2) (ITTC, 2011), where M denotes the number of runs, Sdev denotes
nally, conclusions are presented in Section 4. the sample standard deviation, and K is called coverage factor, the
value of which approximately equals 2. At a forward speed corre-
2. Model test and CFD setup sponding to Fr = 0.4 during the three repeated tests, the precision
limits of the flow rates were observed to be 0.37% and 0.42% for the
2.1. Design of self-propulsion test inner and outer waterjets, respectively.
A vehicle-mounted three-dimensional underwater stereo PIV device
Self-propulsion tests were conducted in a large 108 × 7 × 3.5 m was used to measure the inlet velocity distribution (for Fr = 0.3, 0.4,
towing tank at Harbin Engineering University. The four-component and 0.534) for the waterjet-propelled ship model (Gong et al., 2019).
motion-measuring device and the underwater particle image veloci- The coverage area of the PIV camera was rectangular, with a size of
metry (PIV) devices were used for data collection. The experiments approximately 260 × 320 mm2 (as determined after the calibration
included resistance, self-propulsion, and PIV tests for a ship model with process). This was sufficient to cover the target capture area of the IW

212
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

Fig. 3. Dimensions of the computational domain and the refined mesh.

of the ship hull was used for bare hull simulations, and thus, only two of
the waterjets were included for self-propulsion simulations. With re-
spect to the computational coordinate system, a Cartesian Earth-fixed
Fig. 2. Design of the differential pressure setup (a) locations of pressure- coordinates O-XYZ were used and defined as follows: the positive di-
transducers installation (b) pressure transducers installed on two waterjets. rection of the X-axis indicated the direction from the bow to the stern of
the ship, the positive direction of the Y-axis indicated the direction from
and OW. The sampling frequency was 7.5 Hz in the PIV test, and a total mid-ship to the port side, and the positive direction of the Z-axis in-
of 250 particle image pairs were captured for each speed. The PIV tests dicated the direction from the bottom of the ship to its deck (see Fig. 1).
were carried out in the resistance test to measure the velocity dis- The dimensions of the computational domain (see Fig. 3) along the
tribution of the capture area, which was used to calculate the energy/ three directions were: 2.5Lpp ≦ X ≦ 4Lpp, 0.0 ≦ Y ≦ 2Lpp, and
momentum velocity coefficients (Ce and Cm). Additionally, the results −2.0Lpp ≦ Z ≦ 1.0Lpp. The boundary conditions of the computational
obtained from the PIV test were used to evaluate the accuracy of the domain were set as follows. A ‘symmetry’ boundary condition of y = 0
computational flow fields. was adopted. The inlet was the velocity inlet, as were the top, sides, and
bottom (Wang et al., 2015; Gong et al., 2018), which simulated infinite
2.2. Numerical model far-field boundary conditions. The outlet was the pressure outlet, the
mid-ship section was a symmetry plane, and the hull, channels, im-
CFD simulations were performed using the unsteady RANS flow pellers, shafts, stators, and nozzles were non-slip surfaces.
solver STAR-CCM+ 11.04. The multiphase (air-water) three-dimen- For the self-propelled simulations, the overset grid technique was
sional flows, including free-surface effects, were established to analyse used to simulate the rotation of the impeller, and coupled with the
the hydrodynamic performance in both the bare hull and self-propul- motion of ship model. The overset grid technique can help mitigate the
sion simulations. In terms of motion, two degrees of freedom (heave effects of the very small gap between the blade tips and wall boundary,
and pitch) and free surface effects were considered in order to maintain and it is an effective solution for calculating unsteady problems (see
consistency with the experimental conditions. In addition, the waterjet Fig. 4 (c)). An overset grid allows complex regions to be replaced with
components (including the shaft, impeller, stator, and channel) were multiple overlapping, body-conforming curvilinear grids. The overset
discretised in the numerical simulations to investigate differences be- grid divides a computed flow regime into several sub-regimes, thus
tween the complex internal flow fields of the IWs and OWs. enhancing the flexibility of grid generation. This allows component
grids to be added in a mutually independent manner. Therefore, grid
connectivity only has to be updated locally, and information is propa-
2.2.1. Governing equations and turbulence model gated through the embedding and overlapping of each regional grid,
The governing equations for the turbulent flow field around the while interpolations are used to couple information from the boundary
waterjet-hull system comprised the instantaneous conservation of mass flow fields of the overset grids (Koblitz et al., 2017). The overset grids
(continuity equation) and momentum (RANS) for an incompressible were widely used in previous self-propulsion simulations (Carrica et al.,
Newtonian fluid (Wang, 2004). An additional equation is needed to 2010; Castro et al., 2011; Shen et al., 2015). The overset grid approach
solve the unknown Reynolds stress. The turbulence model is a closure divides the computational domain into three subdomains (one static
equation that combines the fluctuating time averages. In this study, the region and two rotational regions) and the overset regions are coupled
Shear Stress Transport (SST) k − ω turbulence model is selected for all to the background region using the overset mesh interface.
simulations, because it combines the advantages of the standard k − ω It is necessary to consider the level of discretisation in the special
and standard k − ε models and exhibits good behaviour in calculating regions of the computational region during meshing. The Cartesian cut-
the separating flow (Menter, 1994). The governing equations and the cell method was used to generate the unstructured computational grids,
transport equations of turbulence model are discretised using the finite with a boundary layer that consisted of a prism layer mesh and trimmed
volume method with a second-order upwind scheme, while pressure- mesh. The cell size of the computational domain should transition in a
velocity coupling is facilitated through the SIMPLE algorithm. The time gradual manner to avoid numerical dissipations. The surface mesh of
terms are discretised using second-order implicit time discretisation. the waterjet-hull system is shown in Fig. 4. Three refinement volumes
were used around the hull with gradually changed isotropic sizes of 3%
2.2.2. Grid system and boundary conditions Lpp, 1.5% Lpp, and 0.75% Lpp. In order to improve the simulation
To save computational resources and improve efficiency, only half

