Applied Ocean Research 32 (2010) 471-482

C o n t e n t s lists a v a i l a b l e at S c i e n c e D i r e c t A p p L I e
O C E A N
RESEARCH
Applied Ocean Research
journal homepage: www.elsevier.com/locate/apor
ELSEVIER

Seakeeping computations using double-body basis flows

Xinshu Zhang*, Piotr Bandyk, Robert F. Beck
Department of Naval Architecture and Marine Engineering, University of Michigan, Ann Arbor, MI 48109, United States

A R T I C L E I N F O ABSTRACT

Article history: Three-dimensional, time-domain, ship-wave interactions are studied in this paper for problems with
Received 17 December 2009
forward speed. Free surface boundary conditions are derived based on a double-body linearization and
Received in revised form the mixed Euler-Lagrange time stepping technique. The boundary integral equations are solved at each
15 October 2010 time step by distributing desingularized sources above the calm water surface and employing constant-
Accepted 22 October 2010 strength panels on the body surface.
Available online 25 November 2010 Radiadon, diffraction, and free motion results for a Wigley hull and a Series 60 hull are presented and
systematically compared with the experiments and other numerical solutions using the Neumann-Kelvin
Keywords:
approach with simplified m-terms, linearized free surface boundary conditions with double-body
Seakeeping
m-terms, and the time-domain body-exact strip theory. By comparing the present results to experiments
Desingularized source
Double-body flow
and other numerical solutions, it is found that the present computational model using double-body
m-terms linearization gives improved results. It is also demonstrated that the m-terms are very important to
obtain accurate hydrodynamic coefficients, while the leading-order terms included in the free surface
boundary conditions of the present model can also improve the computational accuracy of the cross-
coupling radiation damping.
© 2010 Elsevier Ltd. All nghts reserved.

continue the computations after wave breaking, but they are still
1. Introduction
not robust and can lead to nonphysical soludons.
In ship and offshore designs, the accurate prediction of A compromise between fully nonlinear computations and
wave-induced motions and loads is very important. It requires linear theory is the so-called body-exact approach. In the body-
I<nowledge of the extreme value of various design parameters exact approach, the body boundary condition is sadsfied on the
such as motion amplitudes, bending moments, wave-induced instantaneous wetted surface of the body while the linearized free
hydrodynamic pressures on the hull, local slamming forces, etc. surface boundary conditions are retained. In order to solve for
A mixed Euler-Lagrange time stepping scheme (MEL) was first the hydrodynamic forces due to large body motions i n the body-
introduced by Longuet-Higgins and Cokelet [1] for solving t w o - exact problem, a time-domain approach is preferred. A method to
dimensional fully nonlinear water wave problems. Since then, deal w i t h the exact body boundary condition using a time-domain
MEL methods have also been successfully used to solve fully free surface Green's function has been developed by Beck and
nonlinear, three-dimensional wave and wave-body interaction Magee [14] for a submerged body performing arbitrary modons.
problems [2-12] either using a boundary element method (BEM) Other researchers such as Lin and Yue [15], Bingham [16] have
or a finite element method (FEM). Recently, Liu et al. [13] also successfully obtained results for a surface-piercing body
used a desingularized boundary element method w i t h a MEL using the dme-domain free surface Green's funcdon method.
formulation to study the nonlinear wave scattering by a submerged Huang and Sclavounos [17] investigated nonlinear ship motion
horizontal plate. The problems w i t h the fully nonlinear MEL problems using the body-exact technique and weak-scatter theory.
computations include numerical instabilities of the free surface Sen [18] and Singh and Sen [19] solved large amphtude free motion
and wave breaking. The instabilides can often be eliminated by problems using the dme-domain free surface Green's funcdon
improved numerical techniques, but wave breaking is a natural while considering the incident wave nonlinearities. Recentiy,
phenomenon that is expected to occur in any large body motion Zhang and Beck [20,21] developed a computationally efficient,
or wave situation. Computations normally are forced to stop when time-domain, two-dimensional body-exact model using Rankine
wave breaking occurs. Various techniques have been proposed to sources to solve large amplitude radiation and diffraction problems
including water entry and exit. Comparisons w i t h other numerical
calculations and experiments were good. Also, Zhang [22], Zhang
and Beck [23] presented a three-dimensional model using the
* Corresponding author. Tel.: +1 2816178779. body-exact technique w i t h desingularized sources above the free
E-mail addresses: xinshuz®umich.edu, [email protected] (X. Zhang),
surface and panels on the body surface. As has been found i n the
[email protected] (P. Bandyk), [email protected] (R.F. Beck).

0141-1187/$ - see front matter ® 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apor.2010.10.003
472 X, Zhang et al./Applied Ocean Research 32 (2010) 471-482

Tig. 1. Definition of the problem and coordinate systems.

