You are on page 1of 10

Time domain simulation of ship motions

A. Colagrossi M. Landrini
INSEAN, Italian ship model basin.
Via di Vallerano 139, 00128 Roma { Italy.

1 Background and motivation


As a matter of fact, strip methods for seakeeping still represent a convenient method for pre-
dicting the motion of ships with arbitrary geometry and speed.
Three dimensional panel codes for solving the ship motion problem in frequency domain are
becoming more and mature to be used as a design tool [2] and even nonlinearities related to
the steady basis ow can be incorporated in the formulation. Unfortunately, further inclusions
of unsteady nonlinear e ects seem impractical and diculties related to the enforcement of the
radiation condition presently limit Rankine panel methods in frequency domain to  = U!=g
high enough.
These drawbacks motivate the development of time domain approaches for ships of arbitrary
geometry and speed. One possible approach, renewed by Beck and coauthors [6], is based on
the Green function satisfying the Kelvin free surface condition (see the recent results in [3]).
Sclavounos and his group [12] are pursuing the use of free space Green function (i.e. Rankine
panel) which allows for the inclusion of nonlinearities. Fully nonlinear potential ow and viscous
ow codes are too much time consuming for being of practical use in the near future.
Here we present preliminary results concerning our ongoing e ort to develop a linear time
domain code of general purpose. Having in mind the possible inclusion of nonlinear e ects, the
algorithm is based on Rankine sources. This requires the discretization of the entire boundary
domain (hull and free surface) and to reduce the computational e ort a numerical transient test
technique is adopted to evaluate with a single run the whole behavior of the ship.
Results concerning two mathematical catamarans and a fast container vessel are presented.
In the latter case satisfactory agreement with experimental data is obtained for head and
following waves.

2 Problem de nition...
In the following the unsteady motion of a ship in a seaway is studied by an inviscid model.
In this framework, the statement of the 'exact' (i.e. nonlinear) problem can be found in [11].
Here, we consider the problem of determining the small oscillations =  +
 R of the ship
with respect to the mean con guration attained when it moves with constant forward velocity
U in the x{direction. W will denote the basis{ ow [10]. The perturbation velocity potential '
satis es the Laplace equation
r2' = 0 (1)
together with the standard no{penetration boundary condition
' = ( + r  (  W ))  n
n t '0 ;n (2)
on the mean wetted hull B and the linearized kinematic and dynamic free surface conditions

r  W +  W  k^ + r2Wg  r '
2
 ='
t z z ' = g W  r'
t (3)
respectively, on the mean free surface level F . Lower suces stand for partial derivatives. The
above problem describes the induced motion of the ship under the action of incoming waves
with potential '0 . The uid dynamic problem is coupled with the body motion through the
hydrodynamic loads Z Z
F = p n dS M = p R  n dS (4)
B B
and, in turn, the motion a ects the ow eld via the body boundary condition (2). To evaluate
the pressure p = (' + W  r')= on the hull, the time derivative of the potential, ' , is
t t

computed by solving the auxiliary boundary value problem


r2 ' = 0
t ' = ( +
 W )  n on B
nt tt t ' = g W  r' on F
t (5)
It can be noted that the above problem is linear and that the body boundary condition depends
linearly on the acceleration . This allows for determining, once and for all, the part of the
tt

hydrodynamic loads proportional to and lumping it with the physical inertia matrix.
tt

Alternatively, the classical Haskind decomposition of ' in radiation potentials ' , related
j

to the body oscillations in calm water, and scattering potential '7 , arising from the interaction
with incoming waves, can be introduced as well. This procedure does not require the coupled
solution of the equations of body motion but two di erent time simulations are required.
In the developed algorithm, we take advantage from the possibility of deducing the entire
response operator of a linear system over a wide range of wavelengths by analyzing the single
response to a given compact (in time) wave induced excitation. This procedure results in a
shorter simulation time when a large number of frequency has to be considered.

