Proceedings of the Twenty-first (2011) International Offshore and Polar Engineering Conference

Maui, Hawaii, USA, June 19-24, 2011

Copyright 2011 by the International Society of Offshore and Polar Engineers (ISOPE)
ISBN 978-1-880653-96-8 (Set); ISSN 1098-6189 (Set); www.isope.org

On the Numerical Simulation of a Piston-type Wavemaker

H.B. Gu1,2, D.M. Causon1, C.G Mingham1, L. Qian1, Han-Bao. Chen2
1 Centre for Mathematical Modelling and Flow Analysis
School of Computing, Mathematics and Digital Technology
The Manchester Metropolitan University,
Manchester, UK
2 Tianjin Research Institute for Water Transportation Engineering,
Key Laboratory of Engineering Sediment of Ministry Communications,
Tianjin, China

ABSTRACT
Havelock (1929) derived an analytical solution for waves generated by
piston and flap wavemakers based on linear wave theory. Ursell et al.
(1960) and Flick and Guza (1980) experimentally verified a piston
wavemaker by using first order wavemaker theory. Ottesen-Hansen et
al. (1980), Sand (1982) and Sand and Donslund (1985) discussed
second-order effects such as long waves in experimental models.
Schaffer (1996) derived a complete mathematical model for piston
wavemakers to second order. The linear wavemaker theory has been
extended to a 3D wave basin (Liu 1994, 1996; Newman 2010). In the
implementation of wave making in the laboratory active absorption is
often applied to avoid spurious reflection (Spinneken and Swan 2009a,
2009b).

The piston-type wavemaker is a very popular and important piece of

equipment in coastal and ocean engineering physical wave model
experiments so in this paper we use a numerical model to simulate the
characteristics of the piston-type wavemaker. The numerical model is
based on solving the Navier-Stokes equations, in which a two-fluid
water and air system is adopted. The interface between the water and
air is tracked by a particle level set method and the partial cell
technique combined with the local relatively stationary method is used
to simulate the motion of the wavemaker in the numerical model.
Firstly, linear wave maker theory is implemented and tested for regular
waves at various water depths, wave periods and wave heights, to see
which wave making methods give reasonable results for short wave
cases. The method is then extended to simulate irregular waves where
the wave spectrum is specified and the individual wave components are
superimposed. The output spectrum from a wave gauge in the
numerical flume compared well with the target spectrum. Finally a
focused wave is simulated by the numerical model and compared with
published results. It is shown that the numerical flume can simulate an
experimental flume with a piston-type wavemaker and thus physical
tests can be accurately replicated by numerical modelling.

Because the size of experimental models is limited by the size of wave

tanks leading to scaling effect errors, numerical wave flumes began to
be considered as a possible tool to support the design and regulation of
costal, ocean and near shore structures. Numerical wave models based
on nonlinear shallow water equations (SWE) can be found in Van Gent
(1994), and Hu et al. (2000). Zhang (2005) built a numerical model
based on the Boussinesq equations and compared numerically
generated waves to physical results. As models based on the previous
methods do not show the detailed information in the vertical direction
advanced numerical wave flumes should be based on the Navier-Stokes
equations (NSE).

KEY WORDS: Wavemaker; regular wave; irregular wave; focused

wave; level set; partial cell; local relatively stationary method.

The main challenge in solving NSE for wave generation is how to

locate the free surface. There are two approaches for solving the NSE:
mesh based methods and meshless methods. Meshless methods include
the smoothed particle hydrodynamic (SPH) and moving particle semiimplicit (MPS) methods. Shao et al. (2006) presented an
incompressible SPH model to investigate wave overtopping in coastal
structures. Koshizuka et al. (1995) proposed the MPS method to solve
the NSE used for the simulation of numerical wave flumes.