previous study, for a wall-sided body, the change of hydrodynamic where Tj(x,y,t) represents the free surface elevation; g is the
force due to the instantaneous wetted body surface is trivial, while gravitational acceleration. All the velocity potentials satisfy the
for a non-wall-sided body the body-exact computation can be Laplace equation under the assumption of ideal potential flow.
critical. The exact body boundary condition can be written as
In this paper, we have continued to develop a three-
dimensional model based on a double-body linearization and the n-V0 = t/o(t)n,-I-VH-II-V0'-n onSg (4)
mixed Euler-Lagrange time stepping technique. The motivation of where SB is the instantaneous wetted body surface; Uo(t) is the
the present study is to formulate and develop a more reliable and time-dependent translating velocity of the body in the x direction;
accurate model to simulate ship motions in waves. The second ob- n is the inward unit normal on the body surface (out of fluid); ni
jective of this study is to identify the effects of free surface nonlin- is the component of the unit normal in the x direction; V H is the
eanties and the coupling effects between the steady and unsteady motion velocity including rotational modes of a point on the ship's
f l o w on the prediction of hydrodynamic coefficients by quantify- surface; 0' is the velocity potential for an incident wave.
ing the differences between the present model and other numeri- By applying Green's theorem, using desingularized sources
cal approaches. It should be noted that the present model is based above the calm water surface and constant-strength panels on the
on the assumption that disturbance due to the presence of ship is hull, the velocity potential can be written as (see [ 2 4 ] )
small, which means that ship induced double-body free surface el-
evation is small. This can be justified in the case of a slender ship.
In the present study, free surface boundary conditions are 0(x) = J2 ^)'^(^) + [ f ?)cr(?)ds (5)
applied on the calm water surface, and the body boundary
condition is applied on the mean wetted hull surface. Linear where G = a is the source strength on the boundary; Sf is
hydrostatic restoring and Froude-Krylov forces are also used in the calm water surface; Nf is the number of desingularized sources
the present simulations. The present model can be extended to above the calm water surface.
study the body-exact problem and the added resistance problem
The disturbance potential (/) must satisfy the far field boundary
by satisfying the body boundary condition on the instantaneous
conditions such that there must be no incoming waves and, in
wetted body surface. Those results w i l l be reported elsewhere.
the deep water problem, Vcf> vanishes as z - > —oo. The initial
conditions at t = 0 can be written as
2. Mathematical formulation
0 in the fluid domain (6)
A boundary value problem for a vessel traveling in deep water After solving the boundary value problem, the pressure on the
is solved. The vessel moves with speed U ( t ) = (Uo{t), 0, 0), body due to the disturbance potential can be computed using
and may be undergoing unsteady oscillations in its six degrees of Bernoulli's equation.
freedom. The fluid is assumed to be ideal and the flow irrotational.
Three coordinate systems w i l l be employed: the Xo system is fixed 'dè d(p 1
in space, the x system is fixed to the mean position of the ship (7)
(moving w i t h forward speed U ( t ) along the straight track of the
ship), and the x system is fixed to the ship. The boundary value It should be noted that the pressure diie to the incident waves w i l l
problem is solved in the right hand moving coordinate system be included in the next section.
(x,y,z), as shown in Fig. 1. The x-axis points in the direction of The force and moment acting on the body can be determined by
travel and the z-axis points upward. The origin is on the calm water using
plane at mid-ship.
In the X coordinate system, a velocity potential is introduced to (8)
describe the fluid motion by using the above assumptions such that ffsr'
the fluid velocity can be expressed as the gradient of a potential
function, V(x, t) = = V ( - ü o ( t ) x -\- (j>ix,y, z, 0). where (j) is M )ds (9)
the disturbance velocity potential which may include the radiation
and/or diffraction potential.
The velocity potential <^(x, y,z,t) satisfies the Laplace equation 2.1. Linearization of tiie free surface boundary conditions and body
(pxx + 4>yy + 4>2z = 0 (1) boundary condition ,

The exact nonlinear kinematic and dynamic free surface

The fully nonlinear free surface boundary conditions ( 2 ) and
boundary conditions are
( 3 ) can be linearized using a double-body basis flow The total
dn d<b dn disturbance velocity potential is decomposed into a double-body
-L = -f-V4>-Vii + Uo{t)-^ onz = n(x,y,t) (2) basis flow and other perturbation flows (/>', which may include
dt dz dx
diffraction and radiation wave components.
dé 1 dè
^ = -g'?--V.A-V<^ + t/o(f)f- onz = i^ix,y,t) (3) 0(x, t) = 'i'ix, t ) + 0'(x, t) (10)
dt 2 dx
 X, Zhang et al./Applied Ocean Research 32 (2010) 471-482 473

It is assumed that i^' ~ 0(1) and 0' 0(e), where e « 1 denoting

a small quantity such as slenderness of the ship.
Desingularized sources
The double-body flow satisfies the Laplace equation in the fluid
domain:

I'xx + ^yy + ^zz = 0 (11)

\
A rigid wall condition is applied on the calm water surface.

0 onz = 0 (12)
Panels on ttie hull surface
w i t h a body boundary condition,

n-V^ = Uo(t)nr onSj (13)

Fig. 2. Desingularized source distribution above calm water and constant-strength
where Sj is the mean wetted body surface. ^ also must satisfy a flat panel discretization on the hull.

far field boundai-y condition such that, in the deep water problem,
V</' vanishes as r = y ^ Ï M - y M - ^ _^ For the free motion problem, the linearized body boundary
condition can be written as
It is assumed here that a slender ship is traveling along a
straight line. Therefore, the ship induced double-body free suiface 9</> d<p'
elevation is 0(e), which is in the same order as 4>' for small on Ss (21)
9 'dn
amplitude ship motion. Furthermore, we can assume the total free
surface elevation ri is 0(e). Substituting (10) into Eqs. (2) and (3), It is assumed that <p' ~ 0(e) so that the higher-order inter-
keeping the leading-order terms and dropping the prime, we get actions between the radiation and diffraction potentials are small
the following linearized free surface boundary conditions, and can be neglected. Adding the pressure due to the incident wave
(j)' into Eq. (7), the total pressure acting on the body is computed
9V using Eq. (22), given in Box I .
Tj onz = 0 (14)
"3? The force and moment on the body can be computed using
dd> 90 1 Eqs. (8) and (9) by integrating the pressure over the mean wetted
Vl// yqr. yxi> body surface.
i = UaiO ax 2
onz 0 (15) 2.2. Desingularizetl source and panel method