213
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

self-propulsion point, and the time-step Δt for each rotational speed was
maintained at 2° per step. The different time-step's effect on the com-
puted forces was negligible as the maximum time-step value (Lpp/
100v0) had met the demand of CFL number both in bare hull and self-
propulsion simulations. Five inner iterations were used for the con-
vergence of the flow field equations within each time step. Pressure
equation convergence is reached when the residual imbalance of the
Poisson equation drops by six orders of magnitude. All other variables
are assumed converged when the residuals drop to 10−5.

2.2.3. Grid sensitivity verification

To assess the numerical uncertainty of the simulations, a suite of
meshes was generated to quantitatively test the grid sensitivity, where
Δt/Δmin was kept constant to ensure that the CFL number was also kept
constant for all grid schemes. The two-grid assessment procedure pro-
posed by Roache (1997) was used to evaluate grid sensitivity in these
cases. Grid convergence analysis was performed using the total re-
sistance coefficient (CR), dynamic sinkage (He) and trim (Tr) in the bare
hull cases, while the IW (QjI) and OW (QjO) flow rates were used in the
self-propulsion cases. Values of QjI and QjO are normalised by Q0, which
Fig. 4. Mesh details of the water-hull system (a) Volume mesh inside the pump corresponds to the maximum volume flow rate obtained from the inner
(b) Volume mesh and prism layers of the impeller (c) Volume mesh and prism waterjet for Fr = 0.5 under operation at SPP.
layers of the stator. The disposal of an unstructured mesh is described in Baek et al.
(2015); in a similar manner, the refinement ratio r0 was considered as a
function of the total number of cells to uniformly refine the grid. The
precision of the non-uniform flow field and capture the free surface
function is as follows:
more accurately, three refinement volumes were applied in the gas-li-
quid interface areas with finer sizes (0.375% Lpp, 0.187% Lpp, and Nfine
0.094% Lpp) in the Z direction. In addition, several refinement volumes r0 = ( )1/ d
Nmedium (1)
were set near the stern and trailing flow region to capture the flow
details. To maintain consistency throughout the simulations, the mesh where N denotes the total number of cells in the grid and d denotes the
size and mesh distribution around the hull were kept almost the same dimensionality of the problem (d = 3). The error of the fine grid is
during the self-propulsion and bare hull simulations. estimated as follows:
For the rotational regions in self-propulsion simulations, even finer
grid sizes were employed to discretise the blade surfaces of the impeller f2 − f1
and stator, with a minimum size of 0.5% Di. Further, the blade surfaces E= p
1 − r0 0 (2)
were refined along the feature lines of the leading and trailing edges.
With respect to the volume mesh, a refinement volume was employed in where f1 and f2 respectively denote the solutions on the fine and
the static region to better discretise the region inside the pump with an medium grids, r0 denotes the mesh-refinement factor (r0 = 1.2 ), and p0
isotropic size of 2% Di. Also, a cylindrical refinement volume was em- denotes the formal order of accuracy (p0 = 2). The uncertainty is
ployed in the static region (isotropic size of 1% Di) to match the overset evaluated as follows:
mesh size in the rotational region. With an isotropic size of 0.25% Di,
the tip clearance region (about 2 mm) between the channel and im- UN = Fs|E| (3)
peller blade tips was further discretised. In addition, the boundary layer
thickness δ of the hull was estimated according to the theory of flat- where FS denotes a safety factor that is assumed to equal 3 (Di Mascio
plate boundary layer (White, 1998). Ten prism layers were computed et al., 2014).
and meshed near the surfaces with a total thickness equals to the esti- Computational results obtained from the medium and fine meshes
mated δ (1.26%–1.37% Lpp). Finally, the total number of cells in the fine are listed in Table 2. Based on the values of medium and fine mesh
grid was 12.67 million, while the static and rotational sub-domains systems, Extr represented the predicted values of finer grid system
contained 9.95 million and 2.72 million cells, respectively. employing the same refinement ratio with linear extrapolation, and the
The Volume of Fluid (VOF) model was utilised to capture the free uncertainties UN were normalised by Extr. It is shown in Table 2 that the
surface (Hirt and Nichols, 1981). The High Resolution Interface Cap- UN values of the dynamic sinkage and trim are considered as slightly
turing (HRIC) scheme was used to capture the free surface in STAR- excessive (UN ≈ 9% and 12%) in the bare hull and self-propulsion
CCM+ (Böhm and Graf, 2014). The Dynamic Fluid Body Interaction cases; it is expected that this is associated with the attitude dynamic of
(DFBI) model was employed to simulate the motion of the ship the ship for Fr = 0.4. The results in Table 2 indicate that a considerable
(Ohmori, 1998). It was important that the time-step size satisfied the difference exists between the predicted flow rates for IW and OW. This
demands of the Courant–Friedrichs–Levy number CFL < 1. Thus, a difference is related to the different ingested flow conditions in station
time-step (Δt) of Lpp/100v0 was used in the bare hull simulations, which 1a: The dead rise angle of the hull near the stern (refer Section 3.3.1)
was consistent with the minimum grid spacing (Δmin). Note that v0 leads to non-homogeneous boundary layer distribution from inboard to
refers to the ship speed. Moreover, a two steps calculation strategy bilge area, on the other hand, the existence of the stabilizer fins induces
(Yang et al., 2014) was employed during the self-propulsion simula- local vortices and further alters the velocity distribution in the capture
tions—Step1: a larger time step Lpp/100v0 was used at the initial stage areas of IW and OW. However, when compared against the experi-
to simulate the free surface elevation and the rotational speed of the mental results described in Section 3.2, simulation predictions were
impeller was maintained zero; Step 2: a relatively slow acceleration was observed to be considerably reliable with a numerical uncertainty of
imparted to the impeller to speed up the convergence to achieve the UN < 6%. Thus, the fine grid system was adopted in the following
analysis.

