Vous êtes sur la page 1sur 22

Ocean Engineering 137 (2017) 287308

Contents lists available at ScienceDirect

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

An integrated empirical manoeuvring model for inland vessels MARK


a,b, a c a d
Jialun Liu , Robert Hekkenberg , Frans Quadvlieg , Hans Hopman , Bingqian Zhao
a
Delft University of Technology, Mekelweg 2, 2628 CD Delft, The Netherlands
b
National Engineering Research Center for Water Transport Safety, Wuhan University of Technology, 430063 Wuhan, Hubei, China
c
Maritime Research Institute Netherlands (MARIN), Haagsteeg 2, 6700 AA Wageningen, The Netherlands
d
Wuhan Rules and Research Institute, China Classication Society, 430022 Wuhan, Hubei, China

A R T I C L E I N F O A BS T RAC T

Keywords: Ship manoeuvrability is important for navigation safety. However, few studies have been specically carried out
Inland vessel manoeuvrability for inland vessels. Since most of the empirical methods were generated based on databases of seagoing ships,
Twin-propeller twin-rudder systems the usability of these methods for inland vessels is doubtful. The objective of the present work is to assess the
Rudder forces and moments existing manoeuvrability models and nd the most suitable ones for inland vessels. Furthermore, these models
Manoeuvrability prediction
are integrated into a single new model that can predict the manoeuvring behaviour of benchmark inland vessels
Manoeuvring simulations
without extensive experimental tests. The method aims at inland ships of typical dimensions in the Yangtze
River (inland ships on European waterways have dierent dimensions), which is characterised by a large water
depth. After preselecting the most promising methods through reviewing literature, a selection of the empirical
methods for hull forces and moments is performed by comparing simulation results to model-scale free-running
experiments of various turning and zigzag manoeuvres. Considering the large variety of rudder congurations
for inland vessels, this paper describes a procedure of using 2D open-water RANS results to calculate the rudder
forces and moments. Accordingly, hydrodynamic coecients of benchmark rudder proles are provided to
apply the proposed procedure for dierent rudder congurations.

1. Introduction general, the TPTR system is preferred in shallow water due to the
possibility to have restricted draught. To analyse the manoeuvrability
An inland vessel, in this paper, is a self-propelled motor ship that of TPTR inland vessels, the mathematical model needs to consider the
sails in inland waterways, such as rivers, canals, and lakes. The design particulars of the inland vessels and the twin-rudder congurations.
and navigation environment of inland vessels are dierent from those The objective of the present work is to formulate an integrated model,
of seagoing ships. The impacts of these dierences should be carefully which is based on available knowledge and easy to use at the initial
considered when prediction methods that are based on or intended for design stage without requiring extensive model tests.
seagoing ships are used for inland vessels (Liu et al., 2015). Unlike Besides studies on the mathematical modelling of MPMR ships
seagoing ships that commonly equip a single propeller and a single (Yoshimura and Sakurai, 1989; Lee and Fujino, 2003; Hasegawa et al.,
rudder (SPSR), inland vessels commonly feature multiple propellers 2006; Khanr et al., 2008; Di Mascio et al., 2011), several attempts have
and multiple rudders (MPMR). Typical MPMR congurations of inland been made to model the interaction between the hull and the rudder
vessels are single-propeller twin-rudder (SPTR), twin-propeller twin- (Khanr et al., 2011), the interaction between the propeller and the
rudder (TPTR), and twin-propeller quadruple-rudder (TPQR). This rudder (Nakatake et al., 1989), and the ow straightening eect of the hull
paper focuses on manoeuvring modelling of the TPTR system for and the propeller on the rudder (Molland and Turnock, 2002).
inland vessels as it is widely used on both the Rhine and the Yangtze Additionally, a number of studies have been published on the asymmetric
River. manoeuvring behaviour of MPMR ships (Kang et al., 2011; Coraddu et al.,
Normally, TPTR ships have better course keeping and course 2013; Dubbioso et al., 2015). However, the above-mentioned studies were
changing abilities but worse turning abilities than SPSR ships (Kim carried out for seagoing ships, very few manoeuvrability studies have been
et al., 2007). Yoshimura and Sakurai (1989) showed that a wide-beam performed for inland vessels. This paper presents studies based on two
TPTR ship may have an improved turning ability in shallow water standard TPTR inland vessels in the Yangtze River. One is a 6700 t TPTR
instead of a customarily worsened one for conventional SPSR ships. In bulk carrier, and the other one is a 3500 t TPTR tanker.


Corresponding author.
E-mail address: jialunliu@whut.edu.cn (J. Liu).

http://dx.doi.org/10.1016/j.oceaneng.2017.04.008
Received 13 September 2016; Received in revised form 30 December 2016; Accepted 4 April 2017
0029-8018/ 2017 Elsevier Ltd. All rights reserved.
J. Liu et al. Ocean Engineering 137 (2017) 287308

Nomenclature FR Rudder resultant force (N)


FT Rudder tangential force (N)
Abbreviations FX Longitudinal component of rudder induced forces (N)
FY Lateral component of rudder induced forces (N)
COG Centre of gravity Iz Moment of inertial (km2)
MPMR Multiple-propeller multiple-rudder Jz Added moment of inertial (km2)
SPSR Single-propeller single-rudder k Impact factor of the rudder aspect ratio on the rudder
RANS Reynolds-Averaged Navier-Stokes hydrodynamics ()
TPTR Twin-propeller twin-rudder KT Propeller thrust coecient ()
L Ship length between perpendiculars (m)
Greek symbols Loa Ship length over all (m)
m Ship mass (kg)
Angle of attack (rad) mx Added mass due to motion in x-direction (kg)
Ship drift angle (rad) my Added mass due to motion in y-direction (kg)
P Drift angle at propeller position (rad) N Total hydrodynamic moment acting on midship around
Relative deviation of the parameter (%) the z-axis (Nm)
Rudder angle (rad) n Propeller revolution rate (s1)
h Hydrodynamic inow angle of the rudder (rad) NH Hydrodynamic moment due to hull acting on the ship
Rudder turning rate (s1) around z-axis (Nm)
R Ratio of propeller diameter to rudder span, R = DP / BR () NP Hydrodynamic moment due to propeller acting on the
R Ratio of the wake fraction of the propeller to the wake ship around z-axis (Nm)
fraction of the rudder () NR Hydrodynamic moment due to rudder acting on the ship
kP Impact factor of the propeller slipstream on the rudder around z-axis (Nm)
hydrodynamics () nT Number of performed turning manoeuvres ()
kR Impact factor of the end plates on the rudder hydrody- nZ Number of performed zigzag manoeuvres ()
namics () r Yaw acceleration around midship (rads2)
R Flow straightening coecient of the rudder () r Yaw rate around midship (rads1)
R Experimental constant for expressing uR () rC Yaw rate in steady turn (rads1)
Model scale () S Wetted surface (m)
G Rudder geometric aspect ratio () T Ship draught (m)
E Rudder eective aspect ratio () TP Propeller thrust (N)
Ship displacement volume (m3) tP Propeller thrust deduction ()
Ship heading angle (rad) tR Steering resistance deduction factor ()
O1 First overshoot angle (deg) tO1 Time to the rst overshoot angle (s)
O2 Second overshoot angle (deg) tO2 Time to the second overshoot angle (s)
Water density (kgm3) u Ship acceleration in x-direction (ms2)
T Average absolute deviation of the turning criteria (%) u Forward speed in x-direction, u = V cos (ms1)
Z Average absolute deviation of the zigzag criteria (%) uR Longitudinal velocity of the inow to rudder (ms1)
v Ship acceleration in y-direction (ms2)
Roman symbols V Ship velocity on midship, V = u 2 + v 2 (ms1)
v Sway speed in y-direction on midship, v = V sin (ms1)
AD Advance in the turning manoeuvre (m) VA Propeller advance speed (ms1)
AD Non-dimensional advance in the turning manoeuvre, VC Speed in steady turn (ms1)
AD = AD / L () VR Rudder inow velocity (ms1)
aH Rudder force increase factor () vR Lateral velocity of the inow to rudder (ms1)
AR Rudder lateral area without the horn part (m2) VS Service speed (ms1)
ARP Rudder lateral area in the propeller slipstream (m2) wP Wake factor at propeller position in manoeuvring ()
TD Tactical diameter of turning circle test (m) wR Wake factor at rudder position in manoeuvring ()
TD Non-dimensional tactical diameter of turning circle test, wP0 Wake factor at propeller position in straight moving ()
T D = TD / L () X Total hydrodynamic force acting on midship in the x-
B Ship width at the water level (m) direction (N)
BR Rudder span (m) xG Longitudinal position of centre of gravity in o xyz (m)
Cb Block coecient () XH Hydrodynamic force due to hull acting on midship in x-
CD Drag coecient () direction (N)
CL Lift coecient () xH Longitudinal position of acting point of additional lateral
CR Rudder chord length (m) force (m)
C D0 Drag coecient at zero angle of attack () XH ( , r) Longitudinal hull force due to manoeuvring motions
CL 0 Lift coecient at zero angle of attack () expressed by and r on midship in x-direction (N)
DP Propeller diameter (m) XH(u) Longitudinal hull force due to straight moving on midship
FD Rudder drag force (N) in x-direction (N)
FL Rudder lift force (N) XH (v, r) Longitudinal hull force due to manoeuvring motions
FN Rudder normal force (N) expressed by v and r on midship in x-direction (N)
XP Hydrodynamic force due to propeller acting on midship in