In order to avoid waves reflecting back from the downstream To solve the boundary value problem for the disturbance
boundary when using the mixed Euler-Lagrange free surface time velocity potential, desingularized sources are distnbuted above the
stepping scheme, the free surface boundary conditions are further calm water surface and constant-strength panels are employed
written as on the mean wetted hull surface. Typical free surface and hull
discretizations are shown in Fig. 2. The desingularized distance
dn 90 d^^
= UoiO is calculated according to the formula Ds = LdSï, where Ds is
dt dx 9z the desingularized distance, S is the local grid area, and Ld = 1
-Fv- V/, onz :0 (16) is a desingularized parameter. Desingularized isolated sources
9^ 1 have the advantage of rapid computing speed, while the panels
= Uo(t) distributed on the body surface have the advantage of good
dt 9x
reliability in handling a complex hull geometry. At each time step
d^ in the simulation, the boundary integral equations are set up to
V0 onz = 0 (17)
- - d F ^ " find the unknown source strengths through applying a Neumann
condition on the body surface and a Dirichlet condition on the
where -t- V • V is the time derivadve following a fluid
calm water surface. In the usual manner, the boundary integral
pardcle along a prescribed path. The velocity of the particle is equations can be discretized into a system of linear equations,
V = ( - I / o ( f ) + ^ , v , O). Here, v is the prescribed velocity of a which can be solved to determine the panel and isolated source
collocadon point such that it moves along a given path around the strengths. Furthermore, the fluid velocity on the free surface
exact body. Note ^ - » 0 as Lfa(t) ^ a constant forward speed. and the velocity potential on the body surface can be computed
For the forced ship motion/radiation problem, the exact body knowing the source strengths. The solution is stepped in time by
boundary condition can also be linearized about the mean wetted using the free surface boundary condidons and body boundary
body position following Ogilvie and Tuck [25], condition. More details of applying the desingularized source
method in wave-body interaction problems with/without forward
(18) speed can be found in [23,24].
nj + ^jtrij] onSfi
9n
2.3. Calculation of the m-terms and the second derivative 0 - on the
where | j is the motion amplitude in j t h mode. The m-term,
free surface
nij, which represents the coupling between the steady and the
unsteady flow, is very cridcal for evaluating the cross-coupling
The calculations for the m-terms involve the computation of the
hydrodynamic coefficients accurately. Once the double-body flow
second derivative of the double-body potential on the body. The
problem has been solved, the m-terms can be computed using, direct numerical computation using a finite difference method is
(mi,m2,m3) = (n-V)(U-V«^) (19) known to have numerical difficulties and a loss of accuracy. Zhao
and Faltinsen [26] illustrate the deficiencies of direct numerical
(m4,m5,m6) = (n-V)(xx (U-V«^)) (20)
evaluation for the m-terms. Nakos and Sclavounos [27] propose
where U = (Uo(t), 0, 0) represents ship's translating velocity an approach using Stokes' theorem for the direct Green's function
vector. method.
 474 X. Zhang et al./Applied Ocean Research 32 (2010) 471-482

'dè d<P 1 \ /'dè'

(22)

hydrodynamic radiation -I- diffraction+Iiydrostatic Froude-Krylov pressure

Box I .

An integral mettiod is proposed liere to compute tlie second

derivatives of the double-body basis flow. It should be noted that
the dme-dependent double-body basis flow !//(t) only needs to
be solved once since i t can be scaled with the translating velocity
Uo(t). By solving the boundary value problem for the double-body
flow <!' using a standard source distribudon method (constant-
strength panels are employed on the body surface), the first
derivative of I' on the body surface can be computed using the
obtained source strengths cr.

VG*(x; ?)a(|)ds (23)

-fi
where G*(x; ?) = l / r ( x , y , z; ^ ?) + l / r ( x , y , z ; /?, - ? ) .
To find the second derivatives of ^ , a complementary potendal
X is set up

JjT Gt(x; ^)a/(?)ds = Xi(x) i = 1, 2, 3 (24)

where Xi(x) \ax' sy az J 1,2, 3

Fig. 3. Convergence of heave force with number of panels on half of hull surface,
G+(x; ?) = l / r ( x , y , z ; O + l/r(x,y,z; /?, with time step size At = T/lOO, number of desingularized sources above calm
water surface NF = 2500. L is the ship length, B is the full beam, A is the forced
C^(x; ?) = l / r ( x , y , z ; f , ??, f ) + l/'-(x,y, z; f , (25) motion amplitude, forced motion period T = 2ir/a, F„ = 0.3.