INTRODUCTION
Wavemakers are important pieces of equipment in coastal and ocean
engineering laboratory experiments to study wave interaction with
structures such as wharves, breakwaters and any other near-shore or
ocean structures. Generation of waves is one of the most significant
tasks in this kind of laboratory.
The most common way in a physical experimental flume to generate
waves is through the movement of a paddle, which is located at one end
of the flume. Paddles used in flumes can be a flap, a piston or a wedge
type, of which the piston-type is the most popular permitting simple
generation of shallow water waves according to the velocity pattern
near the paddle.

Mesh based methods include the volume of fluid (VOF) method (Hirt
& Nichols 1981, Yongs 1982, Ubbink 1997), the level set method (Gu
et al.2009, 2010; Chen 2009) and the marker and cell method (MAC)
(Harlow & Welch 1965) to capture or track the free surface together
with a NSE solver such as the projection or SIMPLE method. Another

652

 ( u i )
= 0
xi

recent mesh based method is the surface capturing method proposed

initially by Kelecy & Pletcher (1997) and Pan & Chang (2000), based
on the artificial compressibility method with high resolution Riemann
solvers, in which the free surface is treated as a contact discontinuity in
the density field. The moving interface is captured automatically as part
of the numerical solution along with other flow variables. Qian et al.
(2003, 2006) and Gao (2007) used this method to study wave run up on
a beach, wave overtopping at a vertical wall and fluid interaction with
moving bodies.

(u i )
(u i )
1 ij
p
+ u j
=
+ g i +
t
x j
( ) x i
( ) x j

EQUATIONS

AND

(8)

When =1 the above equations reduce to equations (1) and (2).

Numerical solution in the fluid domain

In the spatial domain, a staggered-grid system is used within which
scalars are defined at cell centres and vectors are defined at the centres
of cell faces. Finite differences are used to discretize the spatial
derivatives on a non-uniform mesh. A two-step projection method is
used to solve equations (7) and (8). A combined upwind and central
scheme is used to discretize the convection terms and a central
difference scheme is used for the diffusion terms (Lin 1998). Then the

The numerical simulation of fluid interaction with moving bodies

makes the simulation of the motion of different wave makers possible.
We have implemented the Fast-Marching particle level set method for
hydrodynamics numerical simulations (Gu 2009, 2010) combined with
a partial cell treatment and the local relative stationary method to
simulate moving bodies (Lin 2007). In this paper, the numerical model
is applied to wave generation by a piston-type wavemaker. In the next
section we briefly describe the numerical model. Then, results for the
case of regular wave generation, irregular wave generation and focused
wave generation are described. Finally some conclusions are drawn.

GOVERNING

(7)

tentative velocities ui are obtained from

~

u i uin
t

= u j

1 ij
(ui )
+
( ) x j
x j

(9)

In general, the tentative velocities do not satisfy the continuity equation

(7). Thus, it has to be updated by the correct pressure information,

NUMERICAL

u in +1 u i

IMPLEMENTATION
where the p n +1

Basic governing equations

1 p n +1
(10)
=
+ g i
( ) x i
t
is obtained by solving the Poisson pressure equation,
~

1 p n +1 1 u i

(11)

=
(g i )
+
x i ( ) x i t x i
x i
Equation (11) is obtained by enforcing equation (7) at the time step
n+1, and (uin+1)/xi=0, ensures the velocities calculated from
equation (10) are divergence free.

The equations of motion for an incompressible two-phase flow can be

written as
u
x
u i
+u
t

= 0

(1)

ij

u i
1 p
1
=
+ gi +
x j
( ) x i
( ) x j

(2)

Local relative stationary method for a moving partial cell

where i=1, 2 for 2D and i=1, 2, 3 for 3D. ui is the velocity component
in i direction, p is pressure, and ij=2()ij , ij is defined as the rate of
deformation tensor.

ij

1
2

u i
u j

+
x
xi
j

There are two major issues that need special attention in the numerical
implementation for a moving body. One is tracking the moving body
and the other is to ensure mass and momentum conservation (Lin
2007).