214
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

Table 2
Grid convergence analysis.
Medium Fine E Extr. UN (%)a

Fr = 0.4, bare hull

CR 7.062 × 10−3 7.081 × 10−3 −4.318 × 10−5 7.100 × 10−3 1.82
He −1.287 × 10−2 −1.305 × 10−2 4.091 × 10−4 −1.323 × 10−2 9.28
Tr 0.113 0.111 4.545 × 10−3 0.109 12.51
Fr = 0.4, self-propulsion
He −1.395 × 10−2 −1.416 × 10−2 4.773 × 10−4 −1.437 × 10−2 9.96
Tr 8.020 × 10−2 7.881 × 10−2 3.159 × 10−3 7.742 × 10−2 12.24
QjIb 0.720 0.725 −1.136 × 10−2 0.730 4.67
QjOb 0.696 0.690 1.363 × 10−2 0.684 5.98

a
UN is %Extr.
b
QjI and QjO are %Q0.

3. Results and discussion Qjt = v1 A1 = v2 A2 (5)

On the other hand, Bernoulli equation is assumed to hold along the

3.1. Definitions of hydrodynamic performance of waterjet-hull system
centre streamline of the nozzle (White, 1998):
Recommended procedures and guidelines for marine vehicles with 1 2 1
P1 + ρv1 + ρgh1 = P2 + ρv22 + ρgh2
waterjets were specified by the ITTC (ITTC 25th, 2008). The control 2 2 (6)
volume approach was employed, in which eight stations (from station
where A1 and A2 are the areas of flow-passing surfaces in two positions,
1a to station 7, see Fig. 5) were defined and measured (pressure or
and v1 and v2 are the velocities of the two sections; the tube is hor-
velocity) in order to calculate the momentum flux and thrust. The dif-
izontal, so h1 = h2.
ferential pressure transducers (DPTs) method was introduced in the
Combining Equations (5) and (6), the theoretical flow rate Qjt is
recommended procedures and used in the present study when mea-
expressed as follows:
suring the flow rates.
The procedures in the waterjet propulsion test to confirm the self- 2(P1 − P2) A12 A22
Qjt =
propulsion point are the same as those in the self-propulsion test for ρ (A12 − A22 ) (7)
ships with conventional propellers (Eslamdoost et al., 2018). The skin
friction correction force Fd was calculated on the basis of the bare hull The flow rate measured using the DPTs method was calibrated prior
resistance results, while considering the differences between the full- to performing the propulsion test. The calibration process was per-
and model-scale frictional resistance coefficients. In accordance with formed in the bollard pull test by means of mass flow measurement
recommended procedures to be followed during the self-propulsion test through weighing. The flow discharged from the nozzle over a time
(Sheng and Liu, 2011), the tow force F, which is measured by means of interval ΔT was collected via a hose connected to a collecting tank
the resistance dynamometer, should locate in the range of 0–1.2Fd placed behind the model, and the mass ΔM of the collected water was
under different rotational speeds for one advance speed. Operation at measured. The volume flow rate Qjt obtained through use of the DPTs
the self-propulsion point (SPP) is realized when F = Fd via interpola- method was calculated using equation (7). Then, the calibration factor
tion. c0 could be defined as:

ρm Rs ΔM
c0 = ( )/ Qjt
Fd = Rm − ρΔT (8)
ρs λ3 (4)
Thus, the volume flow rate Qj can be expressed as:
where ρm and ρs respectively denote the density of water in the towing
tank and actual sea conditions, and λ is the scale factor. Rm and Rs, 2(P1 − P2) A12 A22
Qj = c0⋅
respectively, denote the total resistances of the model and the full scale ρ (A12 − A22 ) (9)
ship. Rs is obtained according to the 2014 ITTC Recommended Proce-
dure of the resistance test, (ITTC, 2014). According to the recommended procedures and guidelines of ITTC
As illustrated in Section 2.1, two micro pulsating pressure trans- (ITTC 24th, 2005), the gross thrust Tg can be calculated according to the
ducers were used as a group to measure the dynamic pressures under change in momentum flux from station 1a through station 6 (nozzle) in
SPP, calculating the flow rates of two waterjets. The nozzle corresponds the X direction. The gross thrust Tg is defined by:
to a contracted contour tube, and by measuring the average dynamic
pressures P1 and P2, an expression of theoretical flow rate Qjt can be
Tg = Mj − Mi = ρQj cm6 vx 6 − ∬ ρvx2 dA
A0 (10)
obtained on the basis of the continuity equation for incompressible
flow: where Mj and Mi represent the momentum flux from the nozzle and
station 1a, Qj denotes the flow rate through the waterjet system, A0
denotes the capture area, and vx denotes the axial velocity distribution
in A0. The parameter cm6 represents momentum correction factor de-
fined as:
1
c m6 =
A6
∬ ( vvxx66 )2dA
A6 (11)

where A6 denotes the nozzle area, vx6 denotes the local axial velocity
and vx6 denotes the area weighted average of vx6 over A6. The value of
cm6 is obtained from the CFD calculation as a reference.
Fig. 5. Definition of the eight stations (station 1a to station 7). Specifically, the thrust deduction fraction and energy/momentum

