You are on page 1of 14

Ocean Engineering 162 (2018) 290–303

Contents lists available at ScienceDirect

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

A time domain boundary element method for wave added resistance of ships T
taking into account viscous effects
Malte Riesner, Ould el Moctar∗
University of Duisburg-Essen, Institute of Ship Technology and Ocean Engineering, Duisburg, Germany

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

Keywords: Linear boundary element methods generally provide accurately enough predictions of ship motions in small
Added resistance in waves amplitude waves. However, when the wave-induced added resistance is investigated, nonlinear effects are to be
Viscous added resistance in waves accounted for. Higher order effects, such as second order wave-induced drift forces, depend largely on ships
Nonlinear boundary element method forward speed. This article presents a nonlinear time-domain Rankine source method to calculate the wave-
Comparative study
induced added resistance of ships advancing at constant forward speed in regular head waves. The fully non-
linear steady flow was computed. Nonlinear Froude-Krylov and hydrostatic forces were obtained by integrating
pressures over the instantaneous wetted surface (undisturbed incident wave). Radiation forces were computed in
time domain using convolution integrals (Cummins approach); diffraction forces, were computed using the
complex force amplitude resulting from the linear seakeeping problem of incident wave diffraction. Radiation
and diffraction effects caused by the changing wetted surface were accounted for. An empirical approach to
account for the viscous wave added resistance was introduced. Numerical results for a 14,000TEU container ship
and a very large tanker were compared to results from a frequency domain method, CFD calculations, and
experimental towing tank measurements.

1. Introduction An alternative to improving ship efficiency is to design its hull and


propulsion system for the representative sea conditions. Generally,
Optimizing and dimensioning a ship's hull and its propulsion system ships are designed for a service speed in calm water. Actually, the ship
are relevant design issues. In this regard, speed loss in waves has be- seldomly operates under these ideal fair weather conditions. In prac-
come important as it effects fuel consumption, which contributes to tice, an experience based allowance called “sea margin” accounts for
operational costs. Speed loss must be considered not only to meet a the additional resistance in actual weather conditions and ensures that
ship's time schedule, but it has become also a safety issue because the ship is maneuverable also in severe seas. However, this approach
maneuverability has to be maintained under adverse weather condi- does not always meet the requirements of a modern ship design and
tions; see, e.g., Papanikolaou et al. (2015), Kwon (2008), Prpić-Oršić may lead to an unnecessary power reserve or to an underpowered ship
and Faltinsen (2012), Sigmund and el Moctar (2018), Kim et al. (2016). that cannot maneuver safely under adverse sea conditions (Rathje et al.,
Indeed, the International Maritime Organization (2013, 2015) requires 2011). Therefore, the maneuverability of ships in waves is of growing
that ships be equipped with a minimum engine output. Furthermore, interest; see, e.g., Skejic and Faltinsen (2008), Zhang et al. (2017).
ship pollution and carbon emissions have come into focus. To reduce Furthermore, ship lines seem to be optimized for sea conditions the ship
the carbon emissions of ships, the IMO (2011) published regulations may rarely experience rather than for the actual sea conditions. Service
and introduced the Energy Efficiency Design Index (EEDI) relating operational conditions should be considered already during the design
carbon emissions and transported amount of cargo to ship speed. To processes to more accurately optimize the ship's resistance in waves;
meet these requirements in future, ships need to become more efficient. see, e.g., Park et al. (2014), Kim et al. (2014, 2015).
One way to improve efficiency regarding EEDI is to reduce ship speed. Modern ship design calls for tools to assess added resistance in
However, reducing a ship's design speed leads to a decrease of the re- waves. Conventional methods comprise model tests, empirical for-
quired maximum engine output, and this raises the question whether mulas, and numerical techniques. These methods differ significantly in
the ship is able to maintain minimum forward speed under adverse accuracy and cost. Model tests are still one of the most reliable
weather condition to ensure safe maneuverability. methods, exemplified by recent tests conducted within the framework


Corresponding author.
E-mail address: ould.el-moctar@uni-due.de (O. el Moctar).

https://doi.org/10.1016/j.oceaneng.2018.05.010
Received 30 December 2017; Received in revised form 25 March 2018; Accepted 5 May 2018
Available online 29 May 2018
0029-8018/ © 2018 Elsevier Ltd. All rights reserved.
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

of the European research projects Shopera (Papanikolaou et al., 2015) integration over the instantaneous wetted surface. We used a frequency
for a cruise ship (Valanto and Hong, 2015) and a containership domain technique of Söding's et al. (2012) to obtain frequency depen-
(Sprenger et al., 2017). However, model tests are expensive and, dent response coefficients and to determine forward speed effects
therefore, the number of test runs has to be limited. Empirical formulas caused by the steady wave system as well as the associated dynamic
need constant updating and, although useful at the preliminary design trim and sinkage. We also accounted for geometrical nonlinearities
stage, are of limited use for the design of new ship types. Numerical caused by radiation and diffraction using the approach of Boese (1970).
methods are generally based on potential theory or field methods that To account for viscous effects, we introduced an empirical proce-
solve the Reynolds-averaged Navier-Stokes (RANS) equations. The dure based on results obtained by Sigmund and el Moctar et al. (2018),
latter method has been proven useful for ship hydrodynamics problems which proved effective especially in short waves. For a 14,000TEU
associated with strong nonlinear effects, such as the changing wetted containership and a very large tanker advancing at constant forward
surface, breaking waves, and viscous effects. el Moctar et al. (2017), speed in regular head waves, we compared our wave added resistance
Sigmund and el Moctar (2018) for example, used a field method to predictions to those obtained from a RANS solver, a frequency domain
demonstrate that forward speed strongly affects wave added resistance method, and model test measurements. Compared to frequency domain
and that viscous effects influence wave added resistance in short waves. predictions, our method resulted in more accurate predictions, espe-
Others, notably Sportelli and Huijsmans (2012), Lyu and el Moctar cially in short and medium length waves.
(2017), Kim et al. (2017), Ley et al. (2014), Gue et al. (2012), and
Sadat-Hosseini et al. (2010, 2013) also used this method to study wave 2. Numerical method
added resistance. In all cases, however, the computational effort was
high, and only an experienced user was able to obtain reliable results. Following standard practice, two right-handed coordinate systems
The obvious advantage of potential theory based methods is their describe ship motions, namely, an inertial system → x = (x , y, z )T moving
computational efficiency, and their use predominated before the de- with the ship's forward speed and a ship bound coordinate system
velopment of modern computers. Maruo (1960, 1963) was one of the →
x = ( x , y , z )T used to simplify the description of external forces acting
first to assess wave added resistance, using the far field method to on the ship. Coordinate x points positive to the bow; y , positive to port;
calculate radiated and diffracted wave energy and momentum flux at and z , positive upward. The origin of these coordinate systems is lo-
infinity. Gerritsma and Beukelman (1972) and Journee (2001) modified cated at the centre of gravity. The following equation solves ship mo-
this approach within a linear strip theory method and demonstrated the tions:
importance of bow wave diffraction for the prediction of wave added t
resistance in short waves. Following Maruo's approach, Kashiwagi [M + A (∞)] →
x¨ (t ) + B (∞)→
x˙ (t ) + ∫ K (t − τ )→x˙ (τ ) dτ
(1995, 2009, 2011) also accounted for the effect of bow wave diffrac- 0
tion. Salvesen (1978), using Gerritsma's and Beukelman's approach ⎯→
⎯ ⎯→
⎯ ⎯→