(3)

In equation (2) is the level set function defined as a distance function

that is positive in the water and negative in air. With this definition of
the level set function, the level set transport equation is
(4)
t + U = 0
where U is the velocity vector. The zero level set is initialized as the
free surface between water and air. From the level set function, the
density () and the viscosity () are written as
( ) = a (1 H ( )) + w H ( )
(5)
( ) = a (1 H ( )) + w H ( )
(6)
where H() is Heaviside function, w and w are density and viscosity
of water, a and a are density and viscosity of air (Gu 2009).

In this study, a rigid body is considered only. The body surface is

expressed by the zero set of a combination of quadratic functions e.g.
Fm(x,y,tn)=0. The new body surface can be determined by geometric
principles at time step n+1, e.g. Fm(x,y,tn+1)=0. The openness function
n+1 can also be updated by a combination of quadratic functions.
To ensure the conservation laws are satisfied during body movement
from one time step to another the following treatment is adopted. We
assume that during the small time interval t the body moves with a
constant velocity uiB, and the moving body problem can be recast into
an equivalent problem where the body is held steady and the fluid
motion is superimposed by means of an additional velocity with the
same magnitude as the moving body but in the opposite direction.
Thus, the continuity equation (7) in the partial cell is revised as follows,

Partial cell technique

(ui uiB )
=0
xi

Normally when a body lies inside the computational domain a

boundary condition is required at the solid surface. However, using the
partial cell technique (PCT) removes the need for a complex boundary
condition (even if the body is a complex surface). The main idea of the
PCT is to define an openness function as the ratio of water volume to
computational cell volume. That is, the whole computational domain is
divided into fluid cells, solid cells and partial cells. When =1 the cell
is a fluid cell; =0 a solid cell; and for 0< <1 a partial cell. With the
openness function so defined, the equations of motion for
incompressible two-phase flow become

=>

(ui ) (uiB )
=
xi
xi

(12)

Substituting the above modified continuity equation into the two-step

method, a modified Poisson pressure equation can be obtained,

xi

n +1 1 p n +1 1

=
( ) xi t

n +1 ui uiB

+ (g )
i
xi
xi

(13)

It is clear that a moving body will essentially change the pressure field
around the body which consequently affects the adjacent flow. Besides

653

the pressure field, another effect of a moving body in a viscous fluid

flow involves obviously the viscous stresses. This is considered in the
first step of the projection method when the convection and diffusion
terms are evaluated. The local relative stationary method is still applied
to calculate the relative velocity gradient near the body. Additional
terms will result that modify the normal and shear stress around the
moving body.

Level set boundary condition for obstacles

In our numerical model, the level set function on the boundary of fixed
or moving bodies must be carefully handled. Several researchers have
described approaches to this problem. One approach is to consider the
angle between the obstacle surface normal direction and water surface
normal direction (Yang & Stern 2009). By setting a constant contact
angle a formula can be found. This could be seen to be artificial and
non-physical. Another approach involves interpolation from nearby
cells which may in principle be better. In our model the water surface
normal direction is used according accurately with the wetted contact at
the obstacle surface. The normal unit vector is easy to obtain from the
level set function. Assuming the level set function at point m with
coordinate (xm, ym) near a boundary point n can be expressed as
m = axm + bym + c
(17)
and is a tangent at that point m is its level set value and c is a

The advantages of the local relative stationary method are that it is very
easy to implement numerically and to deal with multiple bodies without
limitation on their number since a moving coordinate system is
established at an individual cell. No re-meshing is required and the
fluid computation is based on a fixed-grid system. The advancement of
the body is solved by a Lagrangian method and thus is free of
numerical diffusion. Its extension to three-dimensions and more
complicated motion than shown here e.g. with rotation is
straightforward.

constant, (a, b) is the normal unit vector which can be calculated from
/ . The boundary point n has coordinates (xn, yn). From the above
formula the level set value of point n can be obtained by
n = axn + byn + c
(18)
The boundary condition for the level set is also confined in the narrow
band.