Glix; ?) = l / r ( x , y , z ; ?) - l / r ( x , y , z ; -f)
the free surface elevation and the velocity potential on the free
Because of the andsymmetry property of X3 (If) about z 0, a surface at each time step, a fourth-order Runge-Kutta method
special form of G 3 is selected to sadsfy X3 aw Oon thez = 0 is used to integrate the free surface boundary conditions (16)
az
plane. and (17). After each time step, the free surface has to be
By solving the discretized equadon (24), the source strength CT/ regrided, all the values including the free surface elevation and the
can be determined. Therefore, the m-terms can be evaluated using velocity potential w i l l be interpolated into the original translating
Eq.(19). coordinate system w i t h the same domain size using a cubic spline
interpolation.
m, = n • V ( U - Xi) VC/(x; | ) a t ( ? ) d s In this paper, we focus on the problem o f t = > 1/4,
-fi" w i t h no waves propagating upstream, where a)e is the encounter
i= 1,2,3 (26) wave frequency. Therefore, a homogeneous boundary condition
is applied at the upstream open boundaiy. For the problem of
Furthermore, (m4, ms, me) can be obtained using T = ^ < 1/4 where waves w i l l propagate upstream, a special
m = (may - m2z) -F (i/>yn3 - ^^112) treatment of the upstream radiation condition w i l l be needed.
ms = (miz - max) -F (if'^n, - - Lfjns) (27)
mg = (max - miy) + ((«/^, - Lf)n2 - '^^3,n,) 2.5. Numerical solver ofthe equations of motion
where ti2 and are the components of the unit normal in the y
The standard dynamic equations of motion for the body can be
and z directions, respectively.
written as
The second derivative " '^3^y°\ which is needed for the
evaluation of Eq. (16), is computed using the same approach as MhO + CHt)=F(lt^,0 (29)
used for m-term evaluations. where M is the inertia matrix for the hull. C is the hydrostatic
stiffness matrix, which is considered constant based on the mean
d^^(x,y,0) ax3(x,y,0) dclix; ?) wetted hull surface. In future studies, the forward speed effects
(28)
9z2 dz JL.
JJs-^ dz on the hydrostatic restoring forces will be included. The total
hydrodynamic forces, F, are computed at each time step and
where x = (x, y, 0) are the coordinates of the collocation points depend on the body's acceleration, velocity, and displacement.
on the calm water surface, may also be calculated using a The equations of motion and the wave flow field are solved
finite difference method, but Eq. (28) is preferred in order to avoid simultaneously in the free motion simulations.
numerical differentiation.
3. Results
2.4. Solver for the boundary value problem and time stepping ofthe
free surface 3.1. Convergence tests

The mixed boundary value problem as formulated in Section 2.1 Special convergence of the hydrodynamic radiation force act-
is solved by a preconditioned GMRES method. In order to update ing on a V\/igley I hull is illustrated i n Figs. 3 and 4. In order to show
 X. Zhang et al /Applied Ocean Research 32 (2010) 471-482 475

Table 1
Number of nodes in X (longitudinal) and V (transverse) directions for the convergence tests of free surface discretization.

Case no. Number of nodes in X direction Number of nodes in Y direction

1 80 25
2 100 25
3 120 30

0.25 -WF = 2000(80X25)

1
• Np = 2500(100X25)
•Wr=.J6(KI(120X30) j„j
0.2

0.15
0.11 A
0.108
i
0.1

-0.2512 14 15 17

Fig. 5. Convergence of heave force with number of desingularized sources on half

Fig. 4. Convergence of pitch moment with number of panels on half of hull surface,
of free surface, with time step size At = T/100, number of panels on half hull
with time step size At = T/100, number of desingularized sources above calm
NB = 200, i is the ship length, B is the full beam, A is the forced motion amplitude,
water surface = 2500, L is the ship length, B is the full beam, A is the forced
forced motion period T = 27T/CO, F„ = 0.3.
motion amplitude, forced motion period T = 27z/o>, F„ = 0.3.

the convergence more clearly, the hydrodynamic radiation force is -Wr = 2000(80X25) X 10
computed by integrating the radiated wave pressure components • = 2500(100X25) .r
• Nr = 3600(120X30)
p = - p M + M _ ( u - v^). v ^ ] included in Eq. (22). There- 3.51

fore, the steady components due to the potential solution f are

not included i n the obtained force time histories.
The Wigley I hull corresponds to Journée's "Wigley I " , as
referenced in [28]. The mathematical hull form of Wigley I is
defined by

y(x,z) = ^ [ ( l - ( 2 x / L ) 2 ) ( l ( z / r ) 2 ) ( l + 0.2(2x/L)2)] (30)

where L is the model length, B is the full beam, T is the draft. For
the Wigley 1 hull, L/B = 10, and B/T = 1.6.
The non-dimensional frequency co-^L/g = 3.3 w i t h a forward
speed corresponding to f n = = 0.3 is selected for the test.
This selected frequency corresponds to an incoming wavelength
X/L= 1.5. The forced heave motion amplitude is Tl/L = 0.01. The 12 13 14 15 16 17 18
hull is initially moved from rest smoothly to a constant forward
speed. Simultaneously, a forced sinusoidal heave motion is added
on the hull w i t h a ramp function applied. Cosine spacing is used in
Fig. 6. Convergence of pitch moment with number of desingularized sources on
the panelization of the hull surface along its longitudinal direction. half of free surface, with time step size zit = 7/100, number of panels on half hull
Similar discretizations of half o f t h e hull using 100, 200, 300, and HB = 200, L is the ship length, B is the full beam, A is the forced motion amplitude,
400 panels are tested. The steady states of the force time histories forced motion period 7" = 2 ; r / i D , f„ = 0.3.

