You are on page 1of 6

Proceedings of the ASME 2012 31st International Conference on Ocean, Offshore and Arctic Engineering

OMAE 2012
July 1-6, 2012, Rio de Janeiro, Brazil

OMAE2012-83170

COMPARISON OF TWO BASE-POTENTIALS FOR A SLENDER-BODY METHOD

Mahmoud Alidadi
Department of Mechanical Engineering
University of British Columbia
Vancouver, Canada

Sander Calisal
Faculty of Engineering
Piri Reis University
Istanbul, Turkey

ABSTRACT
The effects of two base-potentials on the accuracy of a
slender-body method are studied in this paper. In the formulation for this method which is developed for the slender ships, the
velocity potential is decomposed into a base-potential and a perturbation potential. Then using an order of magnitude analysis,
the three-dimensional flow problem is simplified into a series of
two-dimensional problems for the perturbation potential. These
two-dimensional problems are solved with the linearized free
surface boundary conditions, using a mixed Eulerian-Lagrangian
method. Finally for the two base-potentials, the numerical wave
elevation along a Wigleyull are compared with the experimental
results.

Numerous researchers at the University of British Columbia


adopted these methods for calculation of wave pattern and wave
resistance of slender ships (see [1, 2], [3], [4], [5]). Song and
Maruo [6] used a 2D+t method to calculate the bow impact and
deck wetness on a Wigleyhull. Tulin and Wu [7] numerically
calculated the divergent waves generated by a Wigleyhull using
a 2D+t method.
Development of the slender-body methods involves decomposition of the potential into a base-potential and a perturbation potential, and then using an order of magnitude analysis
to turn the three dimensional flow problem into a series of twodimensional problems which are faster to solve than the original
problem.
The terms neglected in the order of magnitude analysis have
to be as small as possible to have an accurate solution for the
flow field around a slender hull. However, their magnitude are
greatly related to the base-potential. This paper investigates the
effects of two base-potentials free stream and double body on the
accuracy of a slender-body method developed in the following
sections.

NOMENCLATURE
U Ship velocity
L Ship length
B Ship beam
D Ship draft
FR Froude number
Total velocity potential
Perturbation velocity potential
Wave height

FORMULATION FOR THE SLENDER-BODY METHOD


A Cartesian coordinate system (x, y, z) is employed with the
origin at the amidship of centerline and on the undisturbed free
surface. The z direction is vertically upward, and the x-axis coincident with the longitudinal axis of the ship, the bow of which
is oriented in the x-direction (see Fig. 1). This coordinate system
moves with the ship, with steady velocity U in the x-direction.
As the fluid is assumed inviscid and incompressible, the fluid

INTRODUCTION
The slender-body methods (e.g. 2D+t) have been adopted
extensively for flow calculation around slender ships at low
Froude numbers, where the flow nonlinearity is minimum.
1

c 2012 by ASME
Copyright

AP

motion is irrotational and can be specified by the velocity potential which satisfies the Laplace equation
2 = 0

(1)

where
U

2 =

2
2
2
+
+
x2 y2 z2

(2)
FP

The kinematic and dynamic boundary conditions for a steady


free surface represented by z = (x, y) are
x x + y y z = 0

(3)

1 2
( + y2 + z2 U 2 ) + g = 0
2 x

(4)

FIGURE 1.

1
((x + x )2 + (y + y )2 + (z + z )2 U 2 ) + g = 0
2

The boundary condition on the hull is


n = 0

HULL WITH BODY-FIXED COORDINATE SYSTEM

(11)

Similarly the body boundary condition (5) changes to


(5)
x nx + y ny + z nz = (x nx + y ny + z nz )

(12)

where
=

+
+
x y z

(6)
Slender-Body Simplification
If we define non-dimensionlized variables

and n is the normal vector on the hull pointing outwards.