Solution of the level set equation

In the present study, we use a semi-Lagrangian method (Enright 2005)
to solve the level set transport equation (4). Here we define the level set
function as a distance function (x,t)= min |x-y |, y (t). The
level set function has the following properties, (x,t)>0 for x and
(x,t)<0 for x , where =0 is included with negative values so
that it is not a special case. The interface lies between >0 and <0 and
can be identified as (t)= {x (t): (x,t)= 0}. is the calculation
domain, and is the interface. Note that is a scalar function. The
distance function in 2D at the n+1 time step based on non-uniform grids
can be expressed as,
in, +j 1 = rn+1,s +1 + (1 ) rn, s +1 + (1 ) rn+1,s + (1 )(1 ) rn,s
(14)
where r and s are the left bottom corner index of the grid cell and the
new grid point value comes from that at the n time step according to the
velocities ui,j and vi,,j. Parameters and are the rate of change of the
distance to the left bottom corner in x and y direction respectively. The
formula can be easily extended to 3D. This scheme is unconditionally
stable so the time step is no longer limited by the CFL condition.

SIMULATION OF WAVE-MAKER AND ITS NUMERICAL

RESULTS
In this section, a piston-type wavemaker is considered as a moving
body in the numerical model. A sketch of the numerical model is
shown in Figure 1. The wavemaker is at the left side of the numerical
flume, which is 6L long, and at its right side the radiation boundary
condition is set to let waves pass freely out of this boundary. In the
domain there is water in the lower part and air in the upper part and a
solid wall boundary is imposed on the bottom and the top boundary. A
wave gauge is set in the middle of the flume to record the free surface
height in the simulation.

Regular waves
In our implementation we only calculate the distance function within a
narrow band around the interface. The narrow band is set to a width of
about 4 to 6 times the spatial step. At each time step, the zero level set
must not pass out of this band which places a restriction on the
maximum time step. To ensure the signed distance maintains the signed
distance property, we directly solve the equation
= 1
(15)
to re-initialize the level set function. In order to obtain the solution of
equation (15) in the shortest time, the fast marching approach (Sethian
1998) is used and the particle level set method (Enright 2002) used to
correct the level set function during the calculation. Thus, the level set
calculation is confined to a narrow band around the free surface.

According to linear wavemaker theory described by Dean and

Dalrymple (1984), the moving velocity of the wave board can be
calculated by the following equation for a specified wave.
U=

T ( )

X
t

(19)
where X is the position of the wavemaker, is the angular wave
frequency, is the given wave elevation, and T() is the transfer
function between the motion of wavemaker and the given wave. The
transfer function is expressed as
T ( ) =

4 sinh 2 (kd )
2kd + sinh(2kd )

(20)
where k is wave number corresponding to and d is the still water
depth.

Heaviside function
The Heaviside function is a smooth function, which makes the density
or viscosity smoothly vary from a to w or from a to w within a
narrow band of the interface. There are several expressions of
Heaviside function. They have the same role in principle. We adopt the
following formula,
< 1 .5 h
0,

1
2

H = 0 .5 +
sin
,
+
1 .5 h
(16)
3h 2
3h

> 1 .5 h
1,
in which h is the local minimum size of the spatial step.

To verify the effect of the linear wave theory, results from many test
cases are given in Table 1. In the table we listed the wave conditions in
the first three columns and calculated the parameters of wave length L0
in deep water, wave length L in shallow water, wave celerity c and
wave number k in the following columns. All the regular waves are
generated by the piston wavemaker and example wave histories are
shown in Figure 2 and Figure 3. From wave time histories, a zero-upcross method (Yu 2000) was used to measure wave height, which is
listed in the final column of Table 1. Figure 2 shows the case of wave

654

height 0.05m, wave period 1.0s and water depth 0.5m. Wave heights
are almost constant after 5 wave periods. Figure 3 shows the case of
wave height 0.08m, wave period 1.5s and water depth 0.4m. Again
wave heights are almost constant after 5 periods.