= FDif (t ) + FFK + R (t ) + FWL (t ) (1)
within a strip theory method, showed the importance of an accurate
prediction of ship motions when calculating wave added resistance. The left hand side of (1) includes radiation forces based on the
Most of these potential theory based solvers, however, turned out to approach of Cummins (1962) and Ogilvie (1964). Thus, radiation forces
be unreliable to asses wave added resistance in short waves. Boese in time domain result from Fourier transforms of hydrodynamic re-
(1970) tried to resolve this shortcoming by integrating hydrodynamic sponse forces from a frequency domain solution, namely, the hydro-
pressures. Later, Faltinsen et al. (1980) developed a more accurate dynamic added mass at infinite frequency, A(∞), the hydrodynamic
pressure integration method for short waves. Söding et al. (2014) in- damping at infinite frequency, B(∞), and the convolution integral that
troduced a frequency domain potential flow solver based on Rankine calculates the elapsed history of the radiation forces based on the im-
sources, which couples the effect of the nonlinear stationary forward pulse response function, K(t) , and the body velocities, →x˙ . Times t and τ
speed problem with the linear seakeeping problem. His method directly represent, respectively, current time and time of the past history. In-
integrates the pressure at the hull, accounts for the changing wetted ertial ship reactions to the ship's mass are calculated using ship mass
surface (Boese's approach), and time averages the second order pressure
matrix, M, multiplied by ship acceleration vector, → x¨ . The right hand
field. Lui et al. (2017) compared his results obtained with this method →
site of (1) includes diffraction forces, FDif , based on the complex dif-
to RANS based predictions and model test measurements and found
fraction force amplitude resulting from the linear problem of the in-
generally favorable agreement. However, wave added resistance in
cident wave diffraction as well as the combined nonlinear Froude-
short waves was underestimated strongly for several cases. →
Later developments were characterized by time domain potential Krylov and restoring forces, FFK +R , taking into account the changing
flow solvers. Using such an approach, Kim et al. (2010), for example, wetted surface caused by incident, radiated, and diffracted waves and
calculated the added resistance from diffraction and radiation compo- the dynamic trim and sinkage caused by the ship's constant forward
nents, and Liu et al. (2011) employed a three-dimensional frequency speed.
domain panel method and a hybrid time domain Rankine source solver A frequency domain method that couples the stationary nonlinear
based on Green functions and a far-field approach to calculate wave forward speed problem with the linear seakeeping problem obtains
added resistance. At small Froude numbers, predictions from these radiation and diffraction forces in (1) by integrating first order pres-
methods compared well to experiments; however, at large Froude sures over the mean wetted surface. To accurately calculate wave added

numbers, errors increased. A distinct advantage of time domain solvers resistance, an additional force component, FWL , accounts for the
is that nonlinear Froude-Krylov forces are calculated more accurately changing wetted surface caused by these radiation and diffraction
because the instantaneous wetted surface is accounted for without ex- forces.
cessive numerical effort. These forces often contribute strongly to wave
added resistance. 2.1. Frequency dependent response coefficients
The quintessence of this review is that three aspects should be
considered to accurately predict wave added resistance, namely, for- The Rankine source frequency domain method of Söding et al.
ward speed effects, the changing wetted surface (geometrical non- (2012) and von Graefe (2014) calculates frequency dependent hydro-
linearity), and viscous effects. Our method presented here started from dynamic response coefficients and diffraction forces. This method ac-
the ship motions solver of (Riesner et al., 2016). Following Cummins counts for the interaction between the nonlinear stationary flow in calm
approach (Cummins, 1962), we modelled radiation forces via con- water (including steady ship waves and dynamic trim and sinkage) and
volution integrals and Froude-Krylov as well as restoring forces via its interaction with the periodic flow in waves. Following Hachmann's

291
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

ωmax
approach for seakeeping (Hachmann, 1991), the total potential, ϕt (ωe ) , 2
in waves is written as a superposition of the steady potential, ϕ0 , and
K (t ) =
π
∫ [B (ωe ) − B (∞)]cos(ωe t ) dω
0
the periodic potential, ϕ ̂ , as follows:
1
2 2 πt cos(ωmax t ) ⎤
+ ωmax [B (ωmax ) − B (∞)] ⎡t Si (ωmax t ) − +
⎢ ωmax ⎥
ϕt (→
x , t ) = −ux + ϕt (→
x , t ) = −ux + ϕ0 (→
x ) + Re (ϕ ̂ eiωe t ) π 2
1
(2) ⎣ ⎦
(10)
The dynamic boundary condition is expressed as Where ωe is the wave encounter frequency, ωmax is the angular fre-
quency corresponding to the maximum value obtained from frequency
→  ⎞  0⎞
(∇ϕ0 − U ) ⎜⎛∇ϕ ̂ + →
α × ∇ϕ0 − (∇2 ϕ0)→
v ⎟ + iωe ⎜⎛ϕ ̂ − →
1 1
v ∇ϕ ⎟ domain computations, and Si (x ) = ∫ sin (x )/ xdx .
⎝ ⎠ ⎝ ⎠ The diffraction force is derived from the complex diffraction force

⎯→

 0⎞
+ g ⎜⎛ς 1̂ − →
v ∇ς ⎟ = 0 amplitude, FDif , which is obtained from the frequency domain
⎝ ⎠ (3) (Rankine source method) calculation:


→  ⎯→
⎯ 
⎯→

where α is the amplitude of the ship rotation vector, →
v is the amplitude FDif (t )= Re ⎡ FDif eiwe t ⎤.
of the total motion of a point on the ship, and i is the imaginary unit ⎣ ⎦ (11)
number. The kinematic boundary condition reads as follows: Integrating the hydrostatic pressure, pFK + R , up to the instantaneous
wetted surface of an undisturbed wave (known as the Smith effect)
 0⎞ → ∂ 1̂
iωe ⎛⎜ς 1̂ − →
v ∇ς ⎟ + ∇ς 0∇ϕ ̂ + (∇ϕ0 − U ) ∇^ς −
1 1 while also accounting for wave elevation, ship motions, and the sta-
ϕ
⎝ ⎠ ∂z tionary wave system caused by the steady forward speed in calm water
T ⎯→

 ∂ϕ0 ∂ϕ0 ∂ς 0 ⎤ yields the combined Froude-Krylov and restoring force, FFK + R:
+→
α ⎡− , ,u  =0
+A

⎣ ∂y ∂x ∂y ⎥ ⎯→

⎦ (4) FFK + R (t ) = ∫ pFK +R (→x , t ) →
n (→
x , t ) dS.
(12)
where 
S (t )
A includes all second order derivatives and is written as
The linearized Bernoulli equation is solved to calculate the hydro-
 o
→→ →
 = −v n → static pressure, pFK + R , using Wheeler stretching (Wheeler, 1979) to
n [(∇2 ϕ0 ) →
o o
A → o2
n + (∇2 ς 0 )(∇ϕ0 − U )]
n (5) more accurately calculate the pressure near the free surface. Assuming
infinitely deep water, the hydrostatic pressure, pFK + R , then is expressed
with →
n ° being the unit vector normal to the stationary free surface as follows:
T →
pFK + R ⎛⎜→x , k , ωe , t ⎞⎟ = ρg [−z + ζ 0 (→x ) + ζ 1 (→
∂ς 0 ∂ς 0 1→
→o
n =⎡ , , −1⎤ x , t ) e k (z − ζ ( x , t )) ]

⎣ ∂x ∂y ⎥
⎦ (6) ⎝ ⎠ (13)

where ζ 1 ( x , t ) is the free surface elevation caused by the undisturbed
The corresponding body boundary condition is as follows:
incident wave, and ζ 0 (→ x ) is the stationary free surface elevation caused
by the ship's forward speed. To accurately integrate pressures over the
→  → ⎞
n (→
x ) ⎜⎛∇ϕ ̂ + →
α × U − iωe →
1
v ⎟=0 instantaneous wetted surface, panels intersecting the free surface are
⎝ ⎠ (7)
modified according to the approach of Riesner et al. (2016).
Finally, the frequency dependent part of the first order pressure is
calculated from 2.3. Radiation and diffraction forces caused by changing wetted surface


p1 →  0⎞ A linear frequency domain boundary element method yields radia-
= −(U + ∇ϕ0) ∇ϕ ̂ − iωi ⎜⎛ϕ ̂ − →
1 1
v ∇ϕ ⎟ tion and diffraction pressures, which are then integrated over the mean
ρ ⎝ ⎠ (8)
wetted surface, neglecting variations of hull submergence caused by the
To determine hydrodynamic added mass and damping at infinite incident wave and the ship motions. Relative motions between wave
encounter frequency, a simplified boundary value problem is solved. contour and hull and the associated pressure distribution on the hull
The free surface condition prevails instead of the high frequency con- surface are approximated from first order pressure at the mean water-
dition: line (Boese, 1970). implemented a similar approach into a linear fre-
quency domain boundary element method to successfully calculate
ϕ̂ =0
1
(9) time average drift forces. However, our technique calculates the drift
force in time domain. At a typical ship section shown in Fig. 1, the
This condition is satisfied analytically by placing image sources of relative motion is the distance d (t ) above the mean water surface, WL ,
opposite sign source strengths symmetrically on the undisturbed free and the instantaneous wave elevation, WL (t ) . The pressure at the hull is
surface. In addition, more accurate results are obtained by locating p (→
x , t ) ; the pressure at the mean water line, pWL (t ) . (A linear frequency
mirrored sources on the nonlinear stationary free surface when con- domain calculation neglects pressures within the red colored triangle.)
sidering the steady wave system (Söding, 2009). Although not satisfied Assuming a linear pressure distribution between the mean water
exactly, the associated boundary condition (9) is fulfilled to a sufficient surface, WL , and the instantaneous wave elevation, WL (t ) , the corre-
level of accuracy. →
sponding force summed over all waterline panels, FWL (t) , is calculated
as follows:
2.2. Radiation, diffraction, and Froude-Krylov forces ⎯→