288
J. Liu et al. Ocean Engineering 137 (2017) 287308

Due to the dierences between inland vessels and seagoing ships, Table 1
such as the ratios of the ship main particulars and, especially, the Full-scale and model-scale main particulars of the reference inland vessels.
rudder design choices, current seagoing ship based empirical methods
Parameter 6700 t bulk carrier 3500 t tanker
need to be carefully assessed before they are applied to inland vessels.
This paper reviews the available methods for each parameter that is Full-scale Model-scale Full-scale Model-scale
needed to perform manoeuvring simulations. Dierent combinations of
() 1 1/24.272 1 1/22.815
the available methods are tested and compared to nd the one that best
L (m) 107.5 4.429 94.6 4.146
ts the free-running results of the 6700 t TPTR bulk carrier. However, B (m) 19.2 0.791 17.2 0.754
it is not possible to perform exhaustive tests of all the possible T (m) 4.2 0.173 4.2 0.184
combinations. Therefore, some methods are preselected based on S (m) 2784.30 4.665 2209.30 4.244
theoretical estimation, model test experience, and benchmark values (m3) 7561.70 0.529 5844.82 0.492
Cb () 0.867 0.867 0.855 0.855
in literature. Afterwards, the proposed integrated model is veried with
Cm () 0.999 0.999 0.999 0.999
the 3500 t tanker. Cp () 0.868 0.868 0.856 0.856
Rudder design choices aect ship manoeuvrability, fuel consump- DP (m) 2.50 0.11 2.35 0.11
tion, and rudder cavitation. To account for the large varieties in rudder CR (m) 3.05 0.126 2.60 0.114
AR (m) 10.51 0.018 6.76 0.013
congurations of inland vessels, this paper describes a procedure of
G () 1.13 1.13 1.00 1.00
using 2D open-water Reynolds-Averaged Navier-Stokes (RANS) results VS (ms1) 5 1.015 5 1.047
to calculate the rudder forces and moments. The RANS method was
validated and presented in the authors previous work (Liu et al., 2016).
Accordingly, regression formulas of the hydrodynamic coecients of Table 2
dierent rudder proles are provided to apply the proposed procedure. Typical main particular ratios of seagoing and inland vessels.
In addition, the RANS method is also applicable to analyse other
Ship model KVLCC2 KCS Inland bulk Inland Inland
rudder design choices, such as the number of rudders, the spacing tanker container carrier tanker tanker
among multiple rudders, and the end plates.
Following this introduction, Section 2 presents the main particulars Environment Sea Sea Yangtze Yangtze Rhine
of the reference inland vessel. Section 3 describes the applied math- River River River
L (m) 320.0 230.0 107.5 94.6 110.0
ematical model, reviews the existing empirical methods, describes the B (m) 58.0 32.2 19.2 17.2 11.4
procedure to calculate rudder forces and moments, and preselects T (m) 20.8 10.8 4.2 4.2 3.5
some part models. Since none of the published applicable ranges of the Cb () 0.801 0.651 0.867 0.855 0.883
available regression formulas for the hull forces and moments covers L /B () 5.99 7.14 5.59 5.50 9.65
B /T () 2.57 2.98 4.57 4.10 3.26
the reference ships, Section 4 selects the most suitable regression
L /T () 15.38 21.30 25.64 22.53 31.43
method by comparing the simulation results and the free-running tests
of the 6700 t bulk carrier in various turning and zigzag manoeuvres.
Accordingly, Section 5 performs manoeuvring simulations of the 6700 t the 6700 t bulk carrier have top and bottom end plates while those of
bulk carrier. Section 6 further veries the proposed method with the the 3500 t tanker do not have end plates. The end plates are commonly
3500 t. In the end, Section 7 draws the conclusions and suggests future applied on inland vessels to improve the eective aspect ratios of the
research. rudders and, further, improve the ship manoeuvring performance.
Fig. 1 shows the rudder prole and the end plates.
2. Reference inland vessels With respect to the hull form, the 3500 t tanker ts a bulbous bow
and a slender stern while the 6700 t bulk carrier features a normal bow
Two twin-propeller twin-rudder (TPTR) standard ships from the and a blunt stern. In general, a bulbous bow reduces wave-making
Yangtze River are taken as the reference inland vessels. One is a 6700 t resistance over a narrow range of speed and draught, which results in a
bulk carrier, and the other one is a 3500 t tanker. The main particulars higher speed for the same power or a lower power for the same speed.
and the free-running tests of the reference ships are provided by Most of the previous studies focused on the impacts of the bulbous bow
Wuhan Rules and Research Institute, China Classication Society on powering while no study mentioned its impacts on ship manoeuvr-
(CCS) to study the mathematical modelling of inland vessels. The ability.
model tests include turning manoeuvres with rudder angles of 15, 25, Compared to the particulars of the 3500 t tanker (a bulbous bow, a
and 35 and zigzag manoeuvres with rudder angles/ship heading slender stern, and rudders without end plates), the 6700 t bulk carrier
angles of 10/10, 15/15, and 20/20 on both port and starboard (a normal bow, a blunt stern, and rudders with end plates) is a more
sides. The full-scale and model-scale main particulars of the reference typical inland vessel. Additionally, inland vessels in Europe are more
inland vessels are shown in Table 1. similar to the arrangements of the 6700 t bulk carrier than the 3500 t
The main particulars of the two reference inland vessels are typical tanker. Therefore, in this paper, we use the 6700 t bulk carrier to
for the Yangtze estuary. However, the main particular ratios of the primarily select proper empirical methods and take the 3500 t tanker
reference ships are dierent from those of the European inland vessels as a verication study.
and common seagoing ships as compared in Table 2. Furthermore, the
water depth to the ship draught ratio (H / T ) of the Yangtze River is
commonly much larger than that of the European waterways where 3. Mathematical model
H / T is typically around 1.4. The water depth in the downstream of the
Yangtze River is frequently larger than 12.5 m, where H / T is larger The following sections describe the mathematical models that we
than 3.0 for the reference inland vessels. Therefore, in this paper, compared and applied for inland vessels. Each section starts with a
shallow water eects are neglected for the reference Yangtze inland brief review of the existing empirical methods. Section 3.1 introduces
vessels while they might be important for European inland vessels. the applied coordinate systems. Section 3.2 presents the equations of
The two reference ships are marginally directional unstable. motion and the estimation of (added) mass and (added) moment of
Furthermore, these ships are dierent in the rudder conguration inertia. Section 3.3 lists the available regression methods for hull forces
and the hull form. Regarding the rudder conguration, both vessels and moments, which are determined in Section 4. Section 3.4 estimates
equip twin spade type NACA 0015 rudders. However, the rudders of the propeller forces and moments. Section 3.5 describes the procedure

289
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 1. The NACA 0015 prole for both vessels and the end plates of the 6700 t bulk carrier.

of using 2D open-water RANS results to compute the rudder forces and gravity (COG) of the ship (G) in the o0 x 0y0 z 0 system. COG of the ship
moments. is located at (xG , 0, 0) in the o xyz system. If not particularly specied,
In this paper, non-dimensional parameters are utilised and indi- the parameters, such as velocity (u, v, r, and V), acceleration (u , v , and
cated by superscript prime. Linear velocity (u , v ), angular velocity (r), r ), force (X and Y), and moment (N), are dened on or around midship
force ( X , Y ), moment (N), mass (m, mx , m y ), and moment of inertia (O).
(Iz,Jz) are non-dimensionalised by ship advance speed (V), water
density (), ship length between perpendiculars (L), and ship draught 3.2. Equations of motion
(T) as the following:
3.2.1. Dynamic equations
u v
u, v = , Most of the previous studies on TPTR ship manoeuvrability used
V V
similar modular type models (Yoshimura and Sakurai, 1989; Lee et al.,
rL
r = 2003; Kang and Hasegawa, 2007; Kim et al., 2007; Khanr et al., 2011;
V
X Y Bonci et al., 2015) and proved the usability of the modular model for
X , Y = 2
, 2 the manoeuvring prediction of TPTR ships in both shallow and deep
0.5LTV 0.5LTV
N water. Owing to dierent simplications of the added mass and the
N = location of the body-xed coordinate system (COG or midship), the
0.5L2TV 2
m equations of motion are not entirely identical in the above-mentioned
m mx y
2
m, mx , my = 2
, 2
, literature.
0.5L T 0.5L T 0.5L T
According to Yasukawa and Yoshimura (2014), we apply the
Iz Jz
I z , J z = , . following format of the dynamic equations:
0.5L 4T 0.5L 4T (1)
(m + mx )u (m + m y )vr xGmr 2 = XH + XP + XR


(m + m y )v + (m + mx )ur + xGmr = YH + YP + YR

3.1. Coordinate systems (Iz + xG2 m + Jz )r + xGm(v + ur ) = NH + NP + NR ,
(2)

A three degree-of-freedom model, i.e. surge, sway, and yaw, is where the subscripts H, P, R indicates the hull, the propeller, and the
applied in this paper. Fig. 2 shows applied coordinate systems and sign rudder, m, mx, my are ship mass, added mass in x-direction, and added
conventions. The ship is assumed to sail in the earth-xed coordinate mass in y-direction respectively, Iz and Jz are moment of inertia and
system o0 x 0y0 z 0 with a body-xed coordinate system o xyz on the added moment of inertia around the z-axis, u and v are ship long-
midship point. The predicted trajectory is the path of the centre of itudinal and lateral speed, r is ship yaw rate around midship, and the
dot notation of u, v, and r denotes the derivative of each parameter.

3.2.2. Mass and moment of inertia


Ship mass (m) and moment of inertia (Iz) depend on the ship main
particulars. Routinely, m is calculated as the following:
m = , (3)
where is ship displacement volume. An estimation of the radius of
gyration (iz) is needed for Iz. Motora (1959) estimated Iz as:
iz = 0.2536L


Iz = miz2.

(4)
which is widely used in current manoeuvring studies. Since iz is not
very sensitive to the ship type, in the present work, we apply the
Motora (1959) method for all the following manoeuvring simulations.

Fig. 2. Earth-xed and body-xed coordinate systems for twin-propeller twin-rudder 3.2.3. Added mass and added moment of inertia
ships. According to Hooft and Nienhuis (1994), the added mass and

290
J. Liu et al. Ocean Engineering 137 (2017) 287308

moment, i.e. mx, my, and Jz, can be accurately estimated with the Table 4
charts given by Motora (1959), Motora (1960a), Motora (1960b). Compared methods for the longitudinal force on the hull (XH).
Based on these charts, Zhou et al. (1983) made regression formulas
Case Method for XH (u) Method for XH ( , r) or XH (v, r)
as follows:
No. 1 Holtrop and Mennen (1982) and Neglect XH ( , r) or XH (v, r)
mx 1 T L T
= 0.398 + 11.97Cb1 + 3.73 + 2.89Cb 1 + 1.13 Holtrop (1984)
m 100 B B B No. 2 Holtrop and Mennen (1982) and Yoshimura and Ma (2003) for
Holtrop (1984) XH ( , r)
L 2
T LT No. 3 Holtrop and Mennen (1982) and Ankudinov and Jakobsen (2006) for
+0.175Cb 1 + 0.54 1.107
B B B B
Holtrop (1984) XH (v, r)
No. 4 Holtrop and Mennen (1982) and Kang and Hasegawa (2007) for

my T L Holtrop (1984) XH ( , r)
= 0.882 0.54Cb1 1.6 0.156(1 0.673Cb ) No. 5 Holtrop and Mennen (1982) and Yoshimura and Masumoto (2012)
m B B Holtrop (1984) for XH ( , r)
T L T T L T
+0.826 1 0.678 0.638Cb 1 0.669
B B B B B B
L L paper, we use the Holtrop and Mennen (1982) and Holtrop (1984)
jz = 33 76.85C (1 0.784C ) + 3.43 (1 0.63C b
) method for XH(u).
100
b b
B
Commonly, hydrodynamic coecients obtained from model tests
Jz = mjz2. (5) are used to express XH (v, r) or XH ( , r) (Kijima et al., 1990;
Matsunaga, 1993; Furukawa et al., 2008; Yasukawa and Yoshimura,
Originally for seagoing ships, Clarke et al. (1983) proposed regres-
2014). But, Yoshimura and Ma (2003),Kang and Hasegawa (2007), and
sion formulas based on Planar Motion Mechanism (PMM) tests for my
and Jz . Compared to m, mx is relatively small and Clarke et al. (1983) Yoshimura and Masumoto (2012) proposed regression formulas for the
approximated mx as 36% of m. The values of mx , my , and Jz hydrodynamic coecients in the expression of XH ( , r). Ankudinov
calculated with the Zhou et al. (1983) method and the Clarke et al. and Jakobsen (2006) provided regression formulas to estimate
(1983) method are compared in Table 3. The Zhou et al. (1983) method XH (v, r). With or without XH (v, r) or XH ( , r), 5 methods for XH
and the Clarke et al. (1983) method give comparable results. In this are compared in this paper as listed in Table 4.
paper, we take the Zhou et al. (1983) method as it is based on the
widely used Motora (1959), Motora (1960a), Motora (1960b) charts. 3.3.2. Lateral force and yaw moment on the hull
According to Yoshimura and Sakurai (1989), the hydrodynamic
3.3. Hull forces and moments coecients of the hull, the propeller, and the rudder of a TPTR ship are
not much dierent from those of an SPSR ship. Therefore, the existing
The ship motion through the water induces forces and moments on regression formulas that were developed based on SPSR ships, includ-
the hull. The longitudinal force (XH), lateral force (YH), and yaw ing, but not limited to, Kijima et al. (1990), Matsunaga (1993), China
moment (NH) are commonly expressed in linear or non-linear func- Classication Society (2003), Kijima and Nakiri (2004), Kang and
tions of dimensional or non-dimensional kinematic parameters of the Hasegawa (2007), Furukawa et al. (2008), Yoshimura and Masumoto
ship, such as u, u , v, v , r, r , and . Traditionally, the hydrodynamic (2012), are still applicable for TPTR ships. In this paper, these 8
coecients are derived from series of model tests (Khanr et al., 2011; methods are compared for YH and NH as listed in Table 5, which are
Yasukawa and Yoshimura, 2014). Fast developments of CFD methods tested with the methods for XH (Table 4) in Section 4.
enable the approach of testing model-scale or full-scale ships in a Table 5 also lists the geometric parameters of the reference inland
numerical basin (Broglia et al., 2013; Bonci et al., 2015; Kim et al., vessels and the applicable ranges of the above-mentioned regression
2015; Carrica et al., 2016; He et al., 2016). However, the experimental formulas. To properly use the empirical methods for hull forces and
and numerical tests may be expensive in money and time at the initial moments, the applicable ranges of ship main particulars, such as L / B ,
design stage. In this paper, we take regression formulas of the B / T , L / T , and Cb should be checked as most of the empirical formulas
hydrodynamic coecients from literature. are originally formulated based on databases of seagoing ships.
Moreover, the dynamic parameters, such as the ship speed (V), the
3.3.1. Longitudinal force on the hull yaw rate (r) and the drift angle (), should also be examined. None of
The longitudinal component of the hull force (XH) commonly refers the listed 8 methods perfectly covers the reference inland vessels. In
to the resistance of the bare hull. XH can be described as the general, the applicable range of the Kijima et al. (1990) method, the
component due to the straight moving (XH(u)) and the component Matsunaga (1993) method, the Furukawa et al. (2008), and the Kang
due to manoeuvring (XH (v, r) or XH ( , r)). XH(u) can be obtained and Hasegawa (2007) method t better than the other methods.
through model tests or estimated by the widely used Holtrop and
Mennen method (Holtrop, 1977, 1984, 1978; Holtrop and Mennen, 3.4. Propeller forces and moments
1982). Some authors claim that XH (v, r) or XH ( , r) can be neglected
(China Classication Society, 2003; Prez and Clemente, 2007), but in The propeller forces and moments (XP, YP, and NP) are expressed
this paper, we will explore the consequences of that choice. In this as follows:

XP = (1 tPP )TPP + (1 tPS )TPS


Table 3

Added mass and added moment of inertia coefficients. YP = YPP + YPS

Method 6700 t bulk carrier 3500 t tanker NP = yPP (1 tPP )TPP yPS (1 tPS )TPS .
(6)

mx my Jz mx my Jz where the superscripts P and S indicate the port side and the starboard
side, tP is the propeller thrust deduction in manoeuvring motions, TP is
Zhou et al. 0.0178 0.1550 0.0129 0.0194 0.1708 0.0125 the propeller thrust, YP is the propeller lateral force, and yP is the
(1983)
relative position of the propeller to the centreline of the ship. The
Clarke et al. 0.0094 to 0.1806 0.0113 0.0093 to 0.1941 0.0116
(1983) 0.0187 0.0187 reference inland vessels t two outward-rotating propellers that are
symmetric to the centreline of the ship, where yPP = yPS . In addition, it

291
J. Liu et al. Ocean Engineering 137 (2017) 287308

Table 5
Applicable parameter ranges of the existing regression formulas.

Case Method L /B B /T L /T Cb

6700 t bulk carrier 5.59 4.57 25.64 0.867


3500 t tanker 5.50 4.10 22.53 0.866
No. 1 Kijima et al. (1990) 4.51 to 6.89 2.38 to 4.09 13.66 to 25 0.52 to 0.84
No. 2 Matsunaga (1993) 4.51 to 6.89 2.38 to 4.09 13.66 to 25 0.52 to 0.84
No. 3 Yoshimura and Ma (2003) 2.60 to 5.20 2.17 to 2.70 5.65 to 14.05 0.57 to 0.66
No. 4 China Classication Society (2003) 5.71 to 16.67 3.5 to 6.0 20 to 100
No. 5 Kijima and Nakiri (2004) 2.60 to 5.20 2.17 to 2.70 5.65 to 14.05 0.57 to 0.66
No. 6 Kang and Hasegawa (2007) 5.00 to 6.13 2.43 to 3.31 13.70 to 18.18 0.78 to 0.83
No. 7 Furukawa et al. (2008) 4.51 to 6.89 2.38 to 4.09 13.66 to 25 0.52 to 0.84
No. 8 Yoshimura and Masumoto (2012) 2.60 to 7.10 2.17 to 4.00 5.65 to 28.40 0.51 to 0.65

Table 6 single propeller with single rudder wP0 0.235 while for twin propellers
Wake factor in straight moving (w P0 ) . with a single rudder behind each propeller wP0 0.32 . Harvald (1983)
proposed regression formulas for the wake fraction and the thrust
Method 6700 t bulk carrier 3500 t tanker
deduction factor. Kristensen and Ltzen (2012) indicated that the
Kijima et al. (1990) 0.3835 0.3775 Harvald (1983) method may overestimate these values. Accordingly,
Harvald (1983) 0.3951 0.3951 Kristensen and Ltzen (2012) derived corrections based on the
Kulczyk (1995) 0.32 0.32 Harvald (1983) method for tankers and bulk carriers. According to
Kristensen and Ltzen (2012) 0.2647 0.2647
Kijima et al. (1990),
wP0 = 0.5Cb 0.05. (10)
is assumed that the lateral forces caused by twin propellers are
compensated with each other, i.e. YPP = YPS , therefore YP=0. The values of wP0 which are calculated with the above-mentioned
Following Eq. (6), the propeller thrust (TP) is expressed as the methods are compared in Table 6. In this paper, we take the Kulczyk
following: (1995) method as it is determined based on a database of inland
vessels.
TPP, S = (n P, S )2 (DPP, S )4 KTP, S . (7) The wake fraction during manoeuvring motions (wP) is compli-
where n is the propeller revolution rate, DP is the propeller diameter, cated and related to the drift angle, the yaw rate, the shape of the stern,
and KT is the propeller thrust coecient. As the reference inland the direction of the propeller rotation, and the propeller working load
vessels have two identical propellers rotating at the same constant (Yoshimura and Sakurai, 1989). Kang et al. (2008) concluded that the
number of revolutions per second, for simplicity in the following text, rudder type (single-rudder or twin-rudder) has little inuence on wP0
n = n P = n S and DP = DPP = DPS . Using a usual format, KT is calculated and wP for a single-propeller ship. wP can be estimated based on wP0
by second order polynomials of the propeller advance ratio (JP) as: considering the geometrical inow angle to the propeller in manoeuvr-
ing (P). P is dened as the following:
KTP, S = k 2P, S (JPP, S )2 + k1P, SJPP, S + k 0P, S . (8)
xPP, S
PP, S = r.
where k2, k1, and k0 are propeller open water characteristics in the L (11)
representation of KT. Additionally, these open water characteristics can
where xP is longitudinal position of the propeller. Since xP = xPP
= xPS
be assumed to be constant against the water depth (Yoshimura and P S
for the reference inland vessel, P = P = P . Additionally, the lateral
Sakurai, 1989). Furthermore, JP is expressed as:
position of the propeller also aects P. However, this eect is
u(1 wPP, S ) neglected due to lack of information. Lee et al. (2003) showed that
JPP, S = . 1 wP of an SPSR ship changes systematically with P and reaches a
nDP (9)
minimum around P = 0 , however, 1 wP of a TPTR ship shows a
where u is the forward speed of the ship and wP is the wake fraction at greatly asymmetric trend with the change of P. Matsumoto and
the propeller position in manoeuvring. Thus far, to resolve Eq. (6), the Sueteru (1980) described the relationship of the eective wake fraction
question becomes how to get the wake fraction (Section 3.4.1) and the (1 wP )/(1 wP0 ). Hirano (1980) expressed the relationship of wP and
propeller thrust deduction factor (Section 3.4.2) during manoeuvring wP0 as:
motions. wP
= exp( 4P2 ).
wP0 (12)
3.4.1. Wake fraction at the position of the propeller
Lee et al. (2003) concluded that the propeller's eective wake (1 - The above-mentioned methods are built up originally for SPSR
wP) during manoeuvring and the ow straightening coecient of the ships. For simplicity, Yoshimura and Sakurai (1989) assumed that
rudder in port and starboard turning of the ship are the unique 1 wP = 1 wP0 for each propeller of a TPTR ship. Kang et al. (2008)
parameters of TPTR ships. The wake factor at the position of the described the procedure to obtain t P0 , tP, wP0 , and wP for an SPTR ship
propeller (wP) is discussed here while the ow straightening coecient through model tests. This procedure is also usable for TPTR ships but
of the rudder due to the lateral speed and the yaw rate of the ship (R the model tests at initial stage may not be possible. This paper takes Eq.
and ) will be addressed in Sections 3.5.5 and 3.5.6 respectively. The (12) as it is recommended by China Classication Society (2003).
wake factors at the position of the propeller on port and starboard sides
(wPP and wPS) may be dierent during manoeuvring motions but we 3.4.2. Thrust deduction factor of the propeller
assume wP = wPP = wPS due to lack of information. The wake fraction in A propeller may have dierent deduction factors in straight moving
manoeuvring motions (wP) is commonly estimated based on the wake (t P0 ) and manoeuvring motions (tP). Commonly, tP is not signicantly
factor at propeller position in straight moving (wP0 ) or simply assumed dierent from t P0 for SPSR ships and presumed to be a constant
to be a constant that is identical to wP0 . (Yasukawa and Yoshimura, 2014). According to Kulczyk (1995), for a
For inland vessels, Kulczyk (1995) indicated that for a centre line centre line single propeller with a single rudder t P0 0.27 while for a

292
J. Liu et al. Ocean Engineering 137 (2017) 287308

Table 7
Thrust deduction factor (tP).

Method 6700 t bulk carrier 3500 t tanker

Harvald (1983) 0.3739 0.3739


Kulczyk (1995) 0.2 0.2
Kristensen and Ltzen (2012) 0.3300 0.3300
Holtrop and Mennen, (1978, 1982) 0.2293 0.2293

twin-propeller ship with a single rudder behind each propeller t P0 0.2 .


Lee et al. (2003) showed that tP of a TPTR ship is similar to that of a
SPSR ship. Kang et al. (2008) reported that the variation from t P0 to tP
for SPTR ships appeared to be signicant in the tested cases. However,
additional tests on other ship types are needed to conrm this
dierence (Kang et al., 2008). In this paper, we follow the assumption
of Yoshimura and Sakurai (1989) that tP = t P0 = tPP = tPS . The values of
tP are calculated with the previously mentioned methods and com- Fig. 3. Applied rudder force and angle conventions.

pared in Table 7.
Similar to the calculation of wP, Kristensen and Ltzen (2012)
indicated that the Harvald (1983) method may overestimate tP and where FL and FD are the rudder lift and drag forces, FN and FT are the
puts corrections for tankers and bulk carriers. According to Hollenbach normal and tangential components of the rudder resultant force (FR),
(1999), the Holtrop and Mennen (1978), Holtrop and Mennen (1982) is the applied rudder angle, h is the hydrodynamic inow angle of the
method is more suitable for twin-screw ships than the Harvald (1983) rudder, and R is the eective rudder angle. In this paper, counter-
method. Additionally, the Holtrop and Mennen (1978), Holtrop and clockwise angles are taken as positive.
Mennen (1982) method gives a similar result to the Kulczyk (1995) Normally, FX and FY are calculated based on the rudder normal
method. Again, we take the Kulczyk (1995) method for tP as it is the force (FN) neglecting the rudder tangential force (FT) as follows:
only method that is formed based on inland vessels.
FX = FN sin