Tp =

TH 1 / 3
1 0.132( + 0.2) 0.559

(23)

and =3.3, =0.07 if <=p; =0.09 if >p.

Table1. Regular wave test cases

air

d(m)
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.25
0.25
0.25
0.25
0.25
0.25
0.25
0.25
0.25

Wave
board

x
d

water

eta (m)

Figure 1: Sketch of the numerical set up

0.04
0.03
0.02
0.01
0
-0.01
-0.02
-0.03
0

10

15

20

time (s)

Figure 2: Wave time history for wave height =0.05m, water depth =
0.5m, and wave period =1.0s
0.06
eta (m)

0.04
0.02
0
-0.02
-0.04
-0.06
0

12

16

20

24

tims (s)

Figure 3: Wave time history for wave height =0.08m, water depth =
0.4m, and wave period =1.5s
Table 1 shows that simulated wave heights are all smaller than
corresponding target ones, especially when the water depth is shallow,
wave period is long and wave height is large. The reason is that the test
cases are not completely linear waves and the non-linearity produces
errors. For highly non-linear waves non-linear wavemaker theories are
required to generate accurate waves but for weakly non-linear waves
linear wavemaker theory can give quite accurate results. In the next
section we use linear wavemaker theory to simulate irregular waves and
focused wave under conditions of small wave height, short wave period
and relatively deep water.

L (m)
1.51
1.51
1.51
2.83
2.83
2.83
4.06
4.06
4.06
1.46
1.46
1.46
2.62
2.62
2.62
3.70
3.70
3.70
1.30
1.30
1.30
2.17
2.17
2.17
3.00
3.00
3.00

L0 (m)
1.56
1.56
1.56
3.51
3.51
3.51
6.24
6.24
6.24
1.56
1.56
1.56
3.51
3.51
3.51
6.24
6.24
6.24
1.56
1.56
1.56
3.51
3.51
3.51
6.24
6.24
6.24

c(m)
1.51
1.51
1.51
1.88
1.88
1.88
2.03
2.03
2.03
1.46
1.46
1.46
1.74
1.74
1.74
1.85
1.85
1.85
1.30
1.30
1.30
1.45
1.45
1.45
1.50
1.50
1.50

k
4.15
4.15
4.15
2.22
2.22
2.22
1.55
1.55
1.55
4.29
4.29
4.29
2.40
2.40
2.40
1.70
1.70
1.70
4.82
4.82
4.82
2.89
2.89
2.89
2.09
2.09
2.09

Hc(m)
0.047
0.067
0.082
0.047
0.072
0.085
0.048
0.072
0.086
0.046
0.066
0.078
0.046
0.071
0.081
0.047
0.071
0.084
0.043
0.065
0.075
0.044
0.068
0.076
0.044
0.067
0.079

(t ) = ai cos(2fit + i )
i =1

(24)

where (t) is the water free surface elevation, N is the number of the
wave components with frequency fi and amplitude ai, and i is a random
phase of the ith wave component.
When a target irregular wave with its significant wave height and
corresponding frequency is specified, the wave spectrum can be
expressed by (21). From the wave spectrum wave component with
frequency fi and amplitude ai can be analyzed. Each wave component
can be generated by the wavemaker as a regular wave. Superposition of
all wavemaker motions for the wave components an irregular wave can
be generated.

Most natural waves are irregular. A wave train includes many waves
with different frequencies and amplitudes. Mathematically, there are
several methods to represent an irregular wave train. One of the
methods uses the wave spectrum which describes wave energy
distribution with frequency for specified conditions. Many wave
spectra have been proposed of which the modified JONSWAP (from
the Joint North Sea Wave Project) spectrum (Goda 1999) is often used.
The wave spectrum is easily described if the significant wave height
H1/3 and its corresponding frequency TH1/3 are given.