2
pWL (t ) →
s × (→
n ×→ s)
FWL (t ) = ∑ → →
WL Panels 2 ρ g ( n × s )3 (14)
Regarding the radiation forces, the impulse response functions,
K (t ) , are calculated using hydrodynamic added mass and damping A waterline panel (WL_Panel) is a triangular body panel having two
coefficients computed with a Rankine source frequency domain of its three corners at the mean waterline, →
s is the vector between these
method, (Riesner et al., 2016). Here, only the impulse response func- two corners, and index 3 designates the third component (vertical) of
tions from damping are needed and calculated as follows: the resulting vector. In a linear frequency domain method, p wl (t ) is the

292
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

Table 1
Main particulars of subject containership (DTC) and tanker (KVLCC2).
Description Symbol Unit DTC KVLCC2

Length between perpendiculars Lpp [m] 355 320


Moulded breadth B [m] 51 58
Draft D [m] 14.5 20.8
Displacement Δ [m3] 173467 312622
Block coefficient CB [−] 0.661 0.81
Vertical centre of gravity KG [m] 19.85 18.60
Metacentric height GM [m] 5.1 5.71
Radius of gyration about x rxx [m] 20.3 23.2
Radius of gyration about y ryy [m] 87.3 80.0
Radius of gyration about z rzz [m] 87.4 80.0
Froude Number Fn [−] 0.139 0.142
Model scale factor ϱ [−] 63.65 100

in (16) need to be stored for every waterline panel, whereas (1) already
Fig. 1. Ship section with mean water level and instantiations wave elevation includes the radiation response of the complete ship. Our sample cal-
(Boese, 1970). culations required about 200 waterline panels. The computational time
(using one core of a standard personal computer) for simulations of the
sum of the pressure contributions from radiation (including restoring), freely moving ship was approximately 200s. For a time step of 0.05s,
diffraction, and Froude-Krylov forces. However, our weakly nonlinear the computational time for 20 time steps was about one second.
time domain method already integrates Froude-Krylov and restoring
force pressures over the instantaneous wetted surface. Therefore, we 3. Results
must only include coupling effects between these pressures contribu-
tions, and this leads to the following expression: We compared results from our time domain Rankine source
2 2 2
boundary element method (TDIR) to experimental measurements (EFD)
pWL (t ) = pRad (t ) + pDif (t ) + 2 pRad (t ) pDif (t ) + 2 pRad (t ) pFk + R (t ) and to numerical results from a Reynolds-Averaged Navier Stokes
+ 2 pDif (t ) pFk + R (t ) Equations (RANSE) solver (CFD) and a linear frequency domain method
(15)
(GLR). We focused on wave added resistance in regular head waves of a
The first two terms on the right hand side of (15) result directly from 14,000TEU container ship, here referred to as the Duisburg Test Case
radiation and diffraction and, had linear Froude-Krylov forces been (DTC) (el Moctar et al., 2012), and a very large tanker, here referred to
used (instead of the nonlinear Froude-Krylov forces integrated over the as the KVLCC2 tanker. Table 1 lists their main particulars and forward
instantaneous wetted surface), (15) would also include a term pF2+ R (t ) . speeds.
The last three terms on the right hand side of (15) account for coupling Benchmark model test measurements for the DTC containership
effects between pressures caused by radiation, diffraction, and Froude- were carried out at MARINTEK within the framework of the EU funded
Krylov effects. Pressure pDif (t ) is calculated from the frequency domain project SHOPERA (Papanikolaou et al., 2015, Sprenger et al., 2017).
results using (11) applied over every waterline panel. Pressure pFk + R (t ) The model test results for KVLCC2 were taken from Sadat-Hosseini et al.
is the Froude-Krylov and radiation pressure. Pressure pRad (t ) is calcu- (2013). CFD results were taken from Sigmund and el Moctar (2018).
lated using convolution integrals in a manner similar to the calculation Their CFD simulations were performed using the RANSE solver Open-
of total radiation forces acting on the ship; see (1). Accounting for FOAM and Comet. All comparisons were made for model scale.
contributions from all six rigid body motions, pRad (t ) is calculated as First, we compared time histories of longitudinal hydrodynamic
follows: forces obtained from our TDIR method and the CFD method. Although
∼ ∼ t ∼ the velocity potential of our TDIR method accounted for wave elevation
⎯→
⎯ ⎯→
⎯ ⎯→

pRad (t ) =− A (∞)·→
x¨ (t )− B (∞)·→
x˙ (t )− ∫ K (t − τ )·→x˙ (t ) dτ and the dynamic trim and sinkage obtained from a previous solution of
−∞ (16) the stationary flow problem, the total hydrodynamic force in time do-
main did not include calm water resistance. Therefore, to compare time
The radiation force in (1) acts over the entire ship, whereas (16)
histories from our TDIR to time histories from the CFD method, calm
stands for the radiation pressure acting at a single water line panel.

⎯→
⎯ ∼
⎯→
⎯ water resistance had to be subtracted from the total hydrodynamic
Therefore, A (∞)⋅→ x¨ and B (∞)⋅→ x˙ are here pressure-induced force force obtained from the CFD method.

⎯→

components of added mass and damping and K (t ) accounts for the In general, hydrodynamic force components acting on a ship in
time history of the pressure caused by only an impulse on the ship. Here regular head waves depend on the relation between wave length, λ, and

⎯→
⎯ ∼
⎯→
⎯ ∼
⎯→
⎯ ship length, L. Therefore, we compared time histories obtained in a long
A (∞) , B (∞) , and K (t ) are vectors and multiplied by the velocity

⎯→
⎯ ∼
⎯→
⎯ wave, in a wave of length close to the ship's length, and in a short wave.
and acceleration vector using the scalar product. A (∞) and B (∞) For greater clarity, the Froude-Krylov, radiation, diffraction and wa-
yield from terline forces were plotted separately. Then the time average drift force
∼ (wave added resistance) obtained from TDIR, CFD, EFD and GLR
Ai (∞) = we−2 Re ( 
pi1 (∞))
methods were compared. The associated normalized wave drift force,

Bi (∞) = we−1 Im ( 
pi1 (∞)) (17) here known as the wave added resistance coefficient, CAW(ω), was ex-
pressed as follows:
where index i stands for the rigid body motion components. The im-

→ RAW (ω) Lpp
pulse response function K(t) is calculated similarly to (10). The memory CAW (ω) =
ρgB2ζ 2 (18)
storage necessary to calculate pressures acting on a waterline panel is
six times less than the memory storage to calculate the radiation forces where RAW is the wave added resistance, g the gravity constant, ρ is the
and moments acting on the hull using (1). Nevertheless, the coefficients fluid density, and ζ is the wave amplitude. The time averaged values

293
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

were calculated based on a Fourier analysis. The transition phase (first between a positive peak and a negative peak than the corresponding
periods of the simulation) was not considered in the analysis. time span between a negative and a positive peak.
Our method considered the changing undisturbed wetted surface The force component caused by the diffraction pressure acting on
caused by incident waves and the ship motions. To separate effects from the non-average wetted surface plus its coupling effects with the
radiation and diffraction, we compared time histories of forces on the Froude-Krylov pressure (WL) is three times smaller than the diffraction
fixed ships first and, afterwards, forces on the ships moving freely in force, and its main period equals half of the wave encounter period
heave and pitch. For a fixed ship, radiation forces are zero. For the ship (second order effect). However, its time average value of −8.99N is
being free in heave and pitch, first harmonic motion amplitudes were negative, and its magnitude is almost twice as large as the magnitude of
compared. The dimensionless amplitude of heave z͠ is the time average F-K force of −4.77N. Consequently, although the
waterline force did not strongly influence the amplitude of the total
∼ Z
Z = hydrodynamic force, it needed to be considered for the calculation of
ζ (19)
drift forces.
Where Z is the corresponding heave first harmonic amplitude. The The sum of all forces is the total hydrodynamic force (Total) plotted
dimensionless amplitude of pitch θ͠ is in the lower graph in Fig. 2. The phase shift between the nonlinear
Froude-Krylov force (F-K) and the diffraction force (Dif) caused its
θ
θ͠ = amplitude to be slightly smaller than the amplitude of the F-K force,
kζ (20)
which agrees favorably to the force obtained from the CFD method. The
where k is the wave number. amplitude of their first harmonics is nearly equal. From our TDIR
method, this amplitude is 211.3N; from the CFD method, 219.34N.
3.1. Force components for the fixed DTC containership However, their time average forces differ. From our TDIR method, this
time average force is −13.76N; from the CFD method, −19.70N.
For the fixed containership (DTC) in regular head waves of length For the fixed containership (DTC) in regular head waves of length
corresponding to λ/L = 2.5 and an amplitude of ζ = 6.365m (full corresponding to λ/L = 1.09 and an amplitude of ζ = 5.60m (full
scale), the upper graph in Fig. 2 plots time histories of the nonlinear scale), the upper graph in Fig. 3 plots time histories of the nonlinear
Froude-Krylov force (F-K); the center graph, time histories of the dif- Froude-Krylov force (F-K); the center graph, time histories of the dif-
fraction force (Dif) and its nonlinear contribution (WL). The lower fraction force (Dif) and its nonlinear contribution (WL). Similar to the
graph compares the total hydrodynamic force obtained from our TDIR time histories plots in waves of λ/L = 2.50, the Froude-Krylov force (F-
method (Total) and the force from the CFD method (CFD). K) is again non-harmonic. However, in this case the time span between
The amplitude of the nonlinear F-K force is about six times higher a positive and a negative peak is shorter than the time span between a
than the amplitude of the diffraction force (Dif), and the phase shift is negative and a positive peak.
about 180deg. Furthermore, the diffraction force is harmonic (only The difference between the first order harmonic amplitude of the
linear diffraction pressure acted on the non-moving ship hull), but the Froude-Krylov force (F-K) and the diffraction force (Dif) is smaller than
nonlinear F-K force is non-harmonic, indicated by the longer time span for the case of λ/L = 2.50 (The amplitude of the F-K force is about twice