The velocity potential is assumed to be composed of a
perturbation potential and a basis potential , i.e.

x = x/L
= +

(7)

y = y/B
z = z/D

where the basis potential satisfies the Laplace equation (1), and
z = 0

then one can show that the order of the terms in governing equation and boundary conditions are:

(8)

Then the perturbation potential must satisfy the Laplace


equation
2 = 0

xx = O
L2 x2
 

x = O
Lx
n
nx = O
Lx
 
x = O
Lx

(9)

Replacing potential from the Eqn. (7) in the kinematic and


dynamic boundary conditions results in
(x + x )x + (y + y )y (z + z ) = 0

(10)
2

yy = O
B2 y2
 

y = O
By
 
n
ny = O
By
 

y = O
By

zz = O
D2 z2
 

z = O
Dz
 
n
nz = O
Dz

(13)

c 2012 by ASME
Copyright

An order of magnitude analysis shows that

Comparing the order of magnitudes in Eqn. (13), we can conclude that

zz < zz

xx  yy , zz
x x  y y

After neglecting the nonlinear term

x2  y2 , z2
x nx  y ny , z nz

yz y < zz

for when O(D) O(B) and B/L  1.


These relations simplify the governing Eqn. (9) to
2 2
+ 2 =0
y2
z

the linearized kinematic boundary condition becomes


x x + (y + y )y z zz = 0

(14)

1 2 2 2
1
( + y + z U 2 ) + (y2 + z2 )+
2 x
2
(x x + y y + z z ) + g = 0

1
1 2 2 2
(x + y + z U 2 ) + (y2 + z2 )+
2
2
y y + g + zz z = 0 at z = 0

Free Surface Boundary Conditions in Lagrangian Form


For writing the kinematic and dynamic conditions in Lagrangian form the operator dtd is defined as
(16)
d

= x + (y + y ) + (z + z )
dt
x
y
z

dy
= y + y
dt

(23)

for the velocity in y direction.


Using Eqn. (22) the Lagrangian form for the linearized kinematic and dynamic boundary conditions (20), (21) at z = 0 respectively are

(18)

Since z = 0 at z = 0, and z = (x, y) then


zx = xz = yz = zy = 0

(22)

This choice for the operator leads to the equation

(17)

Linearization of the Free Surface Boundary Conditions


The kinematic and dynamic boundary conditions (15), (16)
can be linearized around z = 0. The Taylor expansion of the Eqn.
(15) up to the first order results in
x x + (y + y )y z z + (xz x + x xz )+
((yz + yz )y + (y + y )yz zz zz ) = 0

(21)

(15)

and the body boundary condition for a slender ship changes to


y ny + z nz = (x nx + y ny + z nz )

(20)

Using the same approach the linearized dynamic boundary condition becomes

The governing equation (14) shows that the flow around a slender
ship can be described by a series of two-dimensional problems in
the cross flow planes along x axis.
The kinematic and and dynamic boundary conditions at z =
(x, y) also convert to
x x + (y + y )y (z + z ) = 0

at z = 0

d
= z + zz
dt

(24)

zx = xz = zy = yz = 0

The above relations simplify the kinematic free surface boundary


condition to
x x + (y + y )y z + (yz y zz zz ) = 0

d 1 2
1
= (U (x2 + y2 )) + (y2 + z2 )
dt 2
2
g (y zy + zz z )

(19)
3

(25)

c 2012 by ASME
Copyright

SOLUTION APPROACH FOR THE SLENDER-BODY


METHOD
The solution approach developed by Longuet-Higgins and
Cokelet [8] is used for the implementation of the slender-body
method. This approach involves a two-step procedure divided
into an Eulerian boundary element method and a Lagrangian
scheme.

where the values for yt , t , and t are determined using Equations


(23), (24), and (25) and the time step dt is
dx
dt =
x

NUMERICAL RESULTS FOR A WIGLEYHULL


The slender-body method explained in the previous sections
was used to calculate the wave elevation along a Wigleyhull at
Froude numbers FR = 0.267 and FR = 0.316. These results were
obtained for the free stream and double body base-potentials, using a computer with Pentium D processor and 4GB of RAM. The
calculation for each of these Froude numbers took about 4 minutes in the case of the free stream potential and 10 minutes in the
case of double body potential.
The Wigleyhull is described by the equation