To test the irregular wavemaker, a case of significant wave height

0.08m and corresponding wave period 1.5s is generated with water
depth 0.4m. The target wave spectrum is described by the modified
JONSWAP spectrum (Goda 1999) as shown in Figure 4 by dotted
line. There are 50 wave components analyzed from the spectrum with
frequency from 0.0047 to 1.88, its peak frequency is 0.623. Numerical
simulation is run up to 150s with a wave gauge to record the wave
profile which is located at a distance of three wavelengths (based on the
significant wave period) from the paddle. Figure 5 shows the wave
history from which the statistically estimated significant wave height is
0.072m which is calculated according to the up-cross-zero method and

5
(exp( ( f / f p 1) 2 / 2 / 2 ))
S ( f ) = H12/ 3Tp4 f 5 exp( (Tp f ) 4 )
(21)
4
0.06238(1.094 0.01915 ln )
0.230 + 0.0336 0.185(1.9 + ) 1

Ht(m)
0.05
0.08
0.10
0.05
0.08
0.10
0.05
0.08
0.1
0.05
0.08
0.10
0.05
0.08
0.1
0.05
0.08
0.10
0.05
0.08
0.10
0.05
0.08
0.10
0.05
0.08
0.10

Another method considers waves in the time domain. An irregular

wave can be expressed by superposition of waves with different
frequency and amplitude,

Irregular waves

where =

T(s)
1.0
1.0
1.0
1.5
1.5
1.5
2.0
2.0
2.0
1.0
1.0
1.0
1.5
1.5
1.5
2.0
2.0
2.0
1.0
1.0
1.0
1.5
1.5
1.5
2.0
2.0
2.0

(22)

655

=0.3 and 0.9 with the peak frequency fc= 0.83. The focusing amplitude
A, which represents the linear sum of the component wave amplitudes
( A= ai), is set to 0.06m by adjusting input significant wave height.
The simulation ran for 20s.

is a little smaller than the target value. Based on the wave history a new
spectrum can be estimated by autocorrelation analysis, which is called
the measured wave spectrum shown in Figure 4 in a black solid line.
Comparing the measured and target spectra, the peak frequency and
spectral shape are very close to each other. The estimated spectrum has
some oscillations around the peak frequency. These may be caused by
numerical errors in the simulation and the discretisation of the target
wave spectrum. However errors are small and so the numerical model
can be used to simulate irregular waves.

A wave gauge is set at the target focus position to record the wave
profiles. Figure 6 and Figure 7 show the free surface elevation with
frequency bandwidths 0.3 and 0.9 respectively. Wave profiles are
similar to results from Zhao and Hu (2010). Comparing Figure 6 to
Figure 7, the frequency bandwidth has an effect on focused wave
profiles. The wider frequency bandwidth case gives a more clearly
defined extreme wave. Therefore the wave frequency bandwidth should
be carefully selected in practical application.

0.0004
Targeted
Simulated

0.00035

spectrum density

0.0003

CONCLUSIONS

0.00025

Linear wavemaker theory is implemented in the numerical model. The

model is based on solving the NSE and using a level set method to
track the water free surface in conjunction with the partial cell method
combined with the local relatively stationary method to simulate the
motion of the wavemaker. A piston-type wavemaker is used to generate
regular waves, irregular waves and focused waves. It is shown that
linear wavemaker theory is suited to simulate short wavelength regular
waves and can also be used to approximate irregular waves and focused
waves. For more accurate wave simulation high order wavemaker
theory will be tested in the future.