are illustrated in Figs. 3 and 4. As seen in Fig. 3, the heave force time
history converges very fast w i t h an increasing number of panels on different discredzations are listed in Table 1. In all three cases, finer
the hull. The pitch moment time history shown in Fig. 4 converges meshes are controlled near the ship's bow and stern. As seen, both
relatively more slowly than the heave force. the heave force and pitch moment converge w i t h the increasing
Free surface discretization convergence tests have been illus- number of desingularized sources above the calm water surface.
trated i n Figs. 5 and 6. The typical free surface discretization has Temporal convergence is illustrated i n Figs. 7 and 8 for a forced
been shown in Fig. 2. Port-starboard symmetry is assumed so that heave motion by using the same number of panels NB = 300
only half the domain is needed in computations. Three similar on half of the hull at a Froude number F„ = 0.3 and a forced
discretizations using 2000, 2500, and 3600 desingularized source modon frequency co^L/g = 3.3. As seen, the convergence of both
points on the free surface are tested. The parameters for the three the heave force and pitch moment w i t h the time step size z i t is
 476 X. Zhang et al./Applied Ocean Research 32 (2010) 471-482

0.25 1 V 1
yoijf/[ = r 2 0,112
YwtlL = 1.5
0,2 - YwtlL = \.i 0.11 -

yts.m
0.15

0,1
/ 0,106

/
13,9/3.95 Xl''
\
14-05 14,t / \
^ 0.05

\ /
^ -0.05 ^ -0.05

-0.1

-0.15 -0.15

-0.2 -0.2 •

-0.25, -0.25
'12 13 14 15 ie 17 12 13 14 15 16 17

Fig. 7. Convergence of heave force with time step size At, number of panels on Fig. 9. Sensitivity of heave force to transverse domain size I'out, number of panels
half hull NB = 3 0 0 , number of desingularized sources above calm water surface on half hull NB = 200, number of desingularized sources above calm water surface
Nf = 2 5 0 0 , L is the ship length, 6 is the full beam,/I is the forced motion amplitude, Np = 2500, with time step size At = T / 1 0 0 , 1 is the ship iength, B is the full beam,
forced motion period T = 2TX/O), F„ = 0 . 3 . A is the forced motion amplitude.

12 13 14 15 16 17 18

ty/Lfg

Fig. 8. Convergence of pitch moment with time step size z\f, number of panels on Fig. TO. Sensitivity of pitch moment to transverse domain size Vout, number of
half hull WB = 300, number of desingularized sources above calm water surface panels on half hull NB = 200, number of desingularized sources above calm water
Np = 2500, L is the ship length, B is the full beam, /\ is the forced motion amplitude, surface Nf = 2500, with time step size zlt = T/lOO, L is the ship length, B is the
forced motion period T = 2TX/lo, F„ = 0.3. full beam, A is the forced motion amplitude.

very fast. Note that the asymmetries o f t h e force time histories in Figs. 9 and 10 illustrate the hydrodynamic force time histories
Figs. 3 through 8 are due to the additional forcing terms created by on the modified Wigley I hull at a Froude number Fn = 0.3
the double-body basis flow retained in the free surface boundary and a forced heave motion frequency a ^ l f g = 3.3, using three
conditions (16) and (17). If uniform flow is chosen as the basis flow, different transverse domain truncation sizes. The forced heave
the asymmetry disappears. motion amplitude is A/l = 0.01. As seen, the prediction of the
hydrodynamic force is not affected by the transverse domain size
3.2. Sensitivity study on the domain size on free surface at Froude number 0.3 as Vout/f- > 1-2.

The sensitivity of the present numencal results to the selection 3.3. Forced motion simulations
of the free surface domain truncation is investigated. The size of
the transverse domain boundary Vout labeled in Fig. 1 is vaned to Forced motion simulations are carried out to validate the
demonstrate the sensitivity of the hydrodynamic force to the free capability of the present model to predict the hydrodynamic
surface domain selection. coefficients. The added mass and damping coefficients are obtained
For the problem with forward speed, an upstream truncation f r o m the forced motion simulation results using the technique
boundary of half a ship length f r o m the bow, and a downstream of Fourier analysis. All the present computational results are
truncation of one and a half ship lengths after the stern are used compared w i t h experiments and shown in Figs. 11 through 14.
for the present computations. In these figures, four sets of results are presented: present
 X. Zhang et al. /Applied Ocean Research 32 (2010) 471-482

1.2 2.5
Present Compulation Present Computation
N - K Approach N - K Approach
1
B E Strip Ttieory B E Strip TTieory
L F S . D B mlerriis - • L F S . D B mterms
0.8 O Experiment - Experiment

^" > 0.6

0.4

0.2

0
2.5 3 3.5 4 4.5 2.5 3 3.5 4 4.5

aj\jL/a us/Ljg

- Present Computation Present Compulation

N - K Approach 0.1 N - K Approach
B E Strip Theory B E Strip Theory
L F S . D B mlerms
0.05
O Experiment

-0.05

-0.1

-0.15

-0.2
2,5 3 3.5 4 4.5

Fig. 11. Hydrodynamic coefficients due to forced heave motion for a modified Wigley 1 hull at Froude number = 0.3.

0.05 0.12
Present Computation Present Computalion
N-K Approach N - K Approach
B E Strip Theory 0.1 B E Strip Theory
0.04
LFS,DB mlerms L F S . D B mterms
O Experiment O Experiment
0.08
0.03

O 0.06

0.02
0.04

0.01
0.02

0_ 0
2.5 3 3.5 4 4.5 2.5 3 3.5 4 4.5

0.1
Present Computalion Present Compulation
N - K Approach — N - K Approach
- •— B E Strip Theory B E Strip Theory
0.05 LFS.DB mlerms L F S . D B mterms
O Experiment O Experiment

-0.05

-0.1
2.5 3 3.5 4 4.5 5

i^\jL/g

Fig. 12. Hydrodynamic coefficients due to forced pitch motion for a modified Wigley 1 hull at Froude number = 0.3.
478 X. Zhang et al./Applied Ocean Research 32(2010) 471-482