Fig. 2. Time histories of longitudinal forces for the fixed DTC containership Fig. 3. Time histories of longitudinal forces for the fixed DTC containership
(Fn = 0.139 ) in waves of λ / L = 2.50 . “F-K” denotes Froude-Krylov forces, “WL” (Fn = 0.139) in waves of λ / L = 1.09 . “F-K” denotes Froude-Krylov forces, “WL”
denotes nonlinear waterline forces, “Dif” denots diffraction forces and “Total” denotes nonlinear waterline forces, “Dif” denots diffraction forces and “Total”
denotes the total hydrodynamic force. These forces were computed using the denotes the total hydrodynamic force. These forces were computed using the
nonlinear time domain method TDIR. nonlinear time domain method TDIR.

294
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

Fig. 5. Wave added resistance coefficient for the fixed DTC containership
(Fn = 0.139).

3.2. Wave added resistance for the fixed DTC containership

Overall, our computed time histories (TDIR) agreed favorably to


those obtained from the CFD method (CFD). However, sometimes the
time average forces from these two methods differed significantly. To
gain additional insight, Fig. 5 plots comparative wave added resistance
coefficient (CAW), i.e., the normalized quadratic transfer function of
wave drift forces for the fixed DTC containership in regular head waves
of differing lengths, corresponding to λ/L ratios ranging between 0.22
and 2.50.
Generally, seakeeping methods based on potential theory under-
Fig. 4. Time histories of longitudinal total force for the fixed DTC containership estimate wave added resistance in short waves, see e.g. Lyu and el
(Fn = 0.139 ) in waves of λ / L = 0.60 . “Total” denotes the total hydrodynamic Moctar et al. (2017). To more easily focus on short waves, we plotted
force. These forces were computed using the nonlinear time domain method the wave added resistance coefficient against not only the wave length
TDIR. ratio λ/L, but also the square root of its inverse, (L/ λ )0.5. Together with
wave added resistance coefficients from our TDIR method (TDIR) and
as large). The time average of the F-K force is −3.33N. The first order the CFD method (CFD), Fig. 5 plots also comparable coefficients from
harmonic amplitude of the nonlinear water line force is still small the linear frequency domain potential method (GLR). As for the case of
compared to the F-K and Dif amplitudes. However, its time average of λ/L = 2.50, coefficients from our TDIR method (TDIR) and the CFD
−8.96N is the largest contribution to the total drift force. method (CFD) differ in long waves, but this difference decreases with
The time history of total hydrodynamic force from our TDIR method decreasing wave length until it increases again in short waves. In long
(Total) compared favorably to the time history from the CFD method waves, coefficients from the potential method (GLR) behave similarly.
(CFD). The first order harmonic amplitude from our TDIR method is However, with decreasing wave length the coefficient from the linear
166.23N; from the CFD method, 177.41N. Furthermore, the time potential method (GLR) decreases markedly, whereas the coefficients
average for this case, i.e., for waves of λ/L = 1.09 agrees more favor- from our TDIR method (TDIR) and the CFD method (CFD) remain ap-
ably to the case of λ/L = 2.50. Our TDIR computed a time average of proximately constant over the whole range of wave lengths, indicating
−12.30N is close to the CFD computed time average of −14.78N. that our TDIR method (TDIR) constituted a significant improvement for
For the fixed containership (DTC) in regular head waves of length the calculation of the wave added resistance coefficient in waves of
corresponding to λ/L = 0.60 and an amplitude of ζ = 3.775m (full almost every wave length, except in very long waves.
scale), the overall behavior of hydrodynamic force components did not
change much compared to the behavior of in waves of λ/L ratios dis- 3.3. Motions for the freely floating DTC containership
cussed above, see Fig. 4.
Compared to the previous case of λ/L = 1.09, the difference be- Generally, our TDIR method reliably predicted ship motions for a
tween amplitudes of the first order harmonic of the Froude-Krylov force freely moving ship. However, the CFD simulations were conducted for
(F-K) and the diffraction force (Dif) increased again, i.e., F-K amplitudes the DTC containership fixed in surge only and free to move in heave and
were about four times higher than Dif amplitudes. The time average of pitch. In contrast to the totally fixed ship, hydrodynamic forces then
the Froude-Krylov force was −1.20N; the time average of the water line included radiation components caused by the ship's heave and pitch
force, −4.41N. motions. For the cases we considered, Table 2 lists the associated wave
For this case of λ/L = 0.60, the total hydrodynamic force from our length via ratio λ/L, wave amplitude ζ, and wave steepness via ratio H/
TDIR method (Total) compares slightly less favorably to this force from λ, where H is wave height (2ζ ) . The wave conditions were selected
the CFD method (CFD) than for the cases of λ/L = 2.50 and λ/L = 1.09. based on the physical model tests, Sprenger et al. (2017).
As seen in Fig. 4, the first harmonic amplitudes of 41.02N from our Depending on the relation of wave length and ship length, the wave
TDIR method (Total) and of 38.17N from the CFD method (CFD) are added resistance was in some cases strongly affected by radiation
still similar, and their respective time averages values of −5.69N from forces. To calculate the wave added resistance more precisely, an ac-
our TDIR method and −6.40N from the CFD method (CFD) correlate curate computation of ship motions is important (Salvesen, 1978).
more closely than for the case of λ/L = 2.50. Therefore, we first calculated and compared ship motions with ex-
perimental results, Riesner et al. (2016). Against wave length to ship
length ratio, λ/L, and the square root of its inverse, (L/ λ )0.5 , Fig. 6 and
Fig. 7 plot transfer functions of normalized heave (see. equation (19))
and pitch motions (see. equation (20)), respectively, obtained from our

295
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

Table 2
Wave length ratio, wave amplitude, and wave steepness for the con-
sidered cases.

λ/L ζ H /λ

2.50 5.902 0.011


1.80 5.902 0.016
1.40 5.902 0.02
1.20 4.074 0.019
1.09 4.481 0.023
1.00 4.710 0.027
0.91 3.800 0.024
0.80 3.972 0.028
0.60 3.004 0.028
0.44 4.551 0.058
Fig. 8. Phase angle between incident wave and heave motion of the DTC
0.36 3.119 0.049
containership (Fn = 0.139 ).
0.28 2.287 0.048

Pitch motions obtained from all three methods agree favorably.