FY = FN cos. (14)
3.5. Rudder forces and moments
where is the rudder angle. In addition, FN = 0.5AR VR2CN ,
where AR is
Di Mascio et al. (2011) pointed out that the rudder forces and the rudder area, VR is the rudder inow velocity, CN is the rudder
moments are very dicult to evaluate as they are strongly aected by normal force coecient. Normally, CN is estimated with the widely
the complex ow in the stern region. Commonly, the interactions applied Fujii method (Fujii, 1960; Fujii and Tsuda, 1961, 1962) as the
between the rudders are not considered (Yoshimura and Sakurai, following:
1989). However, the hydrodynamic characteristics of each rudder in
multiple-rudder congurations are dierent due to the rudder proles G
CN = 6.13sinR .
and the interaction between the twin rudders. Gim (2013) showed that G + 2.25 (15)
the distance between the twin rudders plays an important role in where G is the geometric aspect ratio of the rudder as G = BR / CR or
generating the side force and concluded that the critical distance
G = AR / CR2 . BR and CR are the rudder span and the rudder chord
between the rudders should be less than one chord length reducing
respectively.
the turbulence ow and vortices. Kang et al. (2008) concluded that the
The neglect of FT is applicable for small rudder angles and well-
impact factors on twin-rudder performance are the inow angle to each
streamlined rudder proles, such as the NACA and the HSVA series.
rudder, the interactions between the twin rudders, and the decrement
For high-lift proles like the shtail or the wedge-tail series, it is not
of the inow to the twin rudders. Furthermore, Kang et al. (2008)
proper to neglect FT due to large rudder induced drag. Furthermore,
observed that the eective wake fraction, the rudder inow velocity,
the impacts of the rudder prole, the spacing between the twin rudders,
and ow straightening factor for the twin rudders are asymmetric
and end plates should be considered to compute the rudder forces and
during manoeuvring. Kang et al. (2011) investigated the inow
moments of inland vessels. These impacts are not included in Eq. (15)
characteristics of each rudder in the single-propeller twin-rudder ship
while they are considered in this paper.
showing that the inow is not parallel to the ship centre line. This
In this paper, FX and FY are computed based on the rudder lift
phenomenon leads to asymmetric manoeuvring characteristics, redu-
force (FL) and the rudder drag force (FD) as follows:
cing ship manoeuvrability.
Taking the origin on midship, total rudder forces and moments of a FXP, S = FLP, S sinhP, S + FDP, S coshP, S

TPTR ship (XR, YR, and NR) are expressed as follows:


FYP, S = FLP, S coshP, S FDP, S sinhP, S .
(16)
XR = (1 tRP )FXP (1 tRS )FXS

where h is the hydrodynamic inow angle of the rudder and expressed
P P
YR = (1 + aH )FY (1 + aH )FYS S
as:

NR = (xRP + aHPxHP )FYP + yRP (1 tRP )FXP (xRS + aHS xHS )FYS + yRS (1 tRS )FXS .
v
h = arctan R .
(13) uR (17)
where tR is the steering resistance deduction factor (Section 3.5.7), aH
where uR and vR are longitudinal and lateral components of the rudder
is the rudder force increase factor (Section 3.5.8), xH is the long-
itudinal coordinate of the acting point of the additional lateral force inow speed as VR = uR2 + vR2 (Sections 3.5.2 and 3.5.4). Furthermore,
FL and FD are written as follows:
(Section 3.5.9), xR = xRP = xRS and yRP = yRS are the longitudinal and
lateral positions of the rudder, FX and FY are rudder forces in
FLP, S = 0.5ARP, S CLP, S (VRP, S )2
longitudinal and lateral directions.
P, S P, S P, S P, S 2
FD = 0.5AR CD (VR ) . (18)
The applied force and angle conventions are illustrated in Fig. 3,

293
J. Liu et al. Ocean Engineering 137 (2017) 287308

where AR = ARP = ARS as the twin rudders have the same area. The lift C 2D
coecient (CL) and the drag coecient (CD) are determined on the CLG = kP L sinR + CL20D G
sinR G + k
eective rudder angle (R), which is written as the following:
C 2 D
R = h. (19) G
CD = kP D 2D

sinR + CD0 G
.
sinR G + k (22)
In this paper, we take the general expressions for h and R (Eqs.
(17) and (19)). These expressions were applied by Yoshimura and Ma where k is the rudder aspect ratio impact factor on rudder
(2003), Yoshimura and Masumoto (2012), Yasukawa and Yoshimura hydrodynamics.
(2014) for SPSR ships, Kang et al. (2008) for SPTR ships, Yoshimura 6. The rudders of the reference inland vessel are usually designed with
and Sakurai (1989) for TPTR ships. Other expressions of R were top and bottom end plates as shown in Fig. 1. These end plates are
described by Nagarajan et al. (2008) and Khanr et al. (2011) for congured to enlarge the eective aspect ratio of the rudder,
dierent ship types. These methods may improve the accuracy of the therefore, improve the rudder eectiveness. To account the eect
prediction but require additional parameters that may not be available of the end plates on CL and CD, an additional amplify factor kR is
at the initial design stage, for instance, the rudder angle when the lift added. The nal form of the CL and CD for the calculation of FL and
coecient becomes zero. FD are expressed as follows:

C 2D k
3.5.1. Lift and drag coecients of the rudder
CL = kP L sinR + CL20D R G
CL and CD are aected by the rudder design choices, such as the sinR G + k
rudder prole, the rudder property, and the rudder type. Instead of the
C 2D k
prole independent Fujii formula (Eq. (15)), Liu et al. (2016) described CD = kP D sinR + CD2D0 R G .
a procedure of using 2D open-water RANS results for 3D rudders in sinR G + k (23)
propeller slipstream. In this paper, we adapt this procedure as follows:

1. Run 2D RANS simulations of a selected prole in open water, i.e. the


In this paper, the RANS results are obtained from the authors'
NACA 0015 for the reference inland vessel. The test range of angles
previous work (Liu et al., 2016) for Eq. (20) as listed in Table 8. For
of attack (R) is suggested to be 35 to 35 to show the trend of lift
symmetric rudder proles, CL20D is assumed to be zero in uniform ow
and drag curves.
while it should be changed when asymmetric proles are applied. For
2. Based on the 2D results of the lift coecient (CL2D ) and the drag
well-streamlined slender NACA proles, CD2D0 can be neglected. When
coecient (CD2D ), derive the regression formulas of CL2D and CD2D as
high-lift proles like wedge-tail and shtail rudders, which may cause
the following: more drag than the NACA series, CD2D0 should be calculated accordingly.
CL2D Here, for the NACA 0015 prole, we take CL20D and CD2D0 as zero for
CL2D = sinR + CL20D simplicity. According to the tests performed by Nienhuis (1987), the lift
sinR
and drag curve slopes of rudders in open water are not much dierent
CD2D

CD2D = sinR + CD2D0 . from those of rudders in propeller slipstream. Thus, we assume kP=1.0
sinR (20) for Eq. (21). Following the Fujii method, we set k = 2.25 for Eq. (22).
where CL20D
and CD2D0
are lift and drag coecients at zero angle of From experience, we take kR=1.3 in Eq. (23).
attack respectively. The reference points for Eq. (20) are within The interaction between the rudders and shallow water also
R < = 10 to minimise the impacts of the inaccuracy of the RANS inuence CL and CD (Liu and Hekkenberg, 2016b). As the spacing
model due to strong ow separation at large angles of attack. The between the twin rudders of the reference inland vessel is rather large,
obtained 2D results are, in fact, identical to those of 3D rudders with which is 3.81CR , we assume no signicant interaction eect on the CL
an innite geometric aspect ratio. and CD of each rudder. The shallow-water eect on the rudder
3. Commonly, the stall angle of the rudder in open water is around 15 hydrodynamics (CL and CD) can be accounted by adding an additional
while the stall angle is extended to around 40 when the rudder is in correction factor for the eective rudder aspect ratio, which is similar
propeller slipstream. In this paper, we assume the stall angle of the to the consideration of the end plate eect. However, these impacts
rudder in the propeller slipstream is extended larger than the only applies to extremely small under-keel clearance (H / T < 1.2 ) (Liu
maximum applied rudder angles (35). and Hekkenberg, 2016a). Furthermore, the value of H / T also aects
4. The propeller slipstream may also aect the slopes of the lift and
drag curves. The impact factor of the propeller on the lift and drag Table 8
Regression coefficients for the rudder hydrodynamic characteristics of benchmark
slopes (kP) mainly depends on the propeller working load and
profiles.
relative positions of the propeller and the rudder. Then, Eq. (20)
becomes: Rudder prole CL2D CL20D CD2D
CD2D
0
sinR sinR
C 2D
CL = kP L sinR + CL20D NACA 0012 6.238 0.000 0.033 0.008
sinR
NACA 0015 6.175 0.000 0.032 0.008
C 2D
2D
NACA 0018 6.014 0.000 0.031 0.009
CD = kP
D
sinR + CD0 . NACA 0020 5.845 0.000 0.032 0.010
sinR (21) NACA 0025 5.260 0.000 0.036 0.011
IFS58 TR15 6.852 0.000 0.033 0.014
where CL
and CD
are lift and drag coecients of 3D rudders with IFS61 TR25 7.265 0.000 0.051 0.019
innite aspect ratios in propeller slipstream. IFS62 TR25 7.096 0.000 0.078 0.011
5. We follow the Fujii method to transfer Eq. (21) for lift and drag Wedge-tail 0015 7.159 0.000 0.005 0.065
Wedge-tail 0020 7.569 0.000 0.003 0.055
coecients (CLG and CDG ) of the rudders with specied geometric
Wedge-tail 0025 8.004 0.000 0.014 0.045
aspect ratio (G) in propeller slipstream as follows:

294
J. Liu et al. Ocean Engineering 137 (2017) 287308

the rudder-hull interaction factor aH. However, typical H / T for the according to China Classication Society (2003). wR0 can be estimated
reference ships is larger than 3.0 and thus the shallow water impacts on based on the wake fraction ratio (R) as proposed by Kijima et al.
CL, CD, and aH are insignicant and neglected in this paper. In the (1990),
end, the expression that we use for CL and CD in Eq. (18) are described
1 wR0 B 2 B
as follows: R = = 156.2Cb + 41.6Cb 1.76.
1 wP0 L L (29)
1.3G
CL = 6.175sinR
G + 2.25 In this paper, we utilise the Kijima et al. (1990) method (Eq. (29))
, for 6700 t bulk carrier, to estimate R for wR0 and the Furukawa et al. (2008) method (Eq. (28))
1.3G
CD = 0.032sinR
G + 2.25
(24a)
for wR. These two methods are pure empirical methods and have been
widely applied on in previous studies for dierent ship types.
G Therefore, they are selected for the present work.
CL = 6.175sinR
G + 2.25
, for 3500 t tanker. 3.5.4. Lateral component of the rudder inow velocity
CD = 0.032sinR
G
According to Yoshimura and Sakurai (1989), Yoshimura and
G + 2.25
(24b) Masumoto (2012), and Yasukawa and Yoshimura (2014), the lateral
component of the rudder inow velocity (vR) is written as:
3.5.2. Longitudinal component of the rudder inow velocity vRP, S = RP, SRP, S . (30)
Toxopeus, p 30) (2011) concluded that the determination of the
rudder inow velocity and direction as a consequence of the drift angle where R is the ow straightening factor due to the lateral speed of the
(), the yaw rate (r), and the propeller action is one of the most ship (v) and discussed in Section 3.5.5. R is the eective inow angle
complicated aspects in determining the rudder forces and moments. to the rudder in manoeuvring and expressed as
The longitudinal and lateral inow velocity components to each rudder
RP, S = r. (31)
(uR and vR) aect the amount of the lift and drag forces (Eq. (18)) and
the eective rudder angle (Eq. (19)). According to Yoshimura and where = / L is the ow-straightening factor due to the yaw rate (r),
Sakurai (1989), considering the relative position of the propeller (yP) which is further discussed in Section 3.5.6.
for multiple-propeller ships, uR is expressed as:
3.5.5. Flow straightening factor due to the lateral speed of the ship
uRP, S = (1 wRP, S )(u yPP, S r )
Through wind-tunnel tests, Molland and Turnock (2002) found that