3.5
- Present Computalion Present Computation
1.6 f^-K Approacti hJ-K Approach
3
B E Strip Theory B E Strip Theory
1.4 - L F S . D B mlerms — — L F S . D B mterms
Experiment 2.5 O Experiment
1.2

1 2

0.8
1.5

0.6
1
0.4

0.2 0.5

0 0
4 4

U)^/L/g

0.1 Present Compulation Present Compulation

N - K Approach 0.2 N - K Approach
B E Strip Theory - ~ B E Strip Theory
0.08 0.15
L F S . D B mlerms L F S . D B mterms
O Experimenl O Experimenl
0.06 0.1

to 0.05

^ > 0.04 0

-0.05
0.02
-0.1
0
-0.15

-0.02 -0.2
4 4

0J\/L/g uJ\/L/g

Fig. 13. Hydrodynamic coefficients due to forced heave motion for a Series 60 hull with a block coefficient CB = 0.7 at Froude number = 0.2.

0.1
Present Computation — Present Computalion
N - K Approach 0.16 N - K Approach
0.08 . •- - B E Strip Theory - - • B E Strip Theory
L F S . D B mlerms 0.14 LFS,DB mtenns
O Experiment O Experiment
0.12
1 ^ ~ - — , ^
0.06
1$ 0.1

0.04 s 1> 0.08

0.06

0.02 0.04

0.02

2 3 4 5 6 3 4 - 5

wsflfg uJs/IJg

0.1
- Present Compulation Present Compulation
N - K Approach N - K Approach
B E Strip Theory 0.5 - B E Strip Theory
0.05 • L F S . D B mlerms L F S . D B mterms
Experimenl O Experiment
0.4

-
0.3

cq t?;

-0.05 0.2

0.1

-0.1
4 4
0
ij3%jL/g

Fig. 14. Hydrodynamic coefficients due to forced pitch motion for a Series 60 hull with a block coefficient CB = 0.7 at Froude number = 0.2.
 X. Zhang et al./Applied Ocean Research 32 (2010) 471-482 479

Table 2 3.4. Wave tiiffraction results

Particulars of Series 60 hull with a block coefficient = 0.7, /^wp is the waterplane
area, C„p is the coefficient.
The velocity potential for a linear incident wave of unit
L/B 7.00 amplitude in deep water is given by the real part of the complex
B/T 2.50
potential <p'
CB 0.70
t<y,/L 0.25
Cwp = A v p / a B ) 0.78 0'(x, y, z, t) = ii_e'"e"'(''™^+^™«+''"'^' (32)
Wo