Compared to the results of Riesner et al. (2016), heave and pitch
transfer function from our TDIR method did not change significantly by
including nonlinear radiation and diffraction force components. Ship
motions were only slightly affected because we assumed that ampli-
tudes of waterline force are small compared to amplitudes of nonlinear
Froude-Krylov forces and linear radiation and diffraction forces. How-
ever, we also assumed that the contribution of the waterline force
dominates the wave added resistance also for the freely moving ship.
The comparative time histories of all force components confirmed our
assumptions.
According to Journee (2001) and Kashiwagi (1995, 2009, and
2011), wave bow interactions are important for an accurate calculation
of wave added resistance. Relative motions of a ship's bow and the free
surface result from coupled heave and pitch motions and incident
waves. To investigate this relationship, Fig. 8 and Fig. 9 plot phase
angles between incident waves and heave and pitch motions, respec-
tively, again against wave length to ship length ratio, λ/L, and the
square root of its inverse, (L/ λ )0.5 , obtained from our nonlinear TDIR
Fig. 6. Heave transfer functions for the DTC containership (Fn = 0.139 ), EFD
method (TDIR), form the CFD method (CFD), from the linear frequency
denotes experiments. domain method (GLR). The incident wave elevation was calculated
with respect to the ship's center of gravity. A positive heave motion was
defined as a vertical upward motion and a positive pitch motion was
defined to be bow down. A negative phase angle means that the motion
follows the wave. Fig. 8 shows that in long waves heave is in phase with
the incident wave (φ ≈ 0) and that the phase angle increases with de-
creasing wave length. In Fig. 9 the phase angle between incident wave
and the pitch shows that in long waves the phase angle converges to-
wards −90deg and decreases with decreasing wave length. Overall, our
TDIR phase angles agree favorably with CFD phase angles. Only for the
heave phase angle in waves of (L/ λ )0.5 ≈ 1.1 differences are noticeable.
Compared to GLR predictions, our TDIR phase angles are closer to the
CFD phase angles for most waves. The GLR phase angles differ most in

Fig. 7. Pitch transfer functions for the DTC containership (Fn = 0.139 ), EFD
denotes experiments.

nonlinear TDIR method (TDIR), form the CFD method (CFD), from
linear frequency domain method (GLR), and from Riesner et al. (2016)
(EFD).
In general, the agreement was favorable. Our TDIR method under-
estimated heave motions slightly in waves of 1.1 ≥ (L/ λ )0.5 ≥ 1.4 ; the Fig. 9. Phase angle between incident wave and pitch motion of the DTC con-
GLR method overestimated heave motions in waves of (L/ λ )0.5 ≥ 1.0 . tainership (Fn = 0.139).

296
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

Fig. 10. Time histories of longitudinal forces for the moving DTC containership Fig. 11. Time histories of longitudinal forces for the moving DTC containership
(Fn = 0.139 ) in waves of λ / L = 2.50 . “F-K” denotes Froude-Krylov forces, “WL” (Fn = 0.139) in waves of λ / L = 1.00 . “F-K” denotes Froude-Krylov forces, “WL”
denotes nonlinear waterline forces, “Dif” denots diffraction forces and “Total” denotes nonlinear waterline forces, “Dif” denots diffraction forces and “Total”
denotes the total hydrodynamic force. These forces were computed using the denotes the total hydrodynamic force. These forces were computed using the
nonlinear time domain method TDIR. nonlinear time domain method TDIR.

L 0.5
waves of 1.0 ≤ ()
λ
≤ 1.2 . Fig. 11 compares force components of our computed time histories
for the freely moving containership in regular head waves of λ/
L = 1.00. Usually wave added resistance is greatest in such wave
3.4. Force components for the freely moving containership lengths because of the phase shift between the incident wave and re-
latively large ship motions; see Figs. 8 and 9. The wave hit the bow of
Fig. 10 compares force components of our computed time histories the ship as it is pointing downwards. Consequently, a large and long-
for the freely moving containership in regular head waves of λ / L = 2.5. lasting restoring force (negative) was generated, as seen by the time
In contrast to the fixed ship, the Froude-Krylov force (F-K) oscillates history of the Froude-Krylov (F-K). The negative part of this Froude-
harmonically about its time average of − 3.02N . Its first harmonic am- Krylov force lasts relatively long compared to its positive part. Conse-
plitude (198N) was 157 times higher than its second harmonic ampli- quently, the time average magnitude of the Froude-Krylov force of
tude (1.26N). − 23.13N is relatively large.
The first harmonic amplitude of the diffraction force (30.71N) and The time histories of the radiation (Rad), diffraction (Dif), and
the first harmonic amplitude of the radiation force (40.83N) were waterline (WL) forces revealed that higher order effects declined
smaller than the first harmonic amplitude of the Froude-Krylov force. compared to the previous case of λ/L = 2.50. The first and second
However, their second harmonic amplitudes were higher, being 11.65N harmonic amplitudes of the radiation force amounted to128.8N and
for the diffraction force and 16.38N for the radiation force. These time 6.57N, respectively, and its time average value was −6.15N. The first
histories were influenced by the second harmonic amplitudes of ra- and second harmonic amplitudes of the diffraction force were 89.3N
diation and diffraction forces. Consequently, the time average of the and 6.07N, respectively, and its time average value was −1.39N. The
radiation force (7.82N) and diffraction force (−6.73 N) were higher. first, second and third harmonic amplitudes of the waterline force were
The time average of the radiation force was positive in this long wave of 2.73N, 22.11N, and 1.93N, respectively, and its time average value was
λ/L = 2.5. −2.99N. Consequently, the Froude-Krylov force contributed most to
The first harmonic amplitude of the waterline force was small the time average drift force; however, the other force components had
(0.646N). However, the second harmonic (5.82N) and the third har- to be included.
monic (0.921N) amplitudes were considerable larger, demonstrating Compared to the previous case of λ/L = 2.50, the total hydro-
that the coupled waterline force (WL) accounted for higher order ef- dynamic force from our TDIR method (Total) agreed slightly less fa-
fects. Although the time average of the waterline force (−1.47N) vorably to this force obtained from CFD method. A small oscillation at a
contributed to total wave added resistance, its contribution was the frequency of 4ωe characterizes this latter time history. Our TDIR method
smallest of all other force components. Our TDIR computed total hy- was not able to account for these high frequency oscillations, because
drodynamic force (Total) compared favorably to comparable CFD pre- the relative motion between the ship's bow and the free surface was
dictions (CFD), whereas our associated time average TDIR force excessive for this case of λ / L = 1.0 . Although accounted for in our TDIR
(−3.40N) differed somewhat from the time average CFD force method, this case caused too large changes of the wetted surface.
(−2.75N). However, relative to their first harmonic amplitudes, the Regarding diffraction and radiation, our waterline force accounted for
differences was less than one percent. relative motions between ship hull and free surface. However, our

297
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

Fig. 13. Wave added resistance coefficient for the heaving and pitching DTC
containership (Fn = 0.139 ), EFD denotes experiments.

shear forces for the container ship contributed about six percent.
However, this does not fully explain the differences between TDIR and
CFD. A simplified approach to account for shear force empirically is
presented below.
Fig. 13 plots comparative values of wave added resistance coeffi-
Fig. 12. Time histories of longitudinal forces for the moving DTC containership cient, CAW, obtained from our TDIR method (TDIR), from the CFD
(Fn = 0.139 ) in waves of λ / L = 0.60 . “F-K” denotes Froude-Krylov forces, “WL” method (CFD), from the linear frequency domain method (GLR), and
denotes nonlinear waterline forces, “Dif” denots diffraction forces and “Total” from model test measurements (EFD). These coefficients, associated
denotes the total hydrodynamic force. These forces were computed using the with the normalized wave drift force, are plotted against the wave
nonlinear time domain method TDIR.
length to ship length ratio, λ/L, and the square root of its inverse,
(L/ λ )0.5 . In long waves, coefficients from our TDIR method and from the
approach did not consider angular changes of the hull normal vectors. GLR method compare favorably to those obtained from the CFD method
Our approach assumed that relative motions between ship and free and from measurements. However, in short waves coefficients from our
surface are small. However, here the relative motions were large, which TDIR method and from the GLR method underestimated values based
could have caused our TDIR and the CFD predictions to differ by a small on model test measurements. Nevertheless, our TDIR computed coef-
amount, and their time averages were similar. Our TDIR method ficients agree more closely to measurement based values than the GLR
yielded a time average force of −33.67N; the and CFD method, computed coefficients. Recall that the comparative predictions for the
−30.75N. In contrast to the previous case of λ / L = 2.5, where the fixed containership were similar.
diffraction and radiation forces contributed most to the time average
force, here the Froude-Krylov force provided the largest contribution. 3.5. Force components for the fixed KVLCC2 tanker
The time average value of the waterline force was twice as large as for
the case of λ / L = 2.5. However, its contribution was small compared to For the fixed tanker in regular 6.0m high head waves of λ/L = 2.0,
the Froude-Krylov force. 1.0, and 0.6, Fig. 14 plots time histories of longitudinal force. The
Fig. 12 compares force components of our computed time histories general behavior of each force component did not change significantly
for the freely moving containership in regular head waves of λ/ compared to those for the DTC containership. Therefore, we present
L = 0.60. The relatively small ship motions in waves of this length only our computed total force (Total) and the CFD force (CFD).
caused second harmonic amplitudes of Froude-Krylov, radiation, and In waves of λ/L = 2.0, first harmonic amplitudes of total force from
diffraction forces to be reduced relative to their first harmonic ampli- our TDIR method (59.73N) and from the CFD method (58.54N) are
tudes. similar; however, their respective time average forces of −1.27N and
The first harmonic amplitude of the Froude-Krylov force amounted −2.13 differed. In waves of λ/ L= 1.0 , agreement is seen to be less
to 40.62N; its second harmonic amplitude, 1.00N; its the time average, favorable. Even comparative first harmonic amplitudes differ. Our TDIR
−2.92 N. The first and second harmonic amplitudes of the diffraction method yielded 31.02 N; the CFD method, 39.65 N. Their respective
force were 9.59N and 0.406N, respectively; its time average, 0.371N. time average forces turned out to be −1.71N and −2.32N. In waves of
The first, second, and third harmonic amplitudes of the waterline force λ/ L= 0.6, predictions from our TDIR method agree more closely to
were 0.285N, 5.32N, and 0.0696 N; its time average, −2.93N. predictions from the CFD method. The first harmonic amplitude from
Consequently, the waterline force contributed about the same amount our TDIR method was 39.11N; from the CFD method, 39.88N. Their
to the total drift force as the Froude-Krylov force. respective time average forces were −2.06N and −2.07N.
Our total hydrodynamic force (Total) agreed favorably to the CFD
predictions (CFD). Their amplitudes are almost identical; however, 3.6. Wave added resistance for the fixed KVLCC2 tanker
their time histories differ somewhat, which we were unable explain.
Their time averages form our TDIR method and the CFD method were Fig. 15 plots comparative time histories of the wave added re-
−5.48N and of −6.81N, respectively. Sigmund and el Moctar (2018) sistance coefficient (CAW) for the fixed tanker (KVLCC2) in regular
showed that viscous surface shear forces may contribute considerably head waves of differing lengths, corresponding to λ/L ratios ranging
to wave added resistance in short waves. In waves of λ/ L= 0.6, such between 0.2 and 2.0. Compared to CFD predictions (CFD), our TDIR