3 ...and numerical solution

Numerical Beach

1.0 L

Z
X
Y
S175 Hull

U
1.2 L 0.5 L
Forward Speed

Figure 1: Example of discretized boundary domain for a S175 hull. Typical dimensions are
normalized by the ship length L. The damping layer surrounding the computational domain is
marked by the thick line.
The di erential problem (1{3) is recast in the form of boundary integral equations by de-
coupling it into a kinematic problem and an evolution problem. Namely, at a given instant
of time, both the potential 'jF on the free surface and the normal gradient ' jB on the body
n
boundary are known. A mixed Neumann{Dirichlet boundary value problem for the Laplace
equation can then be stated and the layer ansatz
Z Z
' (Q ) =  G(P ; Q) dS +
P  G(P ; Q) dS P (6)
B F
where G(P ; Q) = 1=4jP Qj is the free space Green function, is assumed to represent the
solution. A set of integral equations is deduced by collocating (6) on on the free surface and on
the body, respectively, and the numerical solution is achieved by a standard low{order panel{
method [5]. First order derivatives on B [ F and second order derivatives on F are evaluated
analytically, while, due to the poor approximation of the m {terms, an extrapolation procedure
j

is developed for the gradient of the velocity on the body boundary.


Once the velocity potential is computed on the boundary domain, the free surface equations,
as well as the equations of body motion, can be stepped forward in time.
As mentioned, the pressure requires the computation of ' and, because of the similarity of
t

the problem (5) with the kinematic problem for ', the same boundary integral equations just
solved for  can be adopted to determine the source strength related to ' . t

Further, to avoid the evaluation of the tangential derivative r (see the rst eq. (3)), we
convert it into r = [r' + r(W r')]=g which requires the already computed rr', while
t

the term r' is obtained by a small extra{e ort after the computation of ' .
t t

Finally, a second order Runge{Kutta scheme is adopted to step{forward in time the evolution
equations.
The continuous problem for ' requires assumptions on the asymptotic behavior at large
distances from the hull. On physical ground, and for nite times, it can be assumed that the
disturbance induced by the body damps out far from it. Actually, due to the nite extent of
the computational domain, unphysical re ections unavoidably appear without the introduction
of suitable adsorbing conditions. In the present work, damping layers [8] are introduced by
modifying the free surface equations (3) according to the scheme originally proposed in [10]

 = ' r  W +  W k^ + r2Wg  r ' 2 + g '


2 2
t z z ' = g W  r'
t

(7)
In gure 1, a typical damping frame surrounding the inner physical free surface is shown. The
damping coecient  is assumed to smoothly increases from zero up to a maximum value
empirically determined.

4 Discussion of results
First computations for catamarans Recently, Bailey et al. (1998) have considered a
'mathematical' catamaran made by two ellipsoids and performed computations in the frequency
domain using two di erent algorithms based on a Green function method. Here, we adopt
that case as a reference result to validate the present numerical analysis based on free motion
transient tests. Further assessment of the (possible) weakness of the model due to the potential
ow hypothesis and the linearization demands for a suitable experimental activity and it is left
for the future.
Upon assuming the ship length L as a reference length, the beam is B=L = 0:125 and the
separation distance between the demihull centerplanes is d=L = 0:4. Three di erent speeds,
F = 0:2, 0.5 and 1, have been considered.
In gure 2 the results for the lower Froude number are summarized. The top plots show
the heave 3 and pitch 5 responses for the catamaran (solid line) and the single demihull
(dashed line). The initial time t = 0 is chosen to coincide with the focusing instant and, at
that time, the mid{ship point is located at the focusing point. Both the two ships start to 'feel'
the action of the incoming wave pulse before it is completely focused. The maximum heave
displacement is for t = 0 and, as it is reasonable, the pitch motion is almost anti-symmetric
with respect to the focusing instant. During the rst stage, the responses for the catamaran
and the monohull are roughly the same and di erences appear only during the oscillatory
tails. In particular, heave oscillations are damped out faster for the monohull while, in pitch,
the decaying rates are almost the same and the main di erence is a phase shift between the
two responses. Consistently, the largest di erences between the catamaran and the monohull
response amplitude operators appear for the heave motion (bottom plot in g. 2). In particular,