where is the frequency of the incident wave in the frame of the

double-body model; Neumann-Kelvin approach; an approach of earth-fixed coordinate system, co^ is the encounter frequency, /8 is
using Linear Free Surface boundary conditions (same as that the angle of wave propagation measured f r o m the positive sense of
in the Neumann-Kelvin approach) w i t h Double-Body m-terms the X-axis, k is the incident wave number. The encounter frequency
(LFS-DB m-terms); a time-domain Body-Exact Strip Theory (BE coe is calculated by
Strip Theory). The Neumann-Kelvin approach employs linear
free surface boundary conditions and the simplified m-terms. We = Wo-kU cos p (33)
The simplified m-terms are computed based on uniform basis f l o w The incident wave amplitude A/L = 0.01 is selected for the
= -Uo(t)x instead of using double-body basis flow. Hence, the present computations. Fig. 15 shows the amplitude and phase an-
formulation of the simplified m-terms reduces to, gle of the wave exciting heave and pitch forces acting on the modi-
fied Wigley I hull over a range of incident wavelengths. The present
( m i , m 2 , nis, m4, m j , mg) = (0, 0,0, 0, Uo(t)n3, - U o ( t ) n 2 ) (31)
results are compared w i t h experiments by Journée [28], and the
The rime-domain body-exact strip theory is formulated by numerical solurion using the Neumann-Kelvin approach and the
Zhang et al. [24] and Bandyk [29], However, in order to time-domain body-exact strip theory. As seen in those figures, the
compare with the present linearized model, a small forced motion present results generally agree well w i t h the experiments. The
amplitude is used in the rime-domain body-exact strip theory only exception is in the phase angle prediction of exciting pitch
model. Both a modified Wigley 1 hull and a Series 60 hull w i t h a moment. As also shown in the comparisons, the difference of the
block coefficient Q = 0.7 are tested. The main particulars of the results between using double-body linearization and using Neu-
Series 60 hull are listed in Table 2. mann-Kelvin approach is trivial, although the results of double-
body basis flow is slightly better in general.
The added mass and damping coefficients due to forced
heave/pitch morions for the modified Wigley I hull are illustrated Fig. 16 shows the amplitude and phase angle of the wave excit-
in Figs. 11 and 12, which are compared w i t h the experiments by ing heave and pitch forces acting on a Series 60 hull w i t h a block
coefficient Cg = 0.7 over a range of incident wavelengths. The
Journée [28]. The forced heave motion amplitude is selected to be
present results are compared w i t h the experimental results given
A/L = 0.01, and the forced pitch modon amplitude iSi4 = 0.02 rad.
in [15], and the numerical solutions using the Neumann-Kelvin
The comparisons show that the present results generally agree well
approximation and the time-domain body-exact strip theory. The
w i t h the experiments, especially for cross-coupling hydrodynamic
comparisons show that the present numerical model gives better
coefficients.
agreement w i t h experiments than other methods, especially for
Similar results for a Series 60 hull w i t h CB = 0.7 are illustrated the phase angle prediction.
in Figs. 13 and 14, which are compared w i t h experiments by
Gerritsma et al. [30]. As seen, the present numerical solutions
3.5. Free motion simulation
agree well w i t h experiments over a range of frequencies. The
comparisons for both hulls show that the present results using
Free motion simulations for a modified Wigley 1 hull and a typ-
double-body linearization are in general better than the results
ical Series 60 hull w i t h a block coefficient CB = 0.7 are per-
using the Neumann-Kelvin approach. This is most evident i n the
formed. The incident wave amplitude A/L = 0.01 is selected for
prediction of the cross-coupling hydrodynamic coefficients, and
the present computations. The rime-domain motion responses for
diagonal hydrodynamic coefficients A 5 5 and B 5 5 . The advantage
both hulls in regular head seas of different incident wavelengths
of the present results can be attributed to both the double-body
are obtained. The motion response amplitude and phase RAOs are
m-term computations and the leading-order terms retained in
calculated f r o m the motion response time history using the tech-
the free surface boundary conditions (16) and (17) based on the
nique of Fourier analysis.
double-body linearization. In order to identify the dominant source
Fig. 17 illustrates the heave and pitch response amplitude
of the present model's improved agreement w i t h experiments,
operators over a range of incident wavelengths near the resonance
another model using the linear free surface boundary conditions
period for the modified Wigley I hull. The present results are
and double-body m-terms (LFS-DB m-terms) is added into the
compared w i t h experiments by Journée [28] and the numerical
comparisons. The comparisons between the present numerical
solutions using the Neumann-Kelvin approach and the rime-
results w i t h those using LFS-DB m-terms indicate that the double- domain body-exact strip theory. The present results generally
body m-terms play a more important role (more than 90%) in agree well w i t h experiments, except that the pitch motion
compuring the cross-coupling added mass ( A 3 5 and A 5 3 ) and response is overesrimated a littie bit.
diagonal coefficients {A55 and B 5 5 ) for pitch motion accurately, Similarly, Fig. 18 compares the present results w i t h experi-
while the addirional terms in the free surface boundary condition ments [30] and the numerical solutions using the Neumann-Kelvin
contribute littie (less than 10%) to the improved ^ 3 5 , ^ 5 3 , A55, approach and the time-domain body-exact strip theory for a Series
and B55. By examining the results of B35 and B53, it is found 60 hull. As seen, the comparisons are generally good.
that the double-body m-terms and the leading-order terms i n the
free surface boundary conditions almost contribute equally to the
4. Conclusions
prediction of the damping coefficient B53. The leading-order terms
in the free surface boundary conditions are the most important
A three-dimensional, time-domain, ship-wave interaction
source for improving the computational accuracy of the damping model is formulated and developed in this paper based on a
coefficient B35. double-body linearization and the mixed Euler-Lagrange time
 480 X. Zhang et al. /Applied Ocean Research 32 (2010) 471-482

0.8
350
Present Computation - Present Computation
0.7 N - K Approacti N - K Approach
- - BE Strip Ttieory 300 B E Strip Theory
0.6 O Experiment Experiment
250
0.5
200
0.4
150
0.3

0.2 100

0.1 50

Ol—
0.5 1.5 2.5 0.5 1.5 2.5
X/L A/L

0.8
350
- Present Computation
0.7 N - K Approach
• B E Strip Theory 300
0.6 Experimenl
250 I O
0.5
200
S 0-4
150
0.3
100
0.2 Present Compulation
N - K Approach
0.1 50 - - • B E Strip Theory
O Experiment
Ol—
0.5 1.5 2.5 °0.5 1.5 2.5
X/L X/L

Fig. 15. Heave and pitch exciting force acting on a modified Wigley I hull at Froude number = 0,3 through a head sea, C33 and C55 are hydrostatic coefficients, k is the
incident wave number, A is the incident wave amplitude.

Fig. 16. Heave and pitch exciting force acting on a Series 60 hull with CB = 0.7 at Froude number = 0.2 through a head sea, C33 and C55 are hydrostatic coefficients, /(is the
incident wave number, /I is the incident wave amplitude.
X. Zhang et al./Applied Ocean Research 32 (2010) 471-482 481
482 X. Zhang et al./Applied Ocean Research 32 (2010) 471-482