298
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

Fig. 16. Heave transfer functions for the KVLCC2 tanker (Fn = 0.142 ), EFD
denotes experiments.

Fig. 14. Time histories of longitudinal forces for the fixed KVLCC2 tanker
(Fn = 0.142 ) in waves of λ / L = 2.0,  1.0,  and 0.6 . “Total” denotes the total
hydrodynamic force. These forces were computed using the nonlinear time
domain method TDIR.

Fig. 17. Pitch transfer functions for the KVLCC2 tanker (Fn = 0.142 ), EFD de-
notes experiments.

method (TDIR), form the CFD method (CFD), from a linear frequency
domain method (GLR), and from model test measurements (EFD).
Fig. 16 shows that in waves of 0.9 < (L/ λ )0.5 < 1.1, our TDIR method
(TDIR) underestimates heave. In these wave lengths, deck immersion
most likely affected heave. In long waves, heave predicted by the CFD
Fig. 15. Wave added resistance coefficient for the fixed KVLCC 2 tanker method (CFD) and the linear method (GLR) correlate favorably to
(Fn = 0.142 ). measurements.
Fig. 17 shows that pitch predictions from our TDIR method (TDIR)
agree favorably to measurements (EFD), especially in long waves,
method (TDIR), the CFD method (CFD), and the frequency domain
whereas predictions from GLR method (GLR) and CFD method (CFD)
method (GLR) underestimated wave added resistance in long waves.
overestimate pitch compared to measurements (EFD).
However, all predictions converge with decreasing wave length. In
Fig. 18 and Fig. 19 plot phase angles between the incident wave
short waves, coefficients from our TDIR method and from the CFD
and, respectively, the tanker's heave motions and pitch motions. These
method agree well, whereas coefficients from the GLR method still
phase angles behaved similarly to those for the containership.
underestimates wave added resistance.
Fig. 18 shows that, except in waves of (L/ λ )0.5 ≈ 0.9, heave from our
TDIR method (TDIR) generally agree favorably to heave from the CFD
3.7. Motions for the freely floating KVLCC2 tanker method (CFD) and to measured heave (EFD). In waves of (L/ λ )0.5 ≥ 0.9,
heave from the linear GLR method (GLR) overestimated measured
We performed simulations of the tanker free to move in heave and heave.
pitch in regular 3.0m amplitude head waves of 0.4 ≤ λ / L ≤ 2.0 . Against Fig. 19 shows that pitch behaved similarly to heave in that pitch
wave length to ship length ratio, λ/L, and the square root of its inverse, from our TDIR method (TDIR) generally agrees favorably to pitch from
(L/ λ )0.5 , Fig. 16 and Fig. 17 plot transfer functions of normalized heave the CFD method (CFD) and to measured pitch (EFD). As a conclusion,
and pitch motions, respectively, obtained from our nonlinear TDIR our weakly nonlinear TDIR method improved the prediction of motion

299
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

Fig. 18. Phase angle between incident wave and heave motion of the KVLCC2
tanker (Fn = 0.142 ), EFD denotes experiments.

Fig. 19. Phase angle between incident wave and pitch motion of the KVLCC2
tanker (Fn = 0.142 ), EFD denotes experiment.
Fig. 20. Time histories of longitudinal forces for the moving KVLCC2 tanker
phase angles compared to the linear frequency domain GLR method. (Fn = 0.142 ) in waves of λ / L = 2.0,  1.0,  and 0.6 . “Total” denotes the total
For ship motions, Froude-Krylov forces often dominated. To obtain the hydrodynamic force. These forces were computed using the nonlinear time
Froude-Krylov forces, recall that our TDIR method integrated over the domain method TDIR.
exact wetted surface, which contributed significantly to more accurate
predictions.
Fig. 20 plots time histories of longitudinal total force in waves of
λ / L = 2.0,  1.0, and 0.6. Similar to the containership simulations, in
long waves of λ/L = 2.0 our TDIR predictions (TDIR) agree favorably to
CFD predictions (CFD). Their respective first harmonic amplitudes of
53.93N and 56.41N were close. Similar could be shown for their time
average forces of −0.40N and −0.37N. However, these average forces
were small compared to the total force amplitudes.
In waves of λ / L = 1.0 , first harmonic amplitudes from our TDIR
method and the CFD method agree well. However, in waves of this
length, the bow of the tanker emerged and immersed largely. Recall
that our TDIR method was then unable to compute the flow around the
bow, thereby causing a small time shift of the negative part and also a
deviation from the harmonic behavior at times t = 0.5 and 1.5s.
Nevertheless, overall the agreement was accurate enough to compute
first harmonic amplitudes. The time average force of −6.4N from our
TDIR method was slightly overestimated compared to the average force
of −5.18N from the CFD method.
In waves of λ/ L= 0.6, first harmonic amplitudes from our TDIR and Fig. 21. Wave added resistance coefficient for the moving KVLCC2 tanker
CFD are seen to be similar. Although the small time shift of the negative (Fn = 0.142 ), EFD denotes experiments.
part was still present, their respective time average forces of −2.09N
and −2.11N were close together.
(L/ λ )0.5 . Over almost the entire range of wave lengths, coefficients from
Fig. 21 plots comparative values of wave added resistance coeffi-
our TDIR method compare favorably to those obtained from measure-
cient, CAW, obtained from our TDIR method (TDIR), from the CFD
ments. Coefficients from the GLR method compared favorably to mea-
method (CFD), from the linear frequency domain method (GLR), and
surement based coefficients only in long waves and in short waves. In
from model test measurements (EFD). These coefficients, associated
waves of 1.2 ≥ λ / L ≥ 0.6, the GLR method underestimated the wave
with the normalized wave drift force, are plotted against the wave
added resistance.
length to ship length ratio, λ/L, and the square root of its inverse,