0.8 3 /
0.6

0.4

0.2

-0.2
-4 -2 0 2 4 6 8 10
t g/L

2 5 /
1

-1

-2
-4 -2 0 2 4 6 8 10
t g/L

7
1.4 3 / 5 /
6
1.2
1 5
0.8 4
0.6 3
0.4 2
0.2 1
0
0
-0.2
0.5 1 1.5 L/ 2 0.5 1 1.5 L/ 2

3 5
200 200
150 150
100 100
50 50
0 0
-50 -50
-100 -100
-150 -150
-200 -200
0.5 1 1.5 2 0.5 1 1.5 2
L/ L/

Figure 2: Behaviors of a twin{ellipsoid catamaran (solid lines) and an ellipsoidal monohull


(dashed lines) in head sea for Fr = 0:2. Top plots: heave and pitch responses due to a wave
pulse. Bottom plots: response amplitude operators for heave (left) and pitch (right).
while the long and short wavelength limits are the same, in the case of the catamaran a resonance
peak appears for wavelengths close to the ship length. A secondary, small, peak is predicted for
L= ' 1:5 As the speed increases, the interaction e ects between the two demihulls decrease
and the response of the catamaran eventually resembles the one of the single demihull, with
very marginal di erences in amplitudes and phases. This can be seen from gures 3 and 4
concerning Fr = 0:5 and Fr = 1, respectively, thus con rming the possibility of analyzing
the high Froude cases by a non{interacting demihull analysis. These and other, not reported,
results have been compared (successfully) with those reported in [1], showing that the present
4 12
3 / 5 /
3.5 10
3
8
2.5
2 6
1.5 4
1
2
0.5
0 0
0.5 1 L/ 0.5 1 L/

3 5
100 200
150
50
100
0
50
-50 0
-50
-100
-100
-150
-150
-200 -200
0.5 1 L/ 0.5 1 L/

Figure 3: Response amplitude operators for heave (left) and pitch (right) in head sea for
Fr = 0:5; solid lines: twin{ellipsoid catamaran, dashed lines: ellipsoidal monohull.

5
4.5
3 / 14 5 /
4 12
3.5 10
3
8
2.5
2 6
1.5 4
1
2
0.5
0 0
0.5 L/ 1 0.5 L/ 1

3 5
50 200
150
0
100
-50 50
0
-100 -50
-100
-150
-150
-200 -200
0.5 1 0.5 1
L/ L/

Figure 4: Response amplitude operators for heave (left) and pitch (right) in head sea for Fr = 1;
solid lines: twin{ellipsoid catamaran, dashed lines: ellipsoidal monohull.
time{domain analysis by transient tests is fully equivalent, also from a practical point of view,
to a frequency domain approach.

Figure 5: Contour lines of the wave elevation past a Wigley{catamaran advancing from left to
right (Fr = 0:3) just after the interaction with a wave pulse. In the lower part of the gure,
also a portion of the free surface discretization is shown.
As a (slightly) more realistic case, the numerical method is applied to a catamaran made
by two modi ed Wigley hulls (type I in [9]); this test{case, although inspired to, cannot be
quantitatively compared with the similar one in [7] because the Wigley parameters was not
reported there. In spite of the sharp{edged sterns, we omit to enforce a Kutta condition there
because of the marginal role it is proved to have, at least for heave and pitch (see [7, 2]). For
Fr = 0:3, the contour lines of the wave pattern just after the passage through the focusing
point is shown in gure 5, where also a portion of the free surface discretization is shown.
At rst, according to the classical Haskind decomposition, we studied separately the radia-
tion and the di raction problems by transient motion tests and the (restrained) interaction with
a wave packet, respectively. This allows for computing added mass and damping coecients,
g. 6, and the exciting forces, g. 7, which are compared with the demihull characteristics.
In the latter case, experiments from [9] have been previously used to validate the numerical
procedure [4]. Marked di erences between the radiation properties of the two ships are easily
detected, while the mono- and the twin{hulls su er almost the same exciting forces over the
entire range of wavelengths considered.
Although the response amplitude operators for the two ships could be obtained by processing
the above results, gure 8 shows them evaluated by studying the interaction (top plots) of the
free running ships with wave pulses. A distinguishing feature of the catamaran behavior is the
double{peaked pitch response, with the two local maxima shifted below and above the resonant
frequency of the monohull. The amplitudes are slightly smaller in pitch, while the catamaran
heave response is slightly higher with an almost uniform amplitude over an interval centered
about the monohull resonance peak.
S-175 in following waves Because of the numerical enforcement of the radiation condition,
the following waves case is known to be dicult to handle in the frequency domain when
using a Rankine panel method. In principle a time domain approach should not su er this
limitation and we developed a version of the code to deal with waves approaching from an
arbitrary direction. Recently, Korsmayer and Bingham [3] presented results obtained by the
Green function formulation of the time domain problem. Although requiring a larger number of
unknowns, the present approach, based on the free space Green function, is in principle suitable
to deal with more general type of linearization and can be extended to include nonlinear e ects.
The algorithm has been preliminary validated for head waves and Fr = 0:275 (see g. 9,
solid line) by comparison with experiments, 4, and frequency domain computations, , from
[2]. The overall agreement is acceptable but an over determined heave response is obtained.
By comparing with a frequency domain computation with the same linearization, , a similar
behavior is obtained which can therefore ascribed to the linearization adopted and suggests
further improvements of the model.
The following waves problem is then solved and the results are reported in gure 10. For
this problem, we experienced some diculties in using a wave packet to excite the response
and the shown results have been obtained by studying in time the interaction between the free
running hull and (several) mono{chromatic wave trains. The resulting responses are eventually
Fourier analyzed to obtain the RAO which is in a reasonable agreement with the experimental
results.