215
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

velocity coefficients are defined to quantitatively describe the waterjet- As stated earlier, two groups of micro pulsating pressure transducers
hull interaction. The thrust deduction fraction t of a waterjet-propelled were installed on the nozzle in order to measure flow rate Qj. In the self-
craft is calculated basing on the gross thrust, and t is defined as follows: propelled simulations, the pulsating pressures recorded by four probes
Rm − Fd in the corresponding positions were monitored to form a direct com-
t=1− parison. Fig. 8 illustrates the comparison between simulated and mea-
Tg (12)
sured pulsating pressures in the time domain from P1 to P4 (Fig. 2 (b)).
The energy velocity coefficient Ce and momentum velocity coeffi- They were carried out for Fr = 0.3, 0.4, and 0.534 (labelled from (a) to
cient Cm are used to represent the hull's influence on the inflow in the (c)), with the rotational speed at SPP. Table 3 shows a comparison of
capture area of the waterjet. Based on the axial velocity distribution in average pressures for P1 to P4 and the flow rates of IW and OW.
the capture area, Ce and Cm are calculated as follows: As shown in Fig. 8 (a), the CFD results were similar to the experi-
mental results from P1 to P4 for Fr = 0.3. Stronger numerical diffusion
∬ vx3 dA occurred during higher speed conditions, and the differences between
1 A0
Ce = 2 ⋅ the measured pressures were gradually increased. The average relative
v0 ∬ vx dA
A0 (13) error reached a maximum value of 14.88% in P4 for Fr = 0.534 (see
Table 3). Values of flow rates obtained via CFD were calculated by
∬ vx2 dA performing velocity integration within the nozzle to provide more ac-
1 A0 curate values. Except for the inner waterjet value for Fr = 0.5, most of
Cm = ⋅
v0 ∬ vx dA the relative errors of the numerical flow rates of IW and OW remained
A0 (14) under 4% when compared against experimental data, thereby demon-
where A0 and vx have been defined earlier in equation (10). strating an acceptable accuracy of the self-propulsion simulations.

3.2. Comparison of model test and CFD results 3.3. Discrepancies analysis of IW and OW

Firstly, the resistance, heave and pitch curves of the bare hull si- 3.3.1. Flow rates
mulations were compared with the experimental results as a pre- Volume flow rates corresponding to IW and OW are compared at the
liminary verification of the CFD simulations. Moreover, numerical SPP for varying ship speeds (Fr = 0.167 to 0.534). Fig. 9 shows the non-
pulsating pressures recorded by four probes at the nozzles were com- dimensional measured flow rates Qj/Q0 for IW and OW; the values were
pared directly with pressures recorded in the propulsion test, in order to normalised by dividing the maximum flow rate Q0 for Fr = 0.534 (IW).
verify the accuracy of the self-propelled simulations. The IW and OW flow rates increase gradually with increases in ad-
The total resistance (Rm), dynamic sinkage (He, measured at half vanced speed. It is noticeable that OW flow rates were consistently
Lpp), and trim (Tr) were monitored for all numerical cases across the lower than those of IW within the range of advanced speeds. The re-
speed range Fr = 0.2–0.534. The dynamic sinkage was defined as po- lative errors for all speeds varies from 5.44% to 13.28%. All waterjets
sitive when the hull moved upward, and the trim was defined as po- provide the demanded total thrust to balance the resistance of the ship.
sitive when the stern moved downward. As shown in Fig. 6, Rm ex- The differences illustrated above are considered to be related to the
hibited good agreement with the experimental data for all speeds, with matching of the waterjet-hull system and the installation locations of
an average difference of 1.98% D (D denotes experimental data) for the waterjets. The plan line of the hull was designed with a dead rise
lower speeds (0.2 ≤ Fr ≤ 0.334) and 3.89% D for medium speeds angle α near the stern (see Fig. 10). The position of the shaft of the outer
(0.334 < Fr ≤ 0.534). The dynamic sinkage and trim (Fig. 7) exhibited waterjet was lifted to fit the hull, resulting in a shaft height differenceΔh
the same trends with the experimental results both in bare hull (BH) between the IW and OW. Fig. 11 shows the non-dimensional axial ve-
and self-propulsion (SP) simulations. However, the average difference locity distributions (vx/v0) in station 1a without and with stabilizer fins.
for He and Tr in medium speeds (Fr > 0.4) were slightly larger than The dead rise angle α resulted in non-homogeneous boundary layer
those in lower speeds. Similar to the results of Takai et al. (2011), the distribution in station 1a (see Fig. 11(a), the influence of the stabilizer
CFD computations under-predicted the sinkage and over-predicted the fins was excluded). The boundary layer thickness gradually decreased
trim angle. Even so, the CFD simulations provided acceptable results from the inboard to the bilge area. The existence of stabilizer fins fur-
overall regarding the resistance of the ship, and correctly captured the ther increased the nonuniformity by inducing local vortices and thus
heave and pitch motion tendencies over the tested speed range. changed the velocity distribution (see Fig. 11(b), circle A and B) in
station 1a. On the other hand, the shaft height differenceΔh influenced
the head rise H directly; this was related to the flow rates. In addition,
the immersed depth of the inner waterjet was always greater than that
of the outer one, which had a direct effect on the power conversion
capability of the waterjet. According to the ‘momentum flux’ method
for waterjet-propelled ships, the flow rate determines the waterjet's
thrust directly. Therefore, it is necessary to conduct a deeper in-
vestigation into the difference in propulsion performance between the
IWs and OWs.