2
ow straightening eects depend on the type of upstream body, drift
8KTP, S
R1 + R 1 + 1 + (1 R ) . angle, and propeller thrust loading. Kang et al. (2008) showed that R
(J P, S )2
(25) depends on the eective rudder angle and drift angle. Kim et al. (2007)
indicated that the twin rudders have dierent inow angles during
where R is a constant that can be 0.5 (Yasukawa and Yoshimura, manoeuvring. The lee side rudder has larger ow straightening eects
2014) or 0.55 (Yoshimura and Masumoto, 2012). Kang et al. (2008) than the upwind side rudder. Meanwhile, the ow straightening
assumed that R is the same for both single-rudder and twin-rudder coecient R for the TPTR ship is smaller to that for the SPSR ship.
systems when they both have a single propeller. In this paper, R is For SPSR and TPTR ships, R may be slightly asymmetric for port and
estimated by the Yoshimura and Ma (2003) method, which considers starboard manoeuvres (Lee et al., 2003; Kang et al., 2008; Khanr
the ship particulars, as the following: et al., 2011) while for SPTR ships, these asymmetric phenomena
B become signicant (Khanr et al., 2011).
R = 0.55 0.8Cb .
L (26) After reviewing the above-mentioned literature, we assume the R
for SPSR and TPTR ships are similar and not signicantly dierent for
The wake fraction at the position of the rudder (wR) is discussed in
port and starboard manoeuvres. Kijima et al. (1990) and Ankudinov
Section 3.5.3, R is the ratio of the rudder area in the propeller
et al. (1993) proposed dierent formulas for SPSR ships. For merchant
slipstream (ARP) to the area of the rudder moveable area (AR) and R
ships and shing vessels, Yoshimura and Masumoto (2012) described
is commonly estimated as:
R as the following:
ARP D
R = P. B
AR BR (27) R = 2.06Cb + 0.14.
L (32)
Eq. (27) is derived based on the assumption that the trailing edge of the The values of R which are calculated by the Kijima et al. (1990)
rudder is always in the propeller slipstream. The eect of a part of the method, the Ankudinov et al. (1993) method, and the Yoshimura and
rudder that is out of propeller slipstream is not modelled and very little Masumoto (2012) method for the reference inland vessel are compared
work is available in the literature. Quadvlieg (2013) indicated that the in Table 9. According to the experimental results from Yoshimura and
rudders of inland vessels are always in the propeller slipstream, and we Ma (2003), Yoshimura and Masumoto (2012), and Yasukawa and
take this assumption for the reference inland vessel. Yoshimura (2014), the possible range of r is 0.40.6. Therefore, the
Kijima et al. (1990) method and the Ankudinov et al. (1993) method
3.5.3. Wake fraction at the position of the rudder may underestimate R for the reference inland vessel. In this paper, we
The wake fraction at the position of the rudder during manoeuvring take the Yoshimura and Masumoto (2012) method for R as its
motions (wR) is commonly estimated based on the wake fraction of the
rudder in straight moving (wR0 ), the wake fraction at the position of the Table 9
propeller in straight moving (wP0 ) and manoeuvring motions (wP). Flow straightening factor due to the lateral speed of the ship (R).
Furukawa et al. (2008), described wR as the following:
Method 6700 t bulk carrier 3500 t tanker
2
r
wR = RwR0 exp 4.0 + . Kijima et al. (1990) 0.1508 0.1464
2 (28) Ankudinov et al. (1993) 0.2532 0.2680
Yoshimura and Masumoto (2012) 0.4590 0.4603
where R is an empirical coecient and it is assumed to be 1.0

295
J. Liu et al. Ocean Engineering 137 (2017) 287308

Table 10 Table 11
Flow-straightening factor due to the yaw rate of the ship (). Rudder force increase factor (aH).

Method 6700 t bulk carrier 3500 t tanker Method 6700 t bulk carrier 3500 t tanker

Kijima et al. (1990) 1.00 1.00 Kijima et al. (1990) 0.9758 0.9402
Yoshimura and Ma (2003) 0.9368 0.9357 Yoshimura and Masumoto (2012) 0.5575 0.5598
Yoshimura and Masumoto (2012) 0.90 0.90 Quadvlieg (2013) 0.3906 0.3831

prediction lies in the reasonable range of values. inland vessels. As specied by Eloot (2006), xH moves towards
midships with decreasing under keel clearance leading to reduced
3.5.6. Flow straightening factor due to the yaw rate of the ship turning ability in shallow water. Kijima et al. (1990) made a regression
Kijima et al. (1990) expressed R = 2xR r and thus = 2xR 1.0 in formula for xH . Lee and Shin (1998) proposed formulas of xH for low-
general. Yoshimura and Ma (2003) expressed as the following: speed blunt ships with stern bulb and horn type rudders. After
B neglecting the terms that concern the stern bulb, the Lee and Shin
= 1.7Cb 1.2. (1998) equation becomes:
L (33)
For shing vessels and merchant ships, Yoshimura and Masumoto B B 2
(2012) indicated that = 0.90 . The values of calculated with the xH = 6.054 + 58.18 148.44 .
L L (36)
above-mentioned methods are compared in Table 10. Through trial
and error, we found that the manoeuvring performance of the reference The values of xH for the reference vessels are computed with the
inland vessel is not very sensitive to . In this paper, we take the above mentioned method and compared in Table 12. The Lee and Shin
formula proposed by Yoshimura and Ma (2003) as it takes the main (1998) method, the Yoshimura and Ma (2003) method, the Khanr
particulars of the ship into account. et al. (2008) method, and the Khanr et al. (2011) method give similar
results. In this paper, we take the Lee and Shin (1998) method to
3.5.7. Steering resistance deduction factor estimate xH as it concerns the ship main particulars.
According to Matsumoto and Sueteru (1980) and Kijima et al.
(1990), for SPSR ships in deep water, the rudder steering resistance
deduction (tR) can be estimated as the following: 4. Selection of the regression methods for hull forces and
tR = 0.28Cb + 0.45. (34) moments

Koh and Yasukawa (2012) indicated that the inuence of B / T and yP Through reviewing the existing methods for the manoeuvring
on tR is not signicant. Furthermore, Yoshimura (1986) concluded that parameters, the selected methods for the presented manoeuvring
the variation of tR with the change of water depth is negligible. Thus, in simulations are summarised in Table 13, which are the same for both
this paper, we use Eq. (34) for tR and follow the assumption as Khanr reference inland vessels. These methods are rst used to select the most
et al. (2011) made that tR = tRP = tRS . suitable method for the hull forces and moments in this section, and
then utilised for the manoeuvring studies of the reference inland
3.5.8. Rudder force increase factor vessels in Sections 5 and 6. Tables 4 and 5 list the tested methods
Based on the model tests carried out by Kose et al. (1981), for the longitudinal and lateral hull forces and moments. In total, 40
Yasukawa and Yoshimura (2014) indicates the rudder force increase combinations of the methods for XH, YH, and NH are compared
factor (aH) has a common magnitude of 0.30.4 meaning that the against free-running tests for the 6700 t bulk carrier.
lateral force acting on the ship by steering increases about 3040% To compare the performance of each combination, the average
larger than the rudder normal force component. Kijima et al. (1990) absolute deviation of the simulated turning criteria (T) is calculated as
made a regression formula of aH for SPSR seagoing ships in deep the following:
water. Yoshimura and Masumoto (2012) calculated aH for shing
vessels and merchant ships. For inland vessels, Quadvlieg (2013) nT A Sim A Exp TDSim TDExp VCSim VCExp
proposed: T = 100% D D
+ +
i =1 ADExp TDExp VCExp
aH = 0.627Cb 0.153. (35)
rCSim rCExp
+ 24.
As shown by Khanr et al. (2011), aH is dependent on B / T , Cb, and rCExp (37)
yP while aH may be changed signicantly by the arrangement of the
propellers and rudders. However, little information was found. The where AD is the advance, TD is the transfer, VC is the speed in steady
values of aH computed by the above-mentioned methods are compared turn, rC is the yaw rate in steady turn, and nT is the number of the
in Table 11. From literature (Nagarajan et al., 2008; Yasukawa and performed turning manoeuvres, The superscripts Sim and Exp stand
Yoshimura, 2014) and experience, a reasonable value of aH should be for Simulation and Experiment. Furthermore, the average absolute
around 0.4. In the presented simulations, we take the Quadvlieg (2013) deviation of the simulated zigzag criteria (Z) is written as:
method for aH as it is the only method that is intended for inland
vessels. Table 12
Non-dimensional longitudinal coordinate of the acting point of the additional lateral
3.5.9. Longitudinal coordinate of the acting point of the additional force (xH ).
lateral force
Method 6700 t bulk carrier 3500 t tanker
As stated by Khanr et al. (2008), the non-dimensional longitudinal
coordinate of the acting point of the additional lateral force (xH ), Kijima et al. (1990) 0.1369 0.0359
where xH = xH / L , has a general value of 0.40. Furthermore, Khanr Lee and Shin (1998) 0.3980 0.3827
Yoshimura and Ma (2003) 0.45 0.45
et al. (2011) indicated that xH has an almost constant value of 0.37,
Khanr et al. (2008) 0.40 0.40
which is not signicantly aected by B / T , Cb, and yP. Therefore, it is Khanr et al. (2011) 0.37 0.37
reasonable to assume xH is not much dierent for seagoing ships and

296
J. Liu et al. Ocean Engineering 137 (2017) 287308

Table 13 Using 40 dierent combinations of methods for XH, YH, and NH,
Applied methods for the 6700 t bulk carrier and the 3500 t tanker. the average absolute deviation of the simulated criteria in turning and
zigzag manoeuvres (T with Eq. (37) and Z with Eq. (38)) are
Parameter Method Equation
calculated and compared in Table 16. XH No. in the rst row refers
Iz Motora (1959) Eq. (4) to the methods listed in Table 4 while YH and NH No. in the rst
mx , my , and Jz Zhou et al. (1983) Eq. (5) column refers to the methods listed in Table 5. Some combinations in
w P0 Kulczyk (1995) Constant Table 16 fail to resolve the manoeuvring equations and a few methods
wP Hirano (1980) Eq. (12) do not give reasonable results, which are indicated as Fail, due to
tP Kulczyk (1995) Constant inaccurate estimation of the hull forces and moments.
CL and CD Liu et al. (2016) Eq. (24)
uR Yoshimura and Sakurai (1989) Eq. (25)
As the tested overshoot angles are actually small values, Z is larger
R Yoshimura and Ma (2003) Eq. (26) than T in Table 16. We suppose that the model with the smallest T
w R0 Kijima et al. (1990) Eq. (29) and comparable Z is the most suitable combination. Therefore, the
wR Furukawa et al. (2008) Eq. (28) combination of XH No. 5 with YH and NH No. 1 gives the best t in the
vR Yasukawa and Yoshimura (2014) Eq. (30) prediction of turning and zigzag manoeuvres for the 6700 t bulk
R Yoshimura and Masumoto (2012) Eq. (32)
carrier. To verify the selection, a similar procedure is carried out for
Yoshimura and Ma (2003) Eq. (33)
tR Kijima et al. (1990) Eq. (34) the 3500 t tanker as shown in Table 17. Again, the combination of XH
aH Quadvlieg (2013) Eq. (35) No. 5 with YH and NH No. 1 performs best for the 3500 t tanker. Thus,
xH Lee and Shin (1998) Eq. (36) the selected combination is veried.
Using XH No. 1 for XH and dierent methods for YH and NH, the
steady yaw rates of the reference inland vessels in various turning
Table 14 manoeuvres are simulated and compared with the experiment values as
Model-scale initial speed of each test manoeuvre for the 6700 t bulk carrier.
shown in Fig. 4. YH and NH No. 1 can give the best t among the tested
Turning 15 25 35 15 25 35 methods, which proves the usability of this method to predict the
u0 (ms1) 1.09 1.08 1.07 1.14 1.20 1.07 steady manoeuvring characteristics of the two reference inland vessels.
Zigzag 10/10 15/15 20/20 10/10 15/15 20/20 Through the comparison, we determine to use the Holtrop and
u0 (ms1) 1.00 1.03 1.03 1.07 1.00 1.01 Mennen (1978),Holtrop and Mennen (1982) method for XH(u) and the
Yoshimura and Masumoto (2012) for XH ( , r), i.e. XH No. 5 in Table 4,
therefore, XH = XH (u ) + XH ( , r). Furthermore, YH and NH is calcu-
Table 15
Model-scale initial speed of each test manoeuvre for the 3500 t tanker. lated by the Kijima et al. (1990) method, i.e. YH and NH No. 1 in
Table 5. This combination is used to perform manoeuvring simulations
Turning 15 25 35 15 25 35 in the following sections as it gives the best t. It should be noted that
u0 (ms1) 1.08 1.07 1.06 1.16 1.15 1.05 the reference inland vessels are actually out of the database that Kijima
Zigzag 10/10 15/15 20/20 10/10 15/15 20/20
u0 (ms1) 1.04 1.06 1.03 1.06 1.04 0.59
et al. (1990) used to derive the formulas.