5 Conclusion and perspectives


A boundary integral equation method to solve the ship motion problem in time domain is
presented. In particular, the numerical simulation of transient tests in which a free running
ship interacts with a single wave packet is developed to compute, in a relatively short time, the
response operator over a wide range of wavelengths. This approach is systematically adopted
to solve the problem in head waves and some results for monohull and for 'mathematical'
catamaran vessels are presented. Although diculties have been encountered for following
waves when using a transient test technique, the simulation of free motion tests in 'regular'
following waves allows for predicting the response operator quite well.
In spite of the obtained results seem reasonable and a partial validation of the method is
already achieved by comparing with frequency domain computations and experiments, it is
felt that these preliminary results still require a deeper comparison with experimental data.
Reliable geometries, in fact, are characterized by bow are and transom stern and numerical
diculties and lost physical phenomena (e.g. nonlinearities) are expected and will certainly
require further re nements of the model and of the numerics.
This research was supported by the Italian Ministero dei Trasporti e della Navigazione through
INSEAN Research Program 1997{99.

References
[1] P.A. Bailey, D.A. Hudson, W.G. Price, P. Temarel, 1998. Theoretical Validation of the
Hydrodynamics of High Speed Mono{ and Multi{Hull Vessels Traveling in a Seaway.
PRADS "98, The Hague, The Netherlands.
[2] V. Bertram (1998). Numerical investigation of steady ow e ects in 3D seakeeping compu-
tations. 22 Symp. on Naval Hydrodynamics. Washington.
nd

[3] H.B. Bingham, F.T. Korsmeyer, The Forward Speed Di raction Problem, J. of Ship Res.,
Vol. 42, No. 2, June 1998.
[4] A. Colagrossi & M. Greco, 1998. Time domain analysis of ship motions by a Rankine panel
method, Numerical Towing Tank Symposium (NuTTS) 1998, Hamburg, Germany.
[5] J. L. Hess & A. M. O. Smith, 1966 Calculation of non{lifting potential ow about arbitrary
bodies, Prog. Aero. Sci. 8,1{138.
0.7 2
A 33
B33 L /g

0.6
1.5
0.5
1
0.4

0.3 0.5

0.2
0
2 3 4 5 6 2 3 4 5 6
L /g L /g

0.03
A 55 0.045
L2 0.04
0.025 0.035
0.03
0.02
0.025
0.02 B55 L /g
0.015
0.015 L2
0.01 0.01
2 3 4 5 6 2 3 4 5 6
L /g L /g

A 53 0.3
0.02 L
0.2
0 B 53 L /g
-0.02
0.1 L
-0.04 0
A 35
-0.06 L -0.1
B 35 L /g
-0.2
-0.08 L
2 3 4 5 L /g 6 2 3 4 5 L /g 6

Figure 6: Wigley{catamaran. Added mass (left) and damping coecients (right) for Fr = 0:3.
Dashed lines refer to the single demihull.