Boundary Element Method


The first step attacks the solution of Eqn. (14) in a cross flow
plane using the velocity potential 0 known on the free surface
from the previous time step, and the body boundary condition
(17). From the solution of Eqn. (14) the potential at a point q in
the cross flow plane is (see [9])

(q) =

1
2

(
S

ln r
ln r)ds
n
n

(26)

4x2
z2
B
y = Y (x, z) = (1 2 )(1 2 )
2
L
D

where S is the boundary for the cross flow plane and r is the
distance from point q.
The boundary S is discritized into N elements with constant
distribution of n and over each element. For collocation
points on the elements, placed at the center of elements, the integral Eqn. (26) gives the linear algebraic equation
N

Ai j n | j + Bi j j = 0

i = 1, ..., N

where L = 2 m, B = 0.2 m and D = 0.125.


As the hull is symmetric, only half of the hull was considered in the calculation. An impermeable wall was placed 35B
away from the body to make sure waves does not reflect into the
computational domain. The step size was set to dx = 1 mm, and
10 equal length elements were used on the hull surface.
For the starting cross section, the values for the wave elevation and perturbation potential were set to zero. On the free surface of this cross section 75 panels were put which their lengths
increased toward the impermeable wall in a geometric progression, with the first panel size D/10 long.

(27)

j=1

where the coefficients Ai j and Bi j are


1
Ai j =
2

1
ln rds , Bi j =
2
j

Z
j

ln rds
n

Results for the Free Stream Base-Potential


For a ship moving with the free stream velocity U, the free
stream base-potential can be described as

Lagrangian Scheme
The second step involves calculation of the velocity potential
and the free surface for the next cross-flow plane. This marching
from one cross-flow plane to the next one can be viewed as a
time-domain problem where the time step is the amount of time
taken to traverse dx, the spatial step along the longitudinal axis.
Using the Euler method, the values for velocity potential and
free surface location at new cross flow plane are

x = Ux

(30)

This base-potential simplifies the Lagrangian equations (23),


(24), and (25) to
dy
=y
dt
d
=z
dt

d 1 2
= y + z2 g
dt 2

y(t + dt) = y(t) + yt (t)dt


(t + dt) = (t) + t (t)dt
(t + dt) = (t) + t (t)dt

(29)

(28)
4

(31)
(32)
(33)

c 2012 by ASME
Copyright

0.05

0.06
0.05

0.04
Experiment Results

Experiment Results
0.04

Resutls for Free-Stream Potential


Results for Double-Body Potential

0.03

Resutls for Free-Stream Potential


Results for Double-Body Potential

0.03
0.02

0.01

0.02
0.01

0
0
-0.01

-0.01

-0.02
-0.03
-1

-0.02
-0.8

-0.6

-0.4

-0.2

0.2

0.4

0.6

0.8

-0.03
-1

-0.8

-0.6

-0.4

-0.2

2X

FIGURE 2.
FR = 0.267

0.2

0.4

0.6

0.8

2X

FIGURE 3.
FR = 0.316

WAVE ELEVATION ON THE WIGLEYHULL SIDE AT

Numerical results for the free stream base-potential are compared


with the experimental results at Figures 2 and 3. The starting
cross section in these simulations was at the bow. The experimental results for FR = 0.267 was obtained from a research conducted at the department of Naval Architecture and Ocean Engineering at the Yokohama National University [10]. Measured
values for FR = 0.316 are obtained by Maruo and Song for a 6
m Wigleyhull model [11]. As the Figures show the numerical
results for free stream base-potential does not match well with
the experimental results. They especially deviate from the experimental results in the bow and stern area.

WAVE ELEVATION ON THE WIGLEYHULL SIDE AT

bow

x = 3L/4

stern

FIGURE 4. FREE SURFACE GRID FOR DOUBLE BODY POTENTIAL CALCULATION

Results for the Double Body Base-Potential


As Equations (23), (24), and (25) show, the slender-body
method with this base-potential requires the calculation of the
double body potential. We adopted the boundary element method
described by Katz and Plotkin [12] for this purpose, using 50
panels in x direction and 20 panels in z direction.
Adopting the double body potential as the base-potential had
advantage over the free stream base-potential as we were able to
calculate the potential and free surface at the bow position. The
starting cross section for this base-potential was at L/4 upstream
of the bow(i.e. x = 3L/4). Since calculation of the double body
potential for all cross sections was very time consuming, we calculated its values for a grid on the free surface (see Figure 4), and
then interpolated the values to obtain the double body potential
for a cross section. The panel size for this grid was L/50 in the
x direction except in bow and stern which was L/100. In the y
direction, 20 panels were used on this grid, where the panel size
increased in a geometric progression with the initial size D/20.