3.3.2. Gross thrust and thrust deduction

The gross thrust Tg was calculated according to the change in mo-
mentum flux through station 1a to the nozzle exit in the X direction, as
illustrated in Equation (8). The determination of the ingested mo-
mentum flux through station 1a was obtained by integration of the
velocity distribution in the capture area A0, which was obtained from
the PIV test. Using the velocity integral in station 1a and flow rate
measurement from station 6 (see Fig. 5), the geometric shape (semi-
ellipse) was confirmed when the flow rate of the capture area equals
Fig. 6. Comparison between CFD and EFD for total resistance (bare hull). that of the nozzle.

216
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

Fig. 7. Comparison between CFD and EFD results for dynamic sinkage (at 0.5Lpp) and trim.

Fig. 9. Non-dimensional measured flow rates of inner waterjets (IW) and outer
waterjets (OW).

Fig. 8. Comparison of simulated and measured pulsating pressures in time

domain for Fr = 0.3, 0.4, and 0.534.

IWs and OWs contributed differently amounts of thrust to the total

gross thrust under SPP. The Relative magnitudes of gross thrust (Tg-IW/
Tg-OW) contributed by IWs and OWs and the computational torque va-
lues (tor-IW/tor-OW) are depicted in Fig. 12. It is clear that IWs provided
Fig. 10. Stern view of the waterjet-hull system design.
greater thrust (by 18.4% (EFD) and 15.9% (CFD) on average) compared

Table 3
Comparison of average pressures values for P1 to P4 and flow rates of IW and OW.
Fr = 0.3 Fr = 0.4 Fr = 0.534

EFD CFD err. (%) EFD CFD err. (%) EFD CFD err. (%)

P1 [kpa] 7.168 6.791 −5.26 11.851 10.994 −7.23 22.610 19.600 −13.31
P2 [kpa] 3.506 3.270 −6.73 5.841 5.528 −5.36 12.369 10.984 −11.19
QjIa 0.567 0.555 −2.12 0.751 0.725 −3.46 1.000 0.929 −7.10
P3 [kpa] 6.871 6.763 −1.57 11.870 11.278 −4.99 21.760 19.512 −10.33
P4 [kpa] 4.144 3.790 −8.54 7.060 6.101 −13.6 13.724 11.682 −14.88
QjOa 0.493 0.512 3.85 0.673 0.690 2.53 0.904 0.881 −2.54

a
QjI and QjO are %Q0.

217
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

Fig. 11. Non-dimensional axial velocity distribution (vx/v0) in station 1a without and with stabilizer fins.

Fig. 12. Relative magnitudes of gross thrust and torque between IWs and OWs. Fig. 13. Thrust deduction fraction t in the speed range Fr = 0.167–0.534.

to OWs under identical input rotational speeds. In addition, a direct inner and outer waterjets. Station 1a is defined as the inflow capture
comparison of thrust values clarifies that the gross thrust of IWs is area and is located one pump diameter ahead of the inlet tangency,
significantly higher (7.2%–13.1%) than that of OWs at low advance while station 3 is defined as the inflow plane of the pump.
speeds (0.167 ≤ Fr ≤ 0.4), and it declines to 5.1%–7.4% for In accordance with ITTC guidelines (ITTC 24th, 2005), the inlet
0.4 < Fr ≤ 0.534. On the other hand, the observed torque difference is velocity distributions in station 1a were compared and analysed for the
rather small as the value of tor-IW/tor-OW remains approximately equal to bare hull cases. The PIV tests were conducted in the bare hull resistance
1 for all speeds. This indicates that the output power of IWs is larger tests, which provided experimental verification for the flow field ac-
compared to that of OWs under identical input power conditions at low curacy of the numerical simulations. Fig. 14 compares the PIV results
speeds, and that the difference between them gradually declines with and CFD computations in terms of the non-dimensional axial velocity
the increase in advanced speed. contours (vx/v0) in the capture area for Fr = 0.3, 0.4, and 0.534. The
The thrust deduction fraction data is calculated based on the bare coordinate axis of the velocity contours were normalised by Lpp, where
hull resistance, external tow force, and total gross thrust for y/Lpp = 0 corresponds to the central longitudinal section. The laminar
Fr = 0.167–0.534. Fig. 13 shows that the waterjets’ effect on the re- bottom data obtained from the PIV measurements was distorted and
sistance of the hull differed with the advanced speeds. Large positive was therefore replaced by grey bars (Gong et al., 2019). The velocity
values for t were observed in low speed ranges (Fr ≤ 0.3); t decreased distribution in the laminar layer was assumed to follow the one-se-
toward zero for 0.3 < Fr ≤ 0.4 then exhibited small negative values in venth-power law in the calculation of tested Ce and Cm (White, 1998).
the high-speed range 0.4 < Fr ≤ 0.534. The computed t values are The outlines of the hull at station 1a for the three speeds are different
lower than the measured ones over the speed range and maximum because of the heave and pitch motions of the ship model. The capture
difference between the two was observed when Fr = 0.534. The nega- area geometry of CFD cases was determined based on the ingested in-
tive t values were caused by the jet system thrust deduction, as reported flow, which was determined by backward tracing of the streamlines
in Eslamdoost et al. (2018). from the impeller surface in self-propulsion simulations. Fig. 15 shows
the capture areas in self-propulsion simulations for Fr = 0.4 case under
3.3.3. Flow fields SPP in station 1a. The vortex structures around the hull were visualized
To further analyse the hydrodynamic performance discrepancies by means of Q-criterion (Wu, 2013) in Fig. 16. It was observed that the
between IWs and OWs, the internal flow field characteristics of IWs and visualized capture area of OW was more narrow and higher than that of
OWs were compared to provide additional auxiliary information. IW, which reflected the influence of stabilizer fins.
Additionally, the flow around the hull was analysed to reveal the Both PIV and CFD results demonstrate that the effect of the induced
physical mechanism behind the waterjet-hull interaction behaviour. flow in the capture area are apparently different for IWs and OWs. CFD-
The following portion of the paper compares the velocity distributions based and experimentally obtained values of the non-dimensional
of station 1a (x/Lpp = −0.09) and station 3 (x/Lpp = −0.03) for the