5. Manoeuvring simulations of the 6700 t bulk carrier


Sim Exp
nZ
Sim Exp t Sim t Exp
Z = 100% O1 Exp O1 + O2 Exp O2 + O1 Exp O1
i =1 O1 O2 tO1 Manoeuvring simulations are performed for 15, 25, and 35
turning manoeuvres (Section 5.1) and 10/10, 10/10, and 10/10

t Sim t Exp
+ O2 Exp O2 24. zigzag manoeuvres (Section 5.2) on both port and starboard sides. In
tO2 (38) the following sections, we perform simulations with model-scale
parameters, then scale the simulated and tested results to full-scale,
where O1 and O2 are the rst and second overshoot angles, tO1 and and compare them in full-scale. We compare the simulated and tested
tO2 are time at rst and second overshoot angles, and nZ is the number results of trajectories, rudder angles, drift angles, heading angles, ship
of the performed zigzag manoeuvres. speed, and yaw rate. The drift angles of the experiment are not
The benchmark free-running tests include 6 turning (15, 25, and available. Here, we only show the simulated drift angles to provide
35 for starboard and port sides) and 6 zigzag (10, 15, and 20 for full information of the simulations. Table 18 presents the parameters
starboard and port sides) tests. The model-scale initial advance speed used in the simulations.
(u0) is slightly dierent for each test manoeuvre as listed in Tables 14,
15 for the two reference inland vessels. Additionally, the propeller 5.1. Turning simulations of the 6700 t bulk carrier
revolution rates (n) are constants of 20.2 s1 and 18.6 s1 for the 6700 t
bulk carrier and the 3500 t tanker respectively. Using the methods in Table 13 and the parameters in Table 18, the

Table 16
Average absolute deviation of the 6700 t bulk carrier using dierent combinations of methods for hull forces and moments in turning and zigzag manoeuvres.

XH No. 1 XH No. 2 XH No. 3 XH No. 4 XH No. 5

T (%) Z (%) T (%) Z (%) T (%) Z (%) T (%) Z (%) T (%) Z (%)

YH and NH No. 1 6.49 10.59 7.03 10.62 Fail 11.15 12.86 10.57 4.89 10.67
YH and NH No. 2 34.53 60.66 35.34 61.23 Fail 49.41 Fail 59.69 33.15 61.59
YH and NH No. 3 104.74 38.18 32.81 38.48 Fail 33.40 Fail 37.58 35.44 38.84
YH and NH No. 4 29.38 17.37 20.32 17.23 Fail 17.49 40.32 17.36 22.01 17.38
YH and NH No. 5 21.27 46.39 20.23 46.96 Fail 36.25 24.87 45.04 19.98 47.40
YH and NH No. 6 11.71 10.09 8.61 10.13 Fail 10.75 19.66 10.13 8.38 10.13
YH and NH No. 7 21.46 57.77 19.81 58.55 Fail 42.53 25.22 55.36 19.37 59.13
YH and NH No. 8 8.14 13.46 7.56 13.52 Fail 12.53 14.96 13.22 4.86 13.68

297
J. Liu et al. Ocean Engineering 137 (2017) 287308

Table 17
Average absolute deviation of the 3500 t tanker using dierent combinations of methods for hull forces and moments in turning and zigzag manoeuvres.

XH No. 1 XH No. 2 XH No. 3 XH No. 4 XH No. 5

T (%) Z (%) T (%) Z (%) T (%) Z (%) T (%) Z (%) T (%) Z (%)

YH and NH No. 1 9.69 29.81 8.45 29.96 36.76 27.89 14.10 29.51 8.14 30.07
YH and NH No. 2 31.69 97.79 31.91 98.51 Fail 82.75 33.19 95.54 31.58 99.09
YH and NH No. 3 51.23 132.15 42.07 133.87 Fail 103.16 Fail 124.65 43.85 135.66
YH and NH No. 4 34.99 33.01 28.10 32.97 Fail 34.19 46.18 33.18 29.15 33.05
YH and NH No. 5 30.85 162.12 30.66 165.43 Fail 117.01 33.16 152.25 30.55 167.73
YH and NH No. 6 15.14 46.74 13.37 47.01 Fail 40.32 19.95 45.45 13.59 47.30
YH and NH No. 7 32.48 248.50 32.08 259.67 Fail 153.89 34.64 222.69 32.06 262.73
YH and NH No. 8 10.79 24.09 8.16 24.11 40.65 23.56 17.13 24.25 8.35 24.22

Fig. 4. Comparison of the simulated and tested r of the 6700 t bulk carrier and the 6700 t bulk carrier in various turning manoeuvres. (a) 6700 t bulk carrier. (b) 3500 t tanker.

Table 18 the simulated and tested turning manoeuvres. Comparing the free-
Applied parameters in the simulations of the 6700 t bulk carrier. running results of various turning manoeuvres, the starboard side AD
and TD are larger than those of the port side. Additionally, asymmetry
Parameters used in the simulations
behaviour is found in the free-running trajectories while it is less
L 4.429 B 0.791 T 0.173 0.5290 obvious in the simulation results. Furthermore, the simulated AD and
S 4.665 Cb 0.867 xG 0.009 mx 0.0178 TD are mostly smaller on the port side while larger on the starboard
my 0.155 Iz 0.021 Jz 0.013 25 side than the tested AD and TD .

Parameters for hull forces and moments


X 0.0019 X r 0.0608 Xrr 0.0665 X 0.0665 5.2. Zigzag simulations of the 6700 t bulk carrier
Y 0.3395 Yr 0.0973 Y 0.5727 Yrr 0.0050
Y rr 0.1731 Y r 0.3656 N 0.0781 Nr 0.0361 The time histories of simulated and tested motion parameters in the
N 0.0381 Nrr 0.0126 N rr 0.0448 N r 0.1295 10/10, 10/10, 15/15, 15/15, 20/20, and 20/20
zigzag manoeuvres are presented in Figs. 11, 12, 13, 14, 15 and 16
Parameters for propeller forces and moments respectively. Additionally, the applied initial speed is listed in Table 14,
DP 0.110 w P0 0.32 tP 0.2 xP 2.1220
which is dierent for each test. The simulated rst and second over
yPS 0.240 yP P 0.240 k2S 0.1075 k1S 0.3507
shoot angles are larger than the tested overshoot angles. Comparing the
k0S 0.3329 k2P 0.1295 k1P 0.3269 k0P 0.3307
results of port and starboard sides, the asymmetric characteristics of
Parameters for rudder forces and moments the simulated overshoot angles are less signicant than the tested ones.
AR 0.010 R 0.9364 w R0 0.3633 CR 0.126 Table 20 compares the rst overshoot angle (O1) and the second
BR 0.142 G 1.129 xR 2.215 yR S 0.240 overshoot angle (O2) of the simulated and tested zigzag manoeuvres.
yRP 0.240 tR 0.2072 aH 0.3906 xH 0.3980
The dierences in the simulated and tested results are primarily caused
R 0.4590 R 0.7752 R 0.937 R 0.426
by the divergence of the simulated and tested ship speed. The larger
speed drop of the simulated results may be caused by the inaccurate
prediction of the hull forces. The Holtrop and Mennen (1982) and
trajectories and time histories of simulated and tested motion para- Holtrop (1984) method may not well predict the resistance of inland
meters in the 15, 15, 25, 25, 35, and 35 turning manoeuvres vessels and further study is needed. Furthermore, the main particulars
are presented in Figs. 5, 6, 7, 8, 9 and 10 respectively. Table 19 of the reference inland vessel are actually out of the applied range of
compares the non-dimensional advance ( AD ), the non-dimensional the Kijima et al. (1990) method for YH and NH. The inaccurate
tactical diameter (TD ), and the relative deviation in percentage () of estimation of the hull forces has more signicant impact on the

298
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 5. Trajectories and time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 15 turning manoeuvre.

Fig. 6. Trajectories and time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 15 turning manoeuvre.

Fig. 7. Trajectories and time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 25 turning manoeuvre.

simulations of the manoeuvres with small rudder angles. In addition, it rudder angle (R) as shown in Eqs. (17), (19) and (30). In addition, the
is important to accurately predict the inertia terms in Eq. (2), especially model-scale simulations are quite sensitive to the initial status of u0
for zigzag tests. However, very few studies have been specially done for and n, which should be carefully matched with the experiments to
inland vessels. perform the simulations.
The simulation study shows that the prediction is quite sensitive to
the wake fractions (wP and wR) and the ow straightening factor (R). 6. Manoeuvring simulations of the 3500 t tanker
The wake fractions aect the inow speed of the propeller and the
rudder which, inuence the magnitude of the forces and moments of To verify the proposed integrated model, this section presents
the propeller and the rudder. The ow straightening eect aects the manoeuvring simulations of the 3500 t tanker. Table 21 presents the
lateral component of the rudder inow (vR). Furthermore, it inuences parameters used in the simulations. The Holtrop and Mennen (1982)
the hydrodynamic inow angle of the rudder (h) and the eective and Holtrop (1984) method is used for XH(u), the Yoshimura and

299
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 8. Trajectories and time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 25 turning manoeuvre.

Fig. 9. Trajectories and time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 35 turning manoeuvre.

Fig. 10. Trajectories and time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 35 turning manoeuvre.

Table 19
Comparison of the simulated and tested turning indices of the 6700 t bulk carrier.

Turning criteria Sim () Exp () (%) Turning criteria Sim () Exp () (%)

AD ( = 15) 3.74 4.08 8.33 AD ( = 15) 3.82 3.59 6.41


T D ( = 15) 4.71 5.08 7.28 T D ( = 15) 4.79 4.72 1.48
AD ( = 25) 2.68 2.83 5.30 AD ( = 25) 2.79 2.68 4.10
T D ( = 25) 2.85 3.02 5.63 T D ( = 25) 2.89 2.84 1.76
AD ( = 35) 2.23 2.27 1.76 AD ( = 35) 2.23 2.18 2.29
T D ( = 35) 2.11 2.05 2.93 T D ( = 35) 2.11 1.99 6.03

300
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 11. Time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 10/10 zigzag manoeuvre.

Fig. 12. Time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 10/10 zigzag manoeuvre.

Fig. 13. Time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 15/15 zigzag manoeuvre.

Masumoto (2012) method is used for XH ( , r), and the Kijima et al. manoeuvres are quite accurate while those of the 15, 15, 25, and
(1990) method is used for YH and NH. Other applied methods are 25 turning manoeuvres are not that satisfactory. All the initial
listed in Table 13. turning stages (the rst quarters of the turning circles) are well
predicted. Therefore, the inaccuracy is primarily caused by the
6.1. Turning simulations of the 3500 t tanker divergence of the steady turn.
Regarding the constant turning stage, the prediction of the TD in
Simulations of the 3500 t tanker are carried out in 15, 25, and 35 the 15, 15, 25, and 25 turning manoeuvres is inaccurate while
turning manoeuvres on both starboard and port sides, which are the prediction of TD in the 35 and 35 turning manoeuvres is
presented in Figs. 17, 18, 19, 20, 21 and 22 respectively. The non- accurate. It shows the defect of the hull force module to estimate the
dimensional turning criteria of the simulated and tested results are forces and moments when the 3500 t tanker is turning with relatively
compared in Table 22. The prediction of the 35 and 35 turning small drift angles. The drift angles for 15, 25, and 35 on starboard

301
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 14. Time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 15/15 zigzag manoeuvre.

Fig. 15. Time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 20/20 zigzag manoeuvre.

Fig. 16. Time histories of simulated and tested motion parameters of the 6700 t bulk carrier in the 20/20 zigzag manoeuvre.