0.0002
0.00015
0.0001
0.00005
0
0

0.5

1.5

2.5

frequency f

Figure 4: Wave spectrum targeted and estimated

1.2
0.8

The mechanism of wave focusing in two dimensions is related to wave

dispersion. When rapidly travelling long waves catch up with slowly
travelling short waves, a large amplitude wave can appear at some fixed
location or time due to the superposition of wave components. Then an
extreme wave is expressed as

0.4
eta/A

Focused waves

-0.4
-0.8

Nf

( x, t ) = ai cos(ki ( x x p ) 2fi (t t p ))
i =1

(25)

-1.2
5

where xp and tp are focus position and time respectively, Nf is the

number of wave components with frequency fi and amplitude ai and
wave number ki. The number of wave components is chosen as 50 to
approximate a continuous wave spectrum. Wave components are equispaced across a bandwidth of df and centered at frequency fc. The wave
density distribution is calculated by the modified JONSWAP
spectrum (Goda, 1999).

10

11

12

13

14

15

time (s)

Figure 6: Focused wave with df = 0.3, A=0.06m

1.2
0.8

eta/A

0.4

0.08

0
-0.4

0.06
0.04
eta (m)

-0.8

0.02
0

-1.2

-0.02

-0.04

10

11

12

13

14

15

time (s)

Figure 7: Focused wave with df = 0.9, A=0.06m

-0.06
0

10

20

30

40

50

60

70 80

90 100 110 120 130 140 150

time (s)

REFERENCES

Figure 5: Irregular wave time history

Chen Y.G. Price W.G. (2009) "Numerical simulation of liquid sloshing

in a partially filled container with inclusion of compressibility effects,
" Physics of fluids, Vol 21
Enright D., Fedkiw R., Ferziger J., Mitchell I. (2002) "A Hybrid
Particle Level Set Method for Improved Interface Capturing,"
Journal of Computational Physics 183, 83-116
Enright D, Losasso F, Fedkiw R. (2005) "A fast and accurate semi-

In this section, we use the numerical piston wavemaker to simulate the

focused waves in tests cases conducted by Zhao and Hu (2010). The
schematic of the setup is similar to that of Figure 1. The length of the
flume is 20m, with water depth of 0.4m. The focusing position is set at
5m away from the wave paddle and the focusing time is 10s.
Computations are conducted with two different frequency bandwidth df