218
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

Fig. 14. Comparison of velocity distributions in station 1a (x/Lpp = −0.09) between experimental (PIV) and CFD results (bare hull).

Fig. 17. Non-dimensional velocities at IW (y/Lpp = 0.012) and OW (y/

Lpp = 0.036) mid-sections.

Fig. 15. Capture areas of the ingested flow in station 1a for IW and OW.
capture area. Owing to the presence of small-scale vortices and noisy
velocities, the measured contour lines are not as smooth as the nu-
merical results. For the IWs, the turbulence boundary layers at station
1a provide relatively high accuracy, and the velocity distribution re-
flects a consistency between the experimental and numerical results.
However, the difference between the PIV and CFD results is dis-
tinguished by the increasing Reynolds number for the OWs. The gra-
dient of the OW velocity distribution is clearly larger than that of the
IWs, which indicates a higher inhomogeneity in the inflow.
Ce and Cm are summarised in Table 4, including the numerical re-
sults of bare hull cases (CFD) and the experimental results (PIV). The
predictions from the CFD results agree well with the PIV measured
Fig. 16. Tip vortices induced by stabilizer fins.
results, with only a slight quantitative difference. Both Ce and Cm of the
OWs are apparently smaller than those of the IWs, which indicates that
velocity at mid-sections of the capture area of IW (y/Lpp = 0.012) and the hull has a stronger negative effect on the OWs inflow. As mentioned
OW (y/Lpp = 0.036) are plotted in Fig. 17. The velocity demonstrates in section 2.1, two sets of stabilizer fins were mounted on the hull. The
increased stability along the depth direction (negative z) for IW while tip vortices generated by these fins convect downstream and alters the
values of the same slightly fluctuate in the bottom half of the OW velocity distribution of the capture area. The uniformity of the capture

219
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

Table 4 whereas a gradual velocity transition is observed in IW, thereby in-

Calculation of Ce and Cm in station 1a. dicating that IW velocity distribution demonstrates greater uniformity
Fr CFD PIV compared to that of the OW. However, the ingested flow of IW and OW
becomes fully developed under high-speed conditions, thereby causing
IW OW IW OW flow field differences to reduce as the advance speed increases from
Fr = 0.3 to Fr = 0.54. As described in Section 3.3.2, the gross thrust
0.3 Cm 0.895 0.864 0.884 0.852
Ce 0.802 0.748 0.784 0.727 discrepancies between IWs and OWs for Fr = 0.3, 0.4, and 0.534 were
0.4 Cm 0.896 0.866 0.891 0.858 gradually diminished. Therefore, the above analysis provided an aux-
Ce 0.804 0.751 0.795 0.737 iliary explanation for the flow rates and gross thrust results, and better
0.534 Cm 0.902 0.867 0.892 0.861 revealed the flow mechanism of a ship propelled by four waterjets.
Ce 0.815 0.752 0.798 0.743
The experimental and numerical jet streams and wake elevations
are compared in Fig. 20. Because the geometric model and test condi-
areas is, therefore, affected as the stabilizer fin vortex passed through tions were completely consistent, and finer girds were distributed in the
the capture area of the outer waterjet, and smaller Ce and Cm values are nozzle and wake regions, the numerical jet stream morphology and
obtained for OW in comparison to IW. Based on this consideration, wake elevations were very close to those in the self-propulsion test. The
trends observed in Fig. 17 can also be explained. Although there exist jet stream travelled a long distance before hitting the transom wake in
certain differences with regards to flow details in certain areas, values higher speeds. Strong interference was observed between the jet
of the energy/momentum velocity coefficients obtained in the capture streams and the tailing flow from a transom stern. Owing to limitations
area were observed to be consistent between the CFD simulations and with regard to grid resolution, it was difficult for the numerical simu-
PIV tests. lations to accurately capture the finer flow phenomenon manifested
For the perspective of the internal flow field, Fig. 18 shows the during model test (e.g. jet diffusion and the splash zone).
detailed flow characteristics at several longitudinal and transverse
sections inside the inner waterjet for the Fr = 0.4 case under SPP. The 4. Conclusion
ingested flow gets discharged by the nozzle at high speeds as it absorbs
the power delivered by the impeller (see Fig. 18(a)). The distribution of Using a combined experimental and numerical approach, this study
vx/v0 between each blade phase remains consistent across the impeller analysed how waterjet-hull interaction affects the propulsion perfor-
(Fig. 18(b)) and stator plane (Fig. 18(c)). It is noted that the low speed mance of a self-propelled ship with four waterjets. The thrust deduction
region located in the middle of the stator and nozzle planes (Fig. 18(d)) fraction as well as energy and momentum velocity coefficients were
after the stator rectification. determined to quantitatively describe the waterjet-hull interaction.
Additionally, Fig. 19 shows the calculated non-dimensional axial Self-propulsion tests were carried out to investigate the flow rates and
velocity (vx/v0) of IWs and OWs at station 3 for cases corresponding to gross thrust values corresponding to inner and outer waterjets. Further,
Fr = 0.3, 0.4, and 0.534. The non-dimensional velocity distributions of numerical simulations were conducted for bare hull and self-propulsion
IWs and OWs both showed consistency in terms of inflow character- conditions using a discretised impeller/stator with dynamic overset
istics. The shape and location of the high-velocity flow distributed near grids. Major conclusions drawn from this study are listed as follows:
the bottom dead centre (BDC) and the velocity distribution in the lower
half disk were greater than those of the upper half disk. Owing to the (1) OW flow rates consistently measured lower compared to those of
shaft retardation, the axial velocity gradually decreased from the cir- IWs within the considered range of advance speeds, albeit the said
cumference to the shaft surface. Additionally, there exist some differ- difference gradually declined when the inflow was fully developed
ences between local velocity distributions for IWs and OWs in that of under high-speed operation;
the high- and low-velocity regions of OW remain most concentrated, (2) Observed differences between thrusts obtained from IWs and OWs