300
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

3.8. Effect of wave steepness on wave added resistance resistance for three different wave steepness. The effects of wave
steepness were found to be moderate. Similar effects were shown by
The effect of different wave heights on the wave added resistance Yasukawa et al. (2016).
was investigated experimentally by Yasukawa et al. (2016). They The weakly nonlinear time domain solver is able to account for
measured the quadratic transfer function of the wave added resistance geometrical nonlinearities. These nonlinearities may affect the non-di-
for three ship models with the same main dimensions and different mensional wave added resistance slightly when the wave steepness
block coefficients (0.81, 0.84 and 0.87). The experiments were per- varies (see. e.g. Fig. 22). However, the major effect of the wave
formed for two different wave amplitudes. Further on, two loading steepness on the added resistance seemed to result from the nonlinear
conditions, full loading and ballast conditions, were investigated. For behavior of steep waves. This effect is not considered in the BEM
the full loading conditions, the authors shown that the quadratic method TDIR. Consequently, the effect of wave steepness may be
transfer function of the wave added resistance was clearly effected by clearly underestimated by the BEM method as shown by Yasukawa
the wave amplitude. For ballast conditions, effects of wave amplitude et al. (2016) and Sigmund and el Moctar (2018).
on the quadratic transfer function of the wave added resistance could
only be seen for few single wave frequencies. 3.9. Estimation of the viscous wave added resistance
Sigmund and el Moctar (2018) investigated the effect of the wave
steepness on the quadratic transfer function of the wave added re- Generally, it is assumed that wave drift forces are caused by pres-
sistance for the containership (DTC) and the tanker (KVLCC2) using sure variations at the ship's hull. However, Sigmund and el Moctar
CFD. They showed that the quadratic transfer function of the wave (2018) showed that also viscous effects play a significant role in short
added resistance of the tanker was sensitive to the wave steepness. In waves of λ / L ≤ 0.5, and that it vanishes in longer waves when the ship
long and mid-size waves the non-dimensional wave added resistance is free to heave and pitch. Further work regarding the wave added re-
increased with decreasing wave steepness. Similar behavior was shown sistance of the KVLCC2 tanker in short waves can be found in Yang et al.
by Yasukawa et al. (2016). As aforementioned, Yasukawa et al. (2016) (2018) and Seo et al. (2014). So far, our time domain method based on
showed that in short waves the quadratic transfer function of the wave potential theory neglected viscous effects. However, to remedy this
added resistance of the VLCC tanker with a block coefficient of 0.81 defect, we introduced a simplified exponential function to account for
seemed to be only slightly effected by the wave steepness and a de- viscous effects on the wave added resistance in head waves. Thus, the
creasing wave steepness resulted in an increasing non-dimensional total wave added resistance coefficient, CAW ,  Tot (ω) , was calculated as
wave added resistance. However, Sigmund and el Moctar (2018) the sum of the wave induced pressure added resistance coefficient,
showed a different behavior in short waves. The non-dimensional wave CAW (ω), and the viscous wave added resistance coefficient, Cvisc (ω) .
added resistance was strongly affected by the wave steepness in short Relating the viscous wave added resistance to the pressure based wave
waves and decreased with decreasing wave steepness. However, in added resistance
contrast to Yasukawa et al. (2016), Sigmund and el Moctar (2018) in-
Cvisc (ω)
vestigated very short waves with a high steepness. This might be a fvisc (ω) = ,
CAW (ω) (21)
reason for the contrary findings.
In the present study we investigated the effect of wave steepness on the total wave added resistance coefficient follows from
the quadratic transfer function of the wave added resistance for the
CAW ,  Tot (ω) = CAW (ω) (1 + fvisc (ω)) (22)
container ship and tanker using the weakly nonlinear time domain
method. The wave added resistance of the container ship was computed With fvisc (ω) being the relative viscous wave added resistance. We
for the wave conditions listed in Table 2. Since we used a potential flow assumed that the relationship between the pressure (wave induced
solver, we tried to avoid water on deck. Thus, the wave steepness had to part) and viscous wave added resistance does not dependent on ship
be limited and only waves with a small to moderate steepness were speed. We expressed fvisc (ω) by an exponential function with the fol-
investigated. As a result, the added resistance of the containership de- lowing parameters: wave length to ship length ratio λ / L and ship's
pended almost quadratically on the wave amplitude for the investigated block coefficient CB:
waves. Slightly more effects were found for the KVLCC2 tanker. Fig. 22
c1 c 3 λ
plots the corresponding quadratic transfer functions of the wave added fvisc = e L
cBc2 (23)
where c1, c2 and c3 are assumed to be ship independent constants. c1, c2
and c3 were determined by applying the least square fit technique to the
results presented by Sigmund and el Moctar (2018).
0.4 −5.8 λ
fvisc = e L
cB1.67 (24)
For the containership (DTC) and the Tanker (KVLCC2), Fig. 23
compares the relative viscous wave added resistance, fvisc , obtained
from our empirical function (24) with results from the CFD method.
Index emp designates values from (24); index CFD, values from the CFD
method. It is seen that the viscous contribution to total added resistance
was higher for the container ship than for the tanker.
Fig. 24 and Fig. 25 plot comparative values of wave added re-
sistance coefficient, CAW, obtained from our TDIR method including
viscous effects (TDIR + fric), from the CFD method (CFD), from the
linear frequency domain method (GLR), and from model test mea-
surements (EFD). For the containership it is seen that the additional
viscous force component yields a better agreement between TDIR and
CFD predictions as well as measurement based (EFD) predictions. For
Fig. 22. Effects of the wave steepness on the quadratic transfer function of the the tanker, differences were smaller. Furthermore, in short waves, our
wave added resistance for KVLCC2 tanker at Fn = 0.142 . TDIR method slightly overestimated wave added resistance compared

301
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

Wave length strongly influenced the viscous wave added resistance


component, especially in shorter waves. Furthermore, ship motions
were mitigated in shorter waves. From that perspective, it seemed
reasonable to relate viscous wave added resistance to relative ship
motions (heave and pitch) instead of wave length. If wave added re-
sistance were to be calculated for the fixed ship, a significant viscous
components should result also in long waves and perhaps explain why
our TDIR predictions of wave added resistance for the freely moving
ship in long waves agreed so favorably to the CFD predictions, but
differenced for the fixed ship. To clarify whether viscous wave added
resistance is relevant also in long waves when the ship is fixed, addi-
tional CFD computations would be necessary.

4. Conclusions

Fig. 23. Percentage viscous wave added resistance for the containership We presented a partially nonlinear time domain numerical method
(Fn = 0.139 ) and the KVLCC2 tanker (Fn = 0.142 ).
to predict the wave added resistance of ships advancing at constant
forward speed in regular head waves. Following the approach of
Cummins, our method relied on frequency domain coefficients to cal-
culate radiation forces in the time domain, on integration of pressures
over the instantaneous wetted surface of the ships to yield nonlinear
Froude-Krylov and restoring forces, and on an additional force com-
ponent to account for the effect of the changing wetted surface caused
by radiation and diffraction of the undisturbed incident waves.
For a modern container ship and very large tanker, we computed
time histories of wave-induced longitudinal forces and quadratic
transfer functions of wave added resistance coefficients and compared
our predictions to results from a RANSE based CFD method, from a
linear frequency domain technique, and model test based measure-
ments. We considered the fixed ships and the ships free to move in
heave and pitch.
Generally, our computed time histories agreed favorably to those
from the CFD method. Only for the fixed tanker in wave lengths close to
the ship's length did first harmonic amplitudes of longitudinal forces
differ by a relatively small amount. For the freely moving ships, time
average wave added resistances were closer to those from the CFD
method and from model test based measurements. For the fixed ships,
Fig. 24. Wave added resistance coefficient including viscous effects for the DTC
our method underestimated wave added resistance also in long waves.
containership (Fn = 0.139), EFD denotes experiments.
In shorter waves, the differences between our method and the CFD
results decreased. Especially for the tanker, agreement in short waves
was favorable.
Compared to the frequency domain approach, our method yielded
more accurate predictions in waves of almost every wave length.
Especially for the fixed containership in short waves, the frequency
domain method strongly underestimated wave added resistance. For
the freely moving ships, our results agreed favorably to those based on
model test measurements over the whole range of wave lengths.
Especially for the containership in short waves and the tanker in mid to
long waves, our method obtained more accurate predictions compared
to the frequency domain method.
To consider the viscous component of wave added resistance, we
introduced an empirical formula. For the containership in short waves,
its inclusion led to a reduction of the differences between prediction
from our method and those from the CFD method and from model test
based wave added resistance. For the tanker, the viscous wave added
resistance played a minor role and the use of the empirical function did
not change the results significantly. However, for the tanker in short
waves our TDIR results and CFD results were almost identical.
Fig. 25. Wave added resistance coefficient including viscous effects for the
tanker (Fn = 0.142 ), EFD denotes experiments.
Acknowledgments
to CFD predictions, even without considering the viscous force con-
tribution. Nevertheless, including the viscous component improved our The research was supported by the German Research Association
predictions compared to measurement based (EFD) predictions. DFG under contract number EL 611/4-1.