Stepping sctieme. Numerical results of the hydrodynamic coeffi- [8] Cao Y, Lee TH, Beck R F Computation of nonlinear waves generated by floating
cients for a modified Wigley 1 hull and a Series 60 hull are com- bodies. In: 7th international workshop on water waves and floadng bodies.
1992. p. 47-52.
pared w i t h experiments and other numerical solutions. The free
]9] Scorpio SM, Beck RF, Korsmeyer FT. Nonlinear water wave computations using
motion response RAOs in a regular seaway are also presented a mulripole accelerated, desingularized method. In: 21st symposium on naval
and compared to experiments. The present computational results hydrodynamics. 1996. p. 34-43.
are found to be in excellent agreement w i t h experiments. The [10] Bai W, Taylor E. Higher-order boundary element simulation of fully nonlinear
wave radiation by oscillating vertical cylinders. Applied Ocean Research 2006;
comparisons also demonstrate that the present model using a
28:247-65.
double-body basis flow is generally better than the time-domain [11] Bai W, Taylor E. Numerical simulation of fully nonlinear regular and focused
body-exact strip theory and the Neumann-Kelvin approximation wave diffraction around a vertical cylinder using domain decomposition.
model using simplified m-terms. The advantage of the present nu- Applied Ocean Research 2007;29:55-71.
[12] Ma Q, Yan S. Numerical simulation of fully nonlinear interaction between
merical model over the time-domain body-exact strip theory can
steep waves and 2D flosting bodies using the qale-fem method. Journal of
be attributed to the fact that the present three-dimensional model Computational Physics 2007;221:666-92.
handles the forward speed effects inherently instead of using [13] Liu C, Huang Z, Tan SK. Nonlinear scattering of non-breaking waves by a
forward speed corrections. The present approach gives more accu- submerged horizontal plate: experiments and simularions. Ocean Engineering
2009;36:1332-45.
rate results than the Neumann-Kelvin approach using simplified
[14] Beck R F Magee A. Time-domain analysis for predicting ship motions. In:
m-terms because the double-body m-terms capture some of the lUTAM symp. dynamics of marine vehicles and structures in waves. 1990.
coupling effects between the steady and unsteady flows. Moreover, p. 49-65.
the leading-order terms kept in the free surface boundary condi- [15] Lin WM, Yue DKP. Numerical solutions for large-amplitude ship motion in the
time domain. In: 18th symposium on naval hydrodynamics. 1990. p. 41-66.
tions improve the computadonal accuracy of the cross-coupling
[16] Bingham H. Simulating ship morions in the rime domain. Ph.D. thesis. MIT:
hydrodynamic damping terms B53 and B35. The present model can Department of Ocean Engineering; 1994.
be extended to study the body-exact problem and the added re- [17] Huang Y Sclavounos P. Nonlinear ship motions. Journal of Ship Research 1998;
sistance (second-order force) problem, and w i l l be reported in the 42(2): 120-30.
[18] Sen D. Time-domain computation of large amplitude 3D ship motions with
future.
forward speed. Ocean Engineering2002;29:973-1002.
]19] Singh S, Sen D. A comparative linear and nonlinear ship motion study using
Acknowledgement 3-D time domain methods. Ocean Engineering 2007;34:1863-81.
[20| Zhang X, Beck R F 2-D body-exact computations in the rime domain. In: 21st
international workshop on water waves and floadng bodies. 2006. p. 197-200.
This work was partially supported by the Office of Naval [21] Zhang X, Beck R F Computadons for large-amplitude two-dimensional
Research, contracts N00014-05-1-0537, N00014-06-1-0879, and motions. Journal of Engineering Mathematics 2007;41(l-4):177-89.
[22] Zhang X. Large amplitude ship modon computations using a dme dependent
N00014-08-1-0594.
body geometry. Ph.D. thesis. The University of Michigan, Department of Naval
Architecture and Marine Engineering. 2007.
References ]23] Zhang X, Beck RF Three-dimensional large amplitude body motions in waves.
Journal of Offshore Mechanics and Arctic Engineering 2008; 130(4).
[1] Longuet-Higgins IMS, Cokelet CD. The deformation of steep surface waves on [24| Zhang X, Bandyk P, Beck R F Time-domain simulations of radiadon and
water: I.A numerical method of computation. Proceedings ofthe Royal Society diffraction forces. Journal of Ship Research 2010;54(2):79-94.
of London, Series A 1976;350:1-26. [25] Ogilvie T, Tuck E. A rational strip theory for ship motions, part 1. Technical
[2] Bai KJ, Yeung RW. Numerical solutions to free surface-flow problems. In: 10th report 013. Tech. rep. Ann Arbor (Michigan): The Department of Naval
symposium on naval hydrodynamics. 1974. Architecture and Marine Engineering, The University of Michigan; 1969.
[31 Dommermuth DG, Yue DKP. Numerical simulations of nonlinear axisymmetric [26] Zhao R, Faltinsen 0. A discussion of the tn-terms in the wave-current body
flows with a free suiface. Journal of Fluid Mechanics 1987;178:195-219. interacrion problem. In: 3rd international workshop on water waves and
[41 Xue M, Xu H, Liu Y, Yue DKP. Computations of fully nonlinear three floadng bodies. 1989.
dimensional wave-wave and wave-body interactions. Part 1. Dynamics [271 Nakos D, Sclavounos PD. Ship motions by a three-dimensional rankine panel
of steep three dimensional waves. Journal of Fluid Mechanics 2001 ;438: method. In: 18th symposium on naval hydrodynamics. Nadonal Academies
11-39. Press; 1990. p. 21-40.
15] Cao Y, Schultz WW, Beck RF. Three-dimensional, unsteady computations of [28| Journée JM. Experiments and calcularions on four wigley hull forms. Technical
nonlinear waves caused by underwater disturbances. In: 18th symposium on report 909. Tech. rep. Delft (The Netherlands): Delft University of Technology,
naval hydrodynamics. 1990. p. 417-27. Ship Hydromechanics Laboratory; 1992.
[6] Cao Y, Schultz W W , Beck R F A three-dimensional desingularized boundary [29] Bandyk P. A body-exact strip theory approach to ship modon computations.
integral method for potential problems. International Journal of Numerical Ph.D. thesis. The University of Michigan, Department of Naval Architecture and
Methods in Fluids 1991; 11:785-803. Marine Engineering. 2009.
[7] Cao Y. Computations of nonlinear gravity waves by a desingularized boundary [30| Gerritsma J, Beukelman W, Glansdorp C The effect of beam on the
integral method. Ph.D. thesis. The University of Michigan, Department of Naval hydrodynamic characterisrics of ship hulls. In: 10th symposium on naval
Architecture and Marine Engineering. 1991. hydrodynamics. 1974.