The wave elevation results along the hull for this basepotential are also presented in Figures 2 and 3. These results
match the experimental results better than the results for the free
stream base-potential. The values around the bow and stern regions are far better than those predicted for the free stream basepotential. This is because of two reasons:
a) The potential and free stream elevation at the bow were set to
zero for the free stream base-potential, while we were able
to predict those values using the double body base-potential.
b) Since the double body potential is the same as the potential
for a flow with FR = 0, it can be argued that the double body
potential is a better representation of the actual flow than the
free stream flow at low Froude numbers. Therefore one can
conclude that the terms ignored in development of the the
slender-body method (refer to pages 2 and 3) were smaller
for the double body base-potential.

c 2012 by ASME
Copyright

CONCLUSIONS
Two base-potentials free stream and double body were studied for a slender-body method developed for slender ships. The
velocity potential was decomposed into a base-potential and a
perturbation potential in this method. Then using an order of
magnitude analysis the three-dimensional steady flow problem
was transformed into a series of two-dimensional problems in
the cross-flow planes along x axis. The free surface boundary
conditions were then linearized around z = 0. A mixed EulerianLagrangian approach was used to implement the slender-body
method.
The slender-body method was used to calculate the wave
elevation along a Wigleyhull. The results for the double body
potential were better match with the experiments in comparison
with the results for the free stream potential. This is because a)
we were able to predict the potential and free surface elevation at
the bow for the case of double body potential, b) the values for
the terms ignored in development of the slender-body method
were smaller for the case of double body base-potential as it is a
better representation of the actual flow.

wave pattern around a yawed ship model. Graduation thesis, Yokohama National University, Japan.
[11] Maruo, H., and Song, W., 1990. Numerical appraisal of
the new slender ship formulation in steady motion. In Proceedings of the 18th Symposium on Naval Hydrodynamics,
pp. 239257.
[12] Katz, J., and Plotkin, A., 2001. Low Speed Aerodynamics,
second ed. Cambridge University Press.

REFERENCES
[1] Calisal, S., and Chan, J., 1989. A numerical calculation for
ship bow wave. Ship Research, 33(1), March, pp. 2128.
[2] Calisal, S., and Chan, J., 1993. A numerical procedure for
time domain nonlinear surface wave calculations. Ocean
Engineering, January, pp. 1932.
[3] Allievi, A., and Calisal, S., 1993. A semi-implicit, semilagrangian finite element model for nonlinear free surface
flow. In 6th International Conference on Numerical Ship
Hydrodynamics, pp. 195215.
[4] Wong, L., 1994. A numerical solution for potential flows
including the effects of vortex shedding. PhD thesis, University of British Columbia.
[5] Wong, L., and Calisal, S., 1996. Numerical algorithms for
slender bodies with vortex shedding and density stratification. Ship Research, 40(1), March.
[6] Song, W., and Maruo, H., 1993. ow impact and deck wetness: simulation based on nonlinear slender body theory.
In Proceedings of the 3rd International Offshore and Polar
Engineering Conference, Vol. III, pp. 3438.
[7] Tulin, M., and Wu, M., 1996. Diverging bow waves. In
21th Symposium on Naval Hydrodynamics.
[8] Longuet-Higgins, M., and Cokelet, E., 1976. The deformation of steep surface wavs, a numerical method for computation. In Proceedings of the Royal Society, Vol. A350,
pp. 126.
[9] Brebbia, C., and Dominguez, J., 1989. Boundary Elements:
An Introductory Course. Computational Mechanics Publication, Southamption, UK.
[10] Hayashi, K., and Kunishige, Y., 1988. Measurement of
6

c 2012 by ASME
Copyright

You might also like