Fig. 18. Non-dimensional axial velocity contours (vx/v0) for internal flow fields of IW at Fr = 0.4.

220
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

Fig. 19. Comparison of the instantaneous axial velocity distributions at station 3 (x/Lpp = −0.03) for IWs and OWs.

demonstrated a direct impact on the total gross thrust under SPP as the influence of the stabilizer fins lead to manifestation of dif-
conditions; IWs provided greater thrust (on average 18.4% (EFD) ferent interaction behaviours, thereby resulting in discrepancies in
and 15.9% (CFD)) compared to OWs under identical input rota- propulsion performance of a four-waterjet-propelled system. A clear
tional speeds. The thrust deduction fraction demonstrated small understanding of the above-mentioned propulsion performance
negative values (close to −0.05) under high-speed conditions; differences is a must to take full advantage of the same in en-
(3) Consistency was observed in terms of energy and momentum ve- gineering practice.
locity coefficient values between numerical and PIV-measured re-
sults, which in turn, validated the flow field accuracy of the RANS Because our research was conducted and analysed based on ex-
computations. Additionally, the effect of the stabilizer fin vortex on perimental and numerical results obtained for a model scale, sub-
OW inflow was verified by the fact that its Ce and Cm values were sequent studies would use these findings to investigate the scale effect
smaller than those of IW; of the waterjet-hull interaction on the propulsion performance.
(4) Difference in inlet-duct design of the waterjet-hull system, as well

221
J. Gong, et al. Ocean Engineering 180 (2019) 211–222

Fig. 20. Comparison between the experimental and numerical jet stream and wake elevations for Fr = 0.3, 0.4, and 0.534.

Acknowledgments fins. Ocean Eng. 171, 120–130.