2.5
X 3 L/A X 5 L/A
15
g 2 g L

10 1.5

1
5
0.5

0 0
1 2 L/ 3 1 2 L/ 3

200 200
X3 X5
100 100

0 0

-100 -100

-200 -200
1 2 L/ 3 1 2 L/ 3

Figure 7: Wigley{catamaran. Exciting forces in head waves for Fr = 0:3. Dashed lines refer to
the single demihull.
1.2 3 /
0.9
0.6
0.3
0
-0.3
-10 0 10 20
t g/L

3
5 /
2
1
0
-1
-2
-3
-10 0 10 20
t g/L

4 3 / 5 /
15

3
10
2

5
1

0 0
0.4 0.8 1.2
L/ 0.4 0.8 1.2 L/
200
50 3 5
100
0

-50 0

-100
-100
-150
-200
-200 0.4 0.8 1.2
0.4 0.8 1.2
L/ L/

Figure 8: Wigley{catamaran. Top: heave and pitch responses for free running transient tests
in head waves for Fr = 0:3. Bottom: response amplitude operators. Dashed lines refer to the
single demihull.

[6] B.K. King, R.F. Beck and A.R. Magee, 1988. Seakeeping calculations with forward speed
using time{domain analysis, 17 Symp. on Naval Hydr.
th

[7] D. Kring and P. Sclavounos, Wave Pattern and Seakeeping of Multi-Hull Ships, FAST91,
Trondheim, Norway.
[8] M. Israeli and S. Orszag. Approximation of radiation boundary conditions J. of Comp.
Physics, 41:115{135, 1981.
[9] J. M. J. Journee, Experiments and Calculations on Four Wigley Hull Forms, Report No.
909, Ship Hydromechanics Laboratory, Delft University of Technology, Delft, The Nether-
lands, 1992.
[10] D. E. Nakos, D. Kring and P.D Sclavounos Rankine Panel Method for Transient Free-
Surface Flows, 6 Intl. Conf. Num. Ship Hydro., U.Iowa, Iowa City, 1993.
th

[11] J. N. Newman, The Theory of Ship Motions, Advances In Applied Mechanics, 18.
[12] P.D Sclavounos, 1996. Computation of wave ship interactions, in Advances in Marine Hy-
drodynamics (Ed. M. Ohkusu), Computational Mechanics Publications
RAO S175 head waves (heave) Fr=0.275 RAO S175 head waves (pitch) Fr=0.275
3 8
_____
Time domain DM L
5
_______ Time domain NK
A 2 Frequency domain DM A Frequency domain NK
Betram RSM Fully 3D 6 Bertram RSM fully 3D
1.5 Experimental Data Experimental Data
4
1
2
0.5

0 0
1 2 3 0 L/g 4 1 2 3 4
0 L/g

5
3
100 200
Time Domain DM
Frequency domain DM
50 100 Bertram RSM fully 3D
Experimental Data
0
0
-50
Time domain DM
Frequency domain DM
-100 Bertram RSM fully 3D -100
Experimental Data
-150
-200
1 2 3 4
0 L/g 1 2 3 0 L/g 4

Figure 9: Heave (left) and pitch (right) response amplitude operator for a S-175 hull in head
waves (Fr = 0:275). The present free motion transient test (solid lines) are compared with
the experimental results. Linearized (double model) frequency domain results and fully three{
dimensional frequency domain computations from [2] are also reported.

3 RAO S175 following waves (heave) Fr = 0.275 5 RAO S175 following waves (pitch) Fr = 0.275

A 1 TiDOM
kA 1.2 TiDOM
Experimental Data 1 Experimental Data
0.8
0.8
0.6
0.6
0.4 0.4
0.2 0.2

0 0
1 2 3 4 0 1 2 3 4
0 L/g 0 L/g

3
50 5
150
0 100

-50 50
0
-100
-50
TiDOM TiDOM
-150 -100
Experimental Data Experimental Data

-200 -150
0 1 2 3 4 0 1 2 3 4
0 L/g 0 L/g

Figure 10: Heave (left) and pitch (right) response amplitude operators for a S-175 hull in follow-
ing waves (Fr = 0:275). The numerical results (solid lines) are compared with the experimental
data.

You might also like