656

Lagrangian particle level set method, " Computers & Structures 83,
479-490.
Flick R.E., Guza R.T. (1980) "Paddle generated waves in laboratory
channels, " Journal of Waterway, port,coastal and ocean engineering,
American society of civil engineers, 106(1), 79-97.
Gao F., Ingram D.M., Causon D.M., Mingham C.G. (2007) "The
development of a Cartesian cut cell method for incompressible viscous
flows, " Int. J. Numer. Meth. Fluids 54, 1033-1053.
Goda Y. (1999) "A corporative review on the functional forms of
directional wave spectrum," Coastal engineering 41(1), 1-20.
Gu H.B., Causon D.M., Mingham C.G. & Qian L. (2009) "A FastMarching Semi-Lagrangian Level Set Method for Free Surface flows,
" ISOPE-2009
Gu H.B., Causon D.M., Mingham C.G. & Qian L. (2010) "High
Resolution Computation of Free Surface Flows Using a Level Set
Approach," ISOPE-2010
Harlow F, Welch J. (1965) "Numerical calculation of time-dependent
viscous incompressible flow, " Phys of Fluids 8, 2182-2189.
Havelock T.H. (1929) "Forced surface-wave on water," Philosophical
Magzine, 8(7), 569-576.
Hirt C.W., Nichols B.D. (1981) "Volume of fluid (VOF) method for
dynamics of free boundaries," J. Comput. Phys 39, 201-225
Hu K., Mingham C.G., Causon D.M. (2000) "Numerical simulation of
wave overtopping of coastal structures using the non-linear shallow
water equations," Coastal engineering 41(4), 433-465.
Kelecy F.J., Pletcher R.H. (1997) "The development of a free surface
capturing approach for multidimensional free surface flows in closed
containers," Journal of computational physics, 138, 939-980
Koshizuka S., Tamako H., Oka Y. (1995) "A particle method for
imcompressible viscous flow with fluid fragmentation,"
Computational fluid dynamics journal 4, 29-46.
Lin P. (2007) "A fixed-grid model for simulation of a moving body in
free surface flows," Computers & fluids l 36, 549-561
Lin P., Liu P.L.F. (1998) "A Numerical Study of Breaking Waves in the
Surf Zone," J. Fluid Mech. 359, 239-264.
Liu S. (1994) "The method and properties of waves generated by
serpent-type wavemakers," The ocean engineering 12(1), 81-93.
Liu S. (1996) "Extension of the effective test area of multi-directional
waves by side wall reflections," The ocean engineering 14(2), 40-50.
Newman J.N. (2010) "Analysis of wave generators and absorbers in
basins," Applied ocean research 32, 71-82.
Ottesen-Hansen N.E., Sand S.E., Lundgren, H., Sorensen T., Gravesen
H. (1980) "Correct reproduction of long group induced waves,"
Proceedings of the 17th coastal engineering conference, Sydney,
Australia, 784-800.
Pan D., Chang C.H. (2000) "The capture of free surface in
incompressible multi-fluid flows," Int. J. Num. Meth. Fluids 33, 203222
Qian L., Causon D.M., Ingram D.M. & Mingham C.G. (2003) "A
Cartesian cut cell two-fluid solver for hydraulic flow problems,"
ASCE Journal of hydraulic engineering 129 (9), 688-696
Qian L., Causon D.M., Mingham C.G. & Ingram D.M. (2006) "A freesurface capturing method for two fluid flows with moving bodies,"
Proc. R. Soc. A, 462, 21-42.
Sand S.E. (1982) "Long wave problems in laboratory models," Journal
of Waterway, port,coastal and ocean engineering, 108(WW4), 492503.
Sand S.E., Donslund B. (1985) "Influence of the wave board type on
bounded long waves," Journal of hydraulic research 23, 147-163.
Schaffer H.A. (1996) "Second-order wavemaker theory for irregular
waves," Ocean engineering 23, 47-88.
Sethian J. (1998) "Fast marching methods," Dept. of mathematics,
University of California, Berkeley.
Shao S., Ji C., Graham D.I., Reeve D.E., James P.W., Chadwick A.J.

(2006) "Simulation of wave overtopping by an incompressible SPH

model," Coastal engineering 53(9), 723-735.
Spinneken J, Swan C. (2009a) "Second-order wave maker theory using
force-feedback control. Part I: A new theory for regular wave
generation," Ocean engineering 36, 539-548.
Spinneken J, Swan C. (2009b) "Second-order wave maker theory using
force-feedback control. Part II: An experimental verification of
regular wave generation," Ocean engineering 36, 549-555.
Ubbink O. (1997) "Numerical prediction of two fluid systems with
sharp interfaces," Thesis, Department of mechanical engineering,
Imperial college of science, technology & medicine Jannuary
Ursell F., Dean R.G., Yu Y.S. (1960) "Forced small amplitude water
waves: Comparation of theory and experiment," Journal of fluid
mechanics, 7(1), 32-53.
Van Gent M.R.A. (1994) "The modeling of wave action on and in
coastal structures," Coastal engineering 22(3-4), 311-339.
Yang J., Stern F. (2009) "Sharp interface immersed-boundary/level set
method for wave-body interactions," Journal of Computational
Physics 228, 6590-6616
Yu Y. (2000) "Random wave and its applications for engineering,"
Dalian university and technology press, pp 67-68.
Zhang H. (2005) "A deterministic combination of numerical and
physical models for coastal waves," PhD thesis, Department of
mechanical engineering, Technical university of Denmark.
Zhao X., Hu C. (2010) "Numerical simulation of focused wave
generation using CIP method," Proc 20th Intl Ocean and Polar Eng
Conf, Beijing, China, ISOPE, Vol 3,pp 596-603.

657