302
M. Riesner, O. el Moctar Ocean Engineering 162 (2018) 290–303

References Maruo, H., 1960. The drift force on a body floating in waves. J. Ship Res. 1–10.
Maruo, H., 1963. Resistance in waves. Soc. Nav. Archit. Jpn. 8 60th anniversary Series.
Ogilvie, T., 1964. Recent progress toward the understanding and prediction of ship mo-
Boese, P., 1970. A simple method for the calculation of resistance increase of a ship in a tions. In: Proc. Of the 5th Symposium on Navan Hydrodynamics, Bergen, Norway
seaway. J. Ship Technol. Res. 17 (86). ACR-112. pp. 3–79 ser. ONR.
Cummins, W., 1962. The impulse response function and ship motions. Schiffstechnik 9, Papanikolaou, A., Zaraphonitis, G., Bitner-Gregersen, E., Shigunov, V., el Moctar, O.,
101–109. Guedes Soares, C., Devalapalli, R., Sprenger, F., 2015. Energy efficient safe ship
el Moctar, O., Shigunov, V., Zorn, T., 2012. Duisburg test case: post- panamax container operation. In: Proceedings of the SNAME World Maritime Conference, Providence,
ship for benchmarking. Ship Technol. Res. J. 59/3, 50–64 (Duisburg, Germany). USA.
el Moctar, O., Sigmund, S., Ley, J., Schellin, T.E., 2017. Numerical and experimental Park, D.M., Seo, M.G., Lee, J., Yang, K.Y., Kim, Y., 2014. Systematic experimental and
analysis of added resistance of ships in waves. J. Offshore Mech. Arct. Eng. 139. numerical analyses on added resistance in waves. J. Soc. Nav. Archit. Korea 51 (6),
http://dx.doi.org/10.1115/1.4034205. 459–479.
Faltinsen, O.M., Minsaas, K., Liapis, N., Skjørdal, S.O., 1980. Prediction of resistance and Prpić-Oršić, J., Faltinsen, O.M., 2012. Estimation of ship speed loss and associated CO2
propulsion of a ship in a seaway. In: 13th Symposium on Naval Hydrodynamics, pp. emissions in a seaway. J. Ocean. Eng. 44, 1–10.
505–529. Rathje, H., Schellin, T.E., Brehm, A., 2011. Speed loss in waves and wave-induced torsion
Gerritsma, J., Beukelman, W., 1972. Analysis of the resistance increase in waves of a fast of a wide-breadth containership. Proc. Inst. Mech. Eng. Part M. J. Eng. Marit.
cargo ship. Int. Shipbuild. Prog. 19, 285–293. Environ. 225 (4), 387–401.
Guo, B., Steen, S., Deng, G., 2012. Seakeeping prediction of KVLCC2 in head waves with Riesner, M., von Graefe, A., Shigunov, V., el Moctar, O., 2016. Prediction of non-linear
RANS. Appl. Ocean. Res. 35, 56–67. ship responses in waves considering forward speed effects. Ship Technol. Res. 64,
Hachmann, D., 1991. Calculation of pressures on a Ship's hull in waves. J. Ship Tech. Res. 135–145.
38, 111–133. Sadat-Hosseini, H., Carrica, P., Kim, H., Toda, Y., Stern, F., 2010. URANS Simulation and
IMO, 2011. Amendments to the annex of the protocol of 1997 to amend the international Valiation of Added Resistance and Motions of the KVLCC2 Crude Carrier with Fixed
convention for the prevention of pollution from ships, 1973, as modified by the and Free Surge Conditions. Gothenburg. (A Workshop on CFD in Ship
protocol of 1978 relating thereto. MARPOL Annex. 19., ESOLUTION MEPC 203 (62). Hydrodynamics).
IMO, 2013. Interim Guidelines for Determining Minimum Propulsion Power to Maintain Sadat-Hosseini, H., Wu, P.-C., Carrica, P.M., Kim, H., Toda, Y., Stern, F., 2013. CFD
the Manoeuvrability in Adverse Conditions. International Maritime Organisation verification ad validation of added resistance and motions of KVLCC2 with fixed and
(IMO), London. free surge in short and long head waves. J. Ocean. Eng. 59, 240–273.
IMO, 2015. Amendments to the Interim Guidelines for Determining Minimum Propulsion Salvesen, N., 1978. Added resistance of ships in waves. J. Hydrodyn. 12 (1), 24–34.
Power to Maintain the Manoeuvrability of Ships in Adverse Conditions. International Seo, M.-G., Yang, K.-K., Park, D.-M., Kim, Y., 2014. Numerical analysis of added resistance
Maritime Organisation (IMO), London. on ships in short waves. Ocean. Eng. Vol. 87, 97–110.
Journee, J.M.J., 2001. Verification and Validation of Ship Motion Program SEAWAY. Sigmund, S., el Moctar, O., 2018. Advanced numerical and experimental investigation of
Report1213a. Ship hydromechanics laboratory, Delft University of Technology. added resistance of different ship types in short and long waves. Ocean. Eng. 147 (1),
Kashiwagi, M., 1995. Prediction of Surge and its Effect on Added Resistance by Means of 51–67.
the Enhanced Unified Theory. vol. 89. Transactions of West-Japan Society of Naval Skejic, R., Faltinsen, O.M., 2008. A unified seakeeping and maneuvering analysis of ships
Architects, pp. 77–89. in regular waves. J. Mar. Sci. Technol. 13, 371–394.
Kashiwagi, M., 2009. Impact of hull design on added resistance in waves—application of Söding, H., 2009. Computation of springing transfer functions. Proc. Institution Mech.
the enhanced unified theory. In: Proceedings of the 10th International Marine Design Eng. Part M J. Eng. Marit. Environ 223 (3), 291–304 03/2009.
Conference, Trondheim, Norway, pp. 521–535. Söding, H., von Graefe, A., Shigunov, V., el Moctar, O., 2012. Rankine source method for
Kashiwagi, M., 2011. “Hydrodynamic Study on Added Resistance Using Unsteady Wave seakeeping predictions. In: Proc. 31th Int. Conf. On Ocean, Offshore and Arctic Eng,
Analysis”, 34th Weinblum Memorial Lecture. vol. 2012 OMAE2012–83450.
Kim, K.-H., Joncquez, S., Kim, Y.-H., Bingham, H., 2010. Numerical analysis on added Söding, H., Shigunov, V., Schellin, T., el Moctar, O., 2014. A rankine panel method for
resistance of ships in time-domain. In: Proceedings 25th International Workshop on added resistance of ships in waves. J. Offshore Mech. Arct. Eng 136 (3).
Water Waves and Floating Bodies, (Harbin, China). Sportelli, M., Huijsmans, R., 2012. Added resistance in short waves: a ray theory ap-
Kim, H.T., Kim, J.J., Choi, N.Y., Lee, G.H., 2014. A study on the operating trim, shallow proach. In: 27th International Workshop on Water Waves and Floating Bodies.
water and wave effect. SNAK 631–637. Sprenger, F., Maron, A., Delefortrie, G., Van Zwijnsvoorde, T., Cura-Hochbaum, A.,
Kim, Y.C., Kim, K.S., Kim, J., Kim, Y.S., Van, S.H., Jang, Y.H., 2015. Calculation of added Lengwinat, A., Papanikolaou, A., 2017. Experimental studies on seakeeping and
resistance in waves for KVLCC2 and its modified hull form using RANS-based manoeuvrability in adverse conditions. J. Ship Res. SNAME Publ. http://dx.doi.org/
method. In: Proceedings of the 25th International Offshore and Polar Engineering 10.5957/JOSR.170002.
Conference. Int. Soc. Offshore Polar Eng, Hawaii, USA, pp. 924–930. Valanto, P., Hong, Y.P., 2015. Experimental investigation on ship wave added resistance
Kim, M., Hizir, O., Turan, O., Day, S., Incecik, A., 2016. A study on ship speed loss due to in regular head, oblique, beam and following waves. In: Proceedings of the 25th
added resistance in a seaway. In: Proceedings of the 26th International Ocean and International Ocean and Polar Engineering Conference, (Kona, Big Island, Hawaii,
Polar Engineering Conference. Int. Soc. Offshore Polar Eng, Rhodes, Greece, pp. USA).
527–534. von Graefe, A., 2014. A Rankine Source Method for Ship-ship Interaction and Shallow
Kim, M., Hizir, O., Turan, O., Incecik, A., 2017. Numerical studies on added resistance Water Problems. PhD Thesis. University of Duisburg-Essen March.
and motions of KVLCC2 in head seas for various ship speeds. Ocean. Eng. 466–476. Wheeler, J.D., 1979. Method for calculating forces produced by irregular waves. J. Petrol.
Kwon, Y.J., 2008. Speed loss due to added resistance in wind and waves. Nav. Archit. Technol. 22 (3), 359–367.
14–16. Yang, K.-K., Kim, Y., Jung, Y.-W., 2018. Enhancement of asymptotic formula for added
Ley, J., Sigmund, S., el Moctar, O., 2014. Numerical prediction of the added resistance of resistance of ships in short waves. Ocean. Eng. 148, 211–222.
ships in waves. In: 33rd Int. Conf. On Ocean, Offshore and Arctic Engineering, San Yasukawa, H., Matsumoto, A., Ikezoe, S., 2016. Wave Height Effect on Added Resistance
Francisco, USA. of Full Hull Ships in Waves. The japanese Society of Naval Archtects and Ocean
Liu, S., Papanikolaou, A., Zaraphonites, G., 2011. Prediction of added resistance of ships Engineers, JASNAOE, pp. 45–54.
in waves. Ocean. Eng. 38, 641–650. Zhang, W., Zou, Z.-J., Deng, D.-H., 2017. A study on prediction of ship maneuvering in
Lyu, W., el Moctar, O., 2017. Numerical and experimental investigations of wave-induced regular waves,. Ocean. Eng. 137, 367–381.
second order hydrodynamic loads. Ocean. Eng. 197–212.

303

You might also like