Table 20
Comparison of the simulated and the tested overshoot angles of the 6700 t bulk carrier.

Zigzag criteria Sim (deg) Exp (deg) (%) Zigzag criteria Sim (deg) Exp (deg) (%)

O1 (10/10) 3.36 2.43 38.27 O1 ( 10/ 10) 3.30 3.16 4.43


O2 (10/10) 4.02 4.17 3.60 O2 ( 10/ 10) 4.15 3.65 13.70
O1 (15/15) 5.82 4.49 29.62 O1 ( 15/ 15) 5.70 4.99 14.23
O2 (15/15) 6.75 6.65 1.50 O2 ( 15/ 15) 6.81 5.93 14.84
O1 (20/20) 8.75 6.74 29.82 O1 ( 20/ 20) 8.60 7.66 12.27
O2 (20/20) 9.54 9.33 2.25 O2 ( 20/ 20) 9.62 7.66 25.59

302
J. Liu et al. Ocean Engineering 137 (2017) 287308

Table 21 overestimation of the overshoot angles is possibly caused by the fact


Applied parameters in the simulations of the 3500 t tanker. that the hull damping forces are underestimated. The large predicted
speed drop decreases the rudder inow velocity and reduces the yaw
Parameters used in the simulations
checking force induced by the rudder.
L 4.146 B 0.754 T 0.184 0.492
S 4.244 Cb 0.855 xG 0.011 mx 0.0194 7. Conclusions
my 0.1708 Iz 0.02 Jz 0.0125 25

In this paper, an integrated manoeuvring model has been developed


Parameters for hull forces and moments
X 0.0012 X r 0.0462 Xrr 0.0663 X 0.0613
using empirical methods and RANS results for inland vessels (with
Y 0.3571 Yr 0.0973 Y 0.5885 Yrr 0.0016
typical main dimensions for the Yangtze estuary). The presented model
Y rr 0.2105 Y r 0.3370 N 0.0888 Nr 0.0401 uses empirical methods that are publicly available to estimate man-
N 0.0320 Nrr 0.0123 N rr 0.0543 N r 0.1292 oeuvring parameters while further improvements can be obtained by
carrying out model tests or CFD simulations. The regression methods
Parameters for propeller forces and moments for the hull forces and moment are selected by comparing the
DP 0.11 w P0 0.32 tP 0.2 xP 2.007 simulated and tested manoeuvring indexes in various manoeuvres.
yPS 0.228 yP P 0.228 k2S 0.1075 k1S 0.3507 Furthermore, a procedure of using 2D open-water RANS results for
k0S 0.3329 k2P 0.1295 k1P 0.3269 k0P 0.3307 manoeuvring simulations are described, which can be used to consider
Parameters for rudder forces and moments
the rudder prole, the spacing between rudders, and the end plate
AR 0.013 R 0.9319 w R0 0.3663 CR 0.114 eects on the rudder forces and moments. For these simulations in
BR 0.114 G 1.0 xR 2.073 yR S 0.22 deep water, we recommend to use the Holtrop (1978), Holtrop and
yRP 0.22 tR 0.2106 aH 0.3831 xH 0.3827 Mennen (1982) method for XH(u), the Yoshimura and Masumoto
R 0.4603 R 0.9649 R 0.9357 R 0.426 (2012) method for XH ( , r), and the Kijima et al. (1990) method for YH
and NH. Other applied methods for the manoeuvring parameters are
listed in Table 13.
Manoeuvring simulations are performed with two standard twin-
and port sides turning are about 12, 18, 25 respectively. The 3500 t propeller twin-rudder inland vessels in the Yangtze River, which are a
tanker is tted with a bulbous bow. When the ship is turning with a 6700 t bulk carrier and a 3500 t tanker, in turning (15, 25, and 35)
small drift angle, the bulbous bow may signicantly inuence the ow and zigzag (10/10, 15/15, and 20/20) manoeuvres on both port
separation at the bow and further aect the eective rudder angle and and starboard sides. Through the simulation study, it is conrmed that
the rudder inow speed. However, most of the previous studies on the the presented model can roughly capture the manoeuvring character-
bulbous bow focused on its eect on powering and few of them istics of the reference inland vessels. The concluding remarks are
considered its impacts on manoeuvring. The impacts of the bulbous summarised as follows:
bow on manoeuvring forces and moments are not well considered in
the presented model and further research is suggested. The seagoing ship oriented methods are usable for inland vessels
while additional attention should be paid to large Cb, L / B , and B / T
6.2. Zigzag simulations of the 3500 t tanker of inland vessels.
The asymmetric characteristics are not well captured due to
The heading angles and rudder angles of the 3500 t tanker in 10/ inaccurate input parameters, such as wP, tP, wR, and R. These
10, 10/10, 15/15, 15/15, 20/20, and 20/20 are parameters should be dierent for each propeller and each rudder
presented in Figs. 23, 24, 25, 26, 27 and 28 respectively. Additionally, on starboard and port side manoeuvres.
the applied initial speeds are listed in Table 15, which are dierent in The accuracy of the prediction is quite sensitive to the initial status
the turning and zigzag tests. Table 23 compares the simulated and of the simulation, such as the initial speed and the constant
tested overshoot angles. Signicant speed drops are observed. The propeller revolution rate. Therefore, to properly validate the math-
inaccurate prediction of the speed leads to delays in the time when ematical model, it is necessary to accurately implement the actual
maximum heading angles are reached. Similar to the turning man- initial states of the validation experiments.
oeuvres, a primary reason for these dierences is because the bow The presented procedure of using 2D open-water RANS results for
eect on the hull forces and moments is not well estimated. The manoeuvring studies has been proved to be usable for the reference

Fig. 17. Trajectories and time histories of simulated and tested motion parameters of the 3500 t tanker in the 15 turning manoeuvre.

303
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 18. Trajectories and time histories of simulated and tested motion parameters of the 3500 t tanker in the 15 turning manoeuvre.

Fig. 19. Trajectories and time histories of simulated and tested motion parameters of the 3500 t tanker in the 25 turning manoeuvre.

Fig. 20. Trajectories and time histories of simulated and tested motion parameters of the 3500 t tanker in the 25 turning manoeuvre.

inland vessel. Furthermore, regression coecients of benchmark ing performance of the reference 3500 t tanker, especially for zigzag
proles are provided to apply the procedure. It is also possible to use manoeuvres. However, these impacts are not well covered in the
this method to explore the impacts of the rudder prole, the spacing literature. Further investigations and experiments are requested to
between rudders, the dierence between applied rudder angles, the describe the inland vessel resistance (XH(u)), express the longitudinal
end plates, and the shallow water eect on inland vessel manoeuvr- component hull forces due to manoeuvring motions (XH (v, r) or
ability. XH ( , r)), and collect hydrodynamic coecients for lateral force (YH)
and yaw moment (NH).
This paper is based on two marginally directional unstable inland Research on the impacts of the bulbous bow on ship manoeuvr-
vessels from the Yangtze River. Through the presented simulations, it is ability is suggested. Additional tests are needed to determine the
noticed that the bulbous bow has suspected impacts on the manoeuvr- impact factor of the propeller (kP), the rudder (kR), and the aspect

304
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 21. Trajectories and time histories of simulated and tested motion parameters of the 3500 t tanker in the 35 turning manoeuvre.

Fig. 22. Trajectories and time histories of simulated and tested motion parameters of the 3500 t tanker in the 35 turning manoeuvre.

Table 22
Comparison of the simulated and the tested turning indices of the 3500 t tanker.

Turning criteria Sim () Exp () (%) Turning criteria Sim () Exp () (%)

AD ( = 15) 4.46 4.47 0.22 AD ( = 15) 4.60 4.43 3.84


TD ( = 15) 5.23 5.96 12.25 T D ( = 15) 5.33 5.97 10.72
AD ( = 25) 3.33 3.20 4.06 AD ( = 25) 3.44 4.01 14.21
T D ( = 25) 3.46 3.69 6.23 T D ( = 25 ) 3.51 3.75 6.40
AD ( = 35) 2.79 2.63 6.08 AD ( = 35) 2.80 2.63 6.46
T D ( = 35) 2.65 2.61 1.53 T D ( = 35) 2.67 2.64 1.14

Fig. 23. Time histories of simulated and tested motion parameters of the 3500 t tanker in the 10/10 zigzag manoeuvre.

305
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 24. Time histories of simulated and tested motion parameters of the 3500 t tanker in the 10/10 zigzag manoeuvre.

Fig. 25. Time histories of simulated and tested motion parameters of the 3500 t tanker in the 15/15 zigzag manoeuvre.

Fig. 26. Time histories of simulated and tested motion parameters of the 3500 t tanker in the 15/15 zigzag manoeuvre.

ratio (k) on the rudder lift and drag coecients during manoeuvring Acknowledgements
motions. To further investigate the manoeuvring characteristics of
inland vessels, the authors are carrying out research for single- The rst author Jialun Liu is nanced by the China Scholarship
propeller twin-rudder and twin-propeller quadruple-rudder inland Council under Grant 201206950025. The benchmark inland vessels
vessels. In the future, the extension to shallow water may be of and the free-running test results are provided by Wuhan Rules and
importance when these ships are sailing in shallow water, as may be Research Institute, China Classication Society, Wuhan, China. The
suspected for inland ships. authors are grateful for their collaboration and contribution.

306
J. Liu et al. Ocean Engineering 137 (2017) 287308

Fig. 27. Time histories of simulated and tested motion parameters of the 3500 t tanker in the 20/20 zigzag manoeuvre.

Fig. 28. Time histories of simulated and tested motion parameters of the 3500 t tanker in the 20/20 zigzag manoeuvre.

Table 23
Comparison of the simulated and the tested overshoot angles of the 3500 t tanker.

Zigzag criteria Sim (deg) Exp (deg) (%) Zigzag criteria Sim (deg) Exp (deg) (%)

O1 (10/10) 3.82 2.86 33.57 O1 ( 10/ 10) 3.73 3.42 9.06


O2 (10/10) 5.09 4.95 2.83 O2 ( 10/ 10) 5.25 5.28 0.57
O1 (15/15) 6.40 5.64 13.48 O1 ( 15/ 15) 6.23 5.24 18.89
O2 (15/15) 7.96 6.18 28.80 O2 ( 15/ 15) 8.06 5.92 36.15
O1 (20/20) 9.27 8.34 11.15 O1 ( 20/ 20) 8.21 8.34 1.56
O2 (20/20) 10.69 8.81 21.34 O2 ( 20/ 20) 10.30 8.81 16.91