Hirt, C.W., Nichols, B.D., 1981. Volume of fluid (VOF) method for the dynamics of free
boundaries. J. Comput. Phys. 39 (81), 201–225.
This work was in response to the research project titled ‘‘High ITTC 24th, 2005. Report of the specialist committee on validation of waterjet test pro-
Technology Ship Scientific Research Project’’ performed by Ministry of cedures. In: Proceedings of the 24th International Towing Tank Conference.
Industry and Information Technology [MIIT2016, 26] (China) and the Edinburgh, UK.
ITTC 25th, 2008. The specialist committee on waterjets: final report and recommenda-
“Foundation of Key Laboratory of Defense Science and Technology for tions to the 25th ITTC. In: Proceedings of 25th ITTC. Fokuoka, Japan.
Equipment Pre-Research (China) (Grant No. 61422230203182223010). ITTC 26th, 2011. The propulsion committee: final report and recommendations to the
The authors would like to acknowledge the help and support by Prof. 26th ITTC. In: Proceedings of 26th ITTC. Rio de Janeiro, Brazil.
ITTC, 2011. Recommended Procedures and Guidelines, Uncertainty Analysis - Example
Nhan Phan-Thien, Prof. Boo Cheong Khoo and Dr. Wang Di from
for Waterjet Propulsion Test (7.5-02-05-03.3).
National University of Singapore (NUS). ITTC, 2014. Recommended Procedures and Guidelines, Resistance Test (7.5-02-02-01).
Funding: This work was supported by the National Natural Science Jessup, S., Donnelly, M., Fry, D., Wilson, M., 2008. Performance analysis of a four-wa-
terjet propulsion system for large sealift ship. In: 27th Symposium on Naval
Foundation of China (Grant Numbers 51679052, 51639004, and
Hydrodynamics, Seoul, Korea.
51709060). Kandasamy, M., Ooi, S.K., Carrica, P., Stern, F., 2010. Integral force/moment waterjet
model for CFD simulations. J. Fluids Eng. 132 (10) 101103(1–9).
References Koblitz, A.R., Lovett, S., Nikiforakis, N., et al., 2017. Direct numerical simulation of
particulate flows with an overset grid method. J. Comput. Phys. 343, 414–431.
Menter, F.R., 1994. Two-equation eddy-viscosity turbulence models for engineering ap-
Baek, D.G., Yoon, H.S., Jung, J.H., Kim, K.S., Paik, B.G., 2015. Effects of the advance ratio plications. AIAA J. 32 (8), 1598–1605.
on the evolution of a propeller wake. Comput. Fluids 118, 32–43. Molland, A.F., Turnock, S.R., Hudson, D.A., 2017. Ship Resistance and Propulsion, second
Böhm, C., Graf, K., 2014. Advancements in free surface RANSE simulations for sailing ed. Cambridge University Press, Cambridge.
yacht applications. Ocean Eng. 90, 11–20. Ohmori, T., 1998. Finite-volume simulation of flows around a ship in maneuving motion.
Bulten, N.W.H., 2006. Numerical Analysis of a Waterjet Propulsion System, Doctoral J. Mar. Sci. Tchnol. 3, 82–93.
Thesis. Eindhoven University of Technology, Eindhoven, Netherlands. Rispin, P., 2007. Waterjet Self-Propulsion Model Test for Application to a High-Speed
Carrica, P.M., Castro, A.M., Stern, F., 2010. Self-propulsion computations using a speed Sealift Ship. 6300 State University Drive, Long Beach, CA 90815 Suite 220.
controller and a discretized propeller with dynamic overset grids. J. Mar. Sci. Roache, P.J., 1997. Quantification of uncertainty in computational fluid dynamics. Annu.
Technol. 15, 316–330. Rev. Fluid Mech. 29, 123–160.
Castro, A.M., Carrica, P.M., Stern, F., 2011. Full scale self-propulsion computations using Shen, Z.R., Wan, D.C., Carrica, P.M., 2015. Dynamic overset grids in OpenFOAM with
discretized propeller for the KRISO container ship KCS. Comput. Fluids 51 (1), 35–47. application to KCS self-propulsion and maneuvering. Ocean Eng. 108, 287–306.
Coop, H.G., 1995. Investigation of Hull-Waterjet Interaction Effects. Doctoral Thesis. Sheng, Z.B., Liu, Y.Z., 2011. Ship Principles. Shanghai Jiaotong University Press,
University of Canterbury, Christchurch, New Zealand. Shanghai.
Di Mascio, A., Muscari, R., Dubbioso, G., 2014. On the wake dynamics of a propeller Takai, T., 2010. Simulation Based Design for High-Speed SeaLift with Waterjets by High
operating in drift. J. Fluid Mech. 754, 263–307. Fidelity URANS Approach. Master's Thesis. The University of Iowa, Iowa, USA.
Eslamdoost, A., 2014. The Hydrodynamics of Waterjet/Hull Interaction. Doctoral. Takai, T., Kandasamy, M., Stern, F., 2011. Verification and validation study of URANS
Chalmers University of Technology, Gothenburg, Sweden Shipping and Marine simulations for an axial waterjet propelled large high-speed ship. J. Mar. Sci. Technol.
Technology. 16 (4), 434–447.
Eslamdoost, A., Larsson, L., Bensow, R., 2014. A pressure jump method for modelling Van Terwisga, T., 1996. Waterjet-Hull Interaction, Doctoral Thesis. Delft University of
waterjet/hull interaction. Ocean Eng. 88, 120–130. Technology, Delft, Netherlands.
Eslamdoost, A., Larsson, L., Bensow, R., 2016. Net and gross thrust in waterjet propulsion. Wang, F.J., 2004. Computational Fluid Dynamic Analysis – CFD Principles and
J. Ship Res. 60 (2), 78–91. Application. Tsinghua University Press, Beijing.
Eslamdoost, A., Larsson, L., Bensow, R., 2018. Analysis of the thrust deduction in waterjet Wang, Z.Z., Xiong, Y., Wang, R., Shen, X.R., Zhong, C.H., 2015. Numerical study on scale
propulsion – the Froude number dependence. Ocean Eng. 152, 100–112. effect of nominal wake of single screw ship. Ocean Eng. 104, 437–451.
Gong, J., Guo, C.Y., Wu, T.C., Zhao, D.G., 2017. Particle image velocimetry measurement White, F.M., 1998. Fluid Mechanics, fourth ed. McGraw-Hill Higher Education, New York.
of velocity distribution at inlet duct of waterjet self-propelled ship model. J. Wu, P.C., 2013. A CFD Study on Added Resistance, Motions and Phase Averaged Wake
Hydrodyn. 29 (5), 879–893. Fields of Full Form Ship Model in Head Waves. Doctoral Thesis. Osaka University,
Gong, J., Guo, C.Y., Wu, T.C., Song, K.W., 2018. Numerical study on the unsteady hy- Osaka, Japan.
drodynamic performance of a waterjet impeller. J. Coast. Res. 34 (1), 151–163. Yang, Q.F., Wang, Y.S., Zhang, Z.H., 2014. Numerical prediction of the fluctuating noise
Gong, J., Guo, C.Y., Song, K.W., Wu, T.C., 2019. SPIV measurements and URANS simu- source of waterjet in full scale. J. Mar. Sci. Technol. 19 (4), 510–527.
lations on the inlet velocity distribution for a waterjet-propelled ship with stabiliser

222