References asymmetric propeller behaviour by means of free running model tests. Ocean Eng.
68, 4764.
Di Mascio, A., Dubbioso, G., Notaro, C., Viviani, M., 2011. Investigation of twin-screw
Ankudinov, V., Kaplan, P., Jacobsen, B.K., 1993. Assessment and principal structure of naval ships maneuverability behavior. J. Ship Res. 55 (4), 221248.
the modular mathematical model for ship maneuverability prediction and real-time Dubbioso, G., Mauro, S., Ortolani, F., 2015. Experimental and numerical investigation of
maneuvering simulations. In: International Conference on Marine Simulation and asymmetrical behaviour of rudder/propeller for twin screw Ships. In: International
Ship Manoeuvrability (MARSIM'93). St. Johns, Newfoundland, Canada Sep. Conference on Marine Simulation and Ship Maneuverability (MARSIM'15).
Ankudinov, V.K., Jakobsen, B.K., 2006. Physically based maneuvering model for Newcastle upon Tyne, UK Sep.
simulations and test evaluations. In: International Conference on Marine Simulation Eloot, K., 2006. Selection, Experimental Determination and Evaluation of a
and Ship Maneuverability (MARSIM'06).Terschelling, The Netherlands Jun. Mathematical Model for Ship Manoeuvring in Shallow Water. (Ph.D. thesis), Ghent
Bonci, M., Viviani, M., Broglia, R., Dubbioso, G., 2015. Method for estimating University.
parameters of practical ship manoeuvring models based on the combination of Fujii, H., 1960. Experimental researches on rudder performance (1) (in Japanese). J.
RANSE computations and System Identication. Appl. Ocean Res. 52, 274294. Zosen Kiokai 107, 105111.
Broglia, R., Dubbioso, G., Durante, D., Di Mascio, A., 2013. Simulation of turning circle Fujii, H., Tsuda, T., 1961. Experimental researches on rudder performance (2) (in
by CFD: analysis of dierent propeller models and their eect on manoeuvring Japanese). J. Zosen Kiokai 110, 3142.
prediction. Appl. Ocean Res. 39, 110. Fujii, H., Tsuda, T., 1962. Experimental researches on rudder performance (3) (in
Carrica, P.M., Modi, A., Eloot, K., Delefortrie, G., 2016. Direct simulation and Japanese). J. Zosen Kiokai 111, 5158.
experimental study of zigzag maneuver of KCS in shallow water. Ocean Eng. 112, Furukawa, Y., Nakiri, Y., Kijima, K., Hiroshi, I., 2008. The prediction of the
117133. manoeuvrability of KVLCC1 and KVLCC2. In: SIMMAN 2008: Workshop on
China Classication Society, 2003. Guidelines for Inland Vessel Manoeuvrability (in Verication and Validation of Ship Manoeuvring Simulation Methods. Copenhagen,
Chinese). Guidance Notes GD-2003. Denmark, Apr. pp. 914.
Clarke, D., Gedling, P., Hine, G., 1983. Application of manoeuvring criteria in hull design Gim, O.S., 2013. Assessment of ow characteristics around twin rudder with various gaps
using linear theory. Trans. R. Inst. Nav. Archit. 125, 4568. using PIV analysis in uniform ow. Ocean Eng. 66, 111.
Coraddu, A., Dubbioso, G., Mauro, S., Viviani, M., 2013. Analysis of twin screw ships' Harvald, S.A., 1983. Resistance and Propulsion of Ships. John Wiley & Sons.

307
J. Liu et al. Ocean Engineering 137 (2017) 287308

Hasegawa, K., Kang, D., Sano, M., Nabeshima, K., 2006. Study on the maneuverability of cavitation including propeller/rudder interaction and the eects of a tunnel. In:
a large vessel installed with a mariner type Super VecTwin rudder. J. Mar. Sci. Proceedings of the 5th International Symposium on Cavitation. Osaka, Japan 2003.
Technol. 11 (2), 8899. Lee, H.Y., Shin, S.S., 1998. The prediction of ships manoeuvring performance in initial
He, S., Kellett, P., Yuan, Z., Incecik, A., Turan, O., Boulougouris, E., 2016. Manoeuvring design stage. In: Practical Design of Ships and Other Floating Bodies Conference.
prediction based on CFD generated derivatives. J. Hydrodyn. Ser. B 28 (2), 284292. The Hague, The Netherlands, pp. 633639.
Hirano, M., 1980. A practical calculation method of ship maneuvering motion at initial Liu, J., Quadvlieg, F., Hekkenberg, R., 2016. Impacts of the rudder prole on
design stage. Nav. Archit. Ocean Eng. 147, 6880. manoeuvring performance of ships. Ocean Eng. 124, 226240.
Hollenbach, K.U., 1999. Estimating resistance and propulsion for single-screw and twin- Liu, J., Hekkenberg, R., Rotteveel, E., Hopman, H., . 2015. Literature review on
screw ships in the preliminary design. In: Proceedings of the 10th International evaluation and prediction methods of inland vessel manoeuvrability. Ocean Eng.
Conference on Computer Applications in Shipbuilding (ICCAS)., vol. 2. Cambridge, 106, 458471.
Massachusetts, USA, Jun. pp. 237250. Liu, J., Hekkenberg, R., 2016a. 3D RANS simulations of shallow water eects on rudder
Holtrop, J., 1977. A statistical analysis of performance test results. Int. Shipbuild. hydrodynamic characteristics. In: 2016 International Conference on Maritime
Progress. 24 (270), 2328. Technology (ICMT 2016).Harbin, China, Jul. pp. 3539.
Holtrop, J., 1978. Statistical data for the extrapolation of model performance tests. Int. Liu, J., Hekkenberg, R., Jul. 2016b. Interaction eects on hydrodynamic characteristics
Shipbuild. Progress. 25 (588), 122126. of twin rudders. In: 2016 International Conference on Maritime Technology (ICMT
Holtrop, J., 1984. A statistical re-analysis of resistance and propulsion data. Int. 2016).Harbin, China, Jul. pp. 17.
Shipbuild. Progress. 31 (363), 272276. Matsumoto, K., Sueteru, K., 1980. The prediction of manoeuvring performances by
Holtrop, J., Mennen, G.G.J., 1978. A statistical power prediction method. Int. Shipbuild. captive model tests. J. Kansa. Soc. Nav. Archit. 176, 1122.
Progress. 25, 253256. Matsunaga, M., 1993. Method of predicting ship manoeuvrability in deep and shallow
Holtrop, J., Mennen, G.G.J., 1982. An approximate power prediction method. Int. waters as a function of loading condition. Tech. Bull. Nippon Kaiji Kyokai 11, 5159.
Shipbuild. Progress. 29 (335), 166170. Molland, A.F., Turnock, S.R., 2002. Flow straightening eects on a ship rudder due to
Hooft, J.P., Nienhuis, U., 1994. The prediction of the ship's manoeuvrability in the design upstream propeller and hull. Int. Shipbuild. Progress. 49 (3), 195214.
stage. SNAME Trans. 102, 419445. Motora, S., 1959. On the measurement of added mass and added moments of inertia for
Kang, D., Hasegawa, K., 2007. Prediction method of hydrodynamic forces acting on the ship motions (in Japanese). J. Soc. Nav. Archit. Jpn. 105, 8389.
hull of a blunt-body ship in the even keel condition. J. Mar. Sci. Technol. 12 (1), Motora, S., 1960a. On the measurement of added mass and added moments of inertia for
114. ship motions, Part 2: added mass for the longitudinal motions (in Japanese). J. Soc.
Kang, D., Nagarajan, V., Hasegawa, K., Sano, M., 2008. Mathematical model of single- Nav. Archit. Jpn. 106, 5962.
propeller twin-rudder ship. J. Mar. Sci. Technol. 13 (3), 207222. Motora, S., 1960b. On the measurement of added mass and added moments of inertia for
Kang, D., Nagarajan, V., Gonno, Y., Uematsu, Y., Hasegawa, K., Shin, S.C., 2011. ship motions, Part 3: added mass for the transverse motions (in Japanese). J. Soc.
Installing single-propeller twin-rudder system with less asymmetric maneuvering Nav. Archit. Jpn. 106, 6368.
motions. Ocean Eng. 38, 11841196. Nagarajan, V., Kang, D.H., Hasegawa, K., Nabeshima, K., 2008. Comparison of the
Khanr, S., Hasegawa, K., Nagarajan, V., Shouji, K., Lee, S.K., 2011. Manoeuvring mariner Schilling rudder and the mariner rudder for VLCCs in strong winds. J. Mar.
characteristics of twin-rudder systems: rudder-hull interaction eect on the Sci. Technol. 13 (February), 2439.
manoeuvrability of twin-rudder ships. J. Mar. Sci. Technol. 2011 (16), 472490. Nakatake, K., Ando, J., Kataoka, K., Sato, T., Yamaguchi, K., 1989. Study on the
Khanr, S., Hasegawa, K., Lee, S.K., Jang, T.S., Lee, J.H., Cheon, S.J., 2008. propulsive performance of twin screw ship: interaction between propeller and rudder
Mathematical model for maneuverability and estimation of hydrodynamic in a uniform ow (in Japanese). Trans. West-Jpn. Soc. Nav. Archit. 78, 4957.
coecients of twin-propeller twin-rudder ship. In: The Japan Society of Naval Nienhuis, U., 1987. Passieve Manoeuvreerhulpmiddelen: Open Water Proeven met Roer
Architects and Ocean Engineers. No. 7K. Osaka, Japan, Nov. pp. 5760, 2008K-G4- (in Dutch). Tech. rep., Maritime Research Institute Netherlands (MARIN),
3. Wageningen, The Netherlands. Jul.
Kijima, K., Nakiri, Y., 2004. On the practical prediction method for ship manoeuvrability Prez, F.L., Clemente, J.A., 2007. The inuence of some ship parameters on
in restricted water (in Japanese). J. Jpn. Soc. Nav. Archit. Ocean Eng. 107, 3754. manoeuvrability studied at the design stage. Ocean Eng. 34, 518525.
Kijima, K., Katsuno, T., Nakiri, Y., Furukawa, Y., 1990. On the manoeuvring performance Quadvlieg, F., 2013. Theoretische Berekening van Simulatiemodellen voor
of a ship with the parameter of loading condition. J. Soc. Nav. Archit. Jpn. 168 Binnenvaartschepen ten Behoeve van Maatgevende Manoeuvres (in Dutch). Tech.
(Novmber), 141148. rep., Maritime Research Institute Netherlands (MARIN), Wageningen, The
Kim, H., Akimoto, H., Islam, H., 2015. Estimation of the hydrodynamic derivatives by Netherlands.
RANS simulation of planar motion mechanism test. Ocean Eng. 108, 129139. Toxopeus, S., 2011. Practical Application of Viscous-Flow Calculations for the Simulation
Kim, Y.G., Kim, S.Y., Kim, H.T., Lee, S.W., Yu, B.S., 2007. Prediction of the of Manoeuvring Ships (Ph.D. thesis), Delft University of Technology.
maneuverability of a large container ship with twin propellers and twin rudders. J. Yasukawa, H., Yoshimura, Y., 2014. Introduction of MMG standard method for ship
Mar. Sci. Technol. 12 (3), 130138. maneuvering predictions. J. Mar. Sci. Technol. 20 (1), 3752.
Koh, K.K., Yasukawa, H., 2012. Comparison study of a pusher-barge system in shallow Yoshimura, Y., 1986. Mathematical model for the manoeuvring ship motion in shallow
water, medium shallow water and deep water conditions. Ocean Eng. 46, 917. water (in Japanese). J. Kansa. Soc. Nav. Archit., Jpn. 200, 4151.
Kose, K., Yumuro, A., Yoshimura, Y., 1981. Concrete of mathematical model for ship Yoshimura, Y., Sakurai, H., 1989. Mathematical model for the manoeuvring ship motion
manoeuvring (in Japanese). In: Proceedings of the 3rd Symposium on Ship in shallow water (3rd report): Manoeuvrability of a twin-propeller twin-rudder ship.
Manoeuvrability, Society of Naval Architects. pp. 2780. J. Kansa. Soc. Nav. Archit. Jpn. 211 (March), 115126.
Kristensen, H.O., Ltzen, M., 2012. Prediction of Resistance and Propulsion Power of Yoshimura, Y., Ma, N.,2003. Manoeuvring prediction of shing vessels. In: International
Ships. Tech. rep., University of Southern Denmark and Technical University of Conference on Marine Simulation and Ship Maneuverability (MARSIM03).
Denmark, Denmark. Oct. Kanazawa, Japan. Aug.
Kulczyk, J., 1995. Propeller-hull interaction in inland navigation vessel. Trans. Built Yoshimura, Y., Masumoto, Y., 2012. Hydrodynamic database and maneuvering
Environ. 11, 7389. prediction method with medium high-speed merchant ships and shing vessels. In:
Lee, S.K., Fujino, M., 2003. Assessment of a mathematical model for the manoeuvring International Conference on Marine Simulation and Ship Maneuverability
motion of a twin-propeller twin-rudder ship. Int. Shipbuild. Progress. 50 (12), (MARSIM12). Singapore. Apr.
109123. Zhou, Z., Yan, S., Feng, W., 1983. Manoeuvring prediction of multiple-purpose cargo
Lee, H., Kinnas, S.A., Gu, H., Natarajan, S., 2003. Numerical modeling of rudder sheet ships (in Chinese). Ship Eng. 6, 2136.

308