You are on page 1of 13

Applied Ocean Research 30 (2008) 159171

Contents lists available at ScienceDirect


Applied Ocean Research
journal homepage: www.elsevier.com/locate/apor
Review
The interaction of an underwater explosion bubble and an elasticplastic
structure
A.M. Zhang

, X.L. Yao, J. Li
College of Shipbuilding Engineering, Harbin Engineering University, Harbin 150001, China
a r t i c l e i n f o
Article history:
Received 27 February 2008
Received in revised form
22 October 2008
Accepted 9 November 2008
Available online 20 December 2008
Keywords:
Underwater explosion
Bubble
Ring
Jet
Elastic-plasticity
Structure
Surface ship
a b s t r a c t
Based on the potential flow theory, the boundary element method (BEM) is applied to calculate the
dynamics of an underwater explosion bubble near boundaries, and in conjunction with the finite element
method (FEM) it is employed to compute the interaction between a bubble and an elasticplastic
structure. Acomplete 3Dunderwater explosion bubble dynamics code is developed; the simulated results
compare well with an underwater explosion experiment. With this code, the interactions between an
underwater explosionbubble andelasticplastic structures suchas a flat plate, a cylinder andother simple
structures are calculated and analyzed. Besides, the damages caused by the after flow, pulsating pressure,
and jetting load on the structures are also calculated, with or without a free surface. Fromthe time history
of the pressure and stress of the structure, it can be observed that the stress reaches its maximum value
when the bubble collapses, which proves that the pressure and jet impact induced by the collapse of
the bubble can result in severe damage to the structure. In particular, the 3D analysis code is applied to
some engineering problems, for example it is used on a surface ship to study the interaction between a
bubble and a complex elasticplastic structure. Under the bubble load, the low-order eigenfrequency of
the ship is aroused usually, leading to the so-called whipping effect, because the pulsating frequency of
the bubble matches the low-order eigenfrequency of the ship. The ship moves up and down with the
expansion and collapse of the bubble respectively. Meanwhile, the power of the bubble generated by a
near-field underwater explosion in short range is discussed, and some important conclusions which can
be applied to project application field are drawn.
2008 Elsevier Ltd. All rights reserved.
Contents
1. Introduction........................................................................................................................................................................................................................ 160
2. Theory and numerical model ............................................................................................................................................................................................ 160
2.1. Introduction ........................................................................................................................................................................................................... 160
2.2. Boundary-element method (BEM) for the fluid part........................................................................................................................................... 160
2.3. Toroidal bubble model .......................................................................................................................................................................................... 161
2.4. Time-step size control and numerical procedures .............................................................................................................................................. 162
2.5. Finite-element method (FEM) solver for the structural part and coupling with fluid part .............................................................................. 162
3. Results and discussions ..................................................................................................................................................................................................... 163
3.1. Circular plate model .............................................................................................................................................................................................. 164
3.2. Cylinder model....................................................................................................................................................................................................... 165
3.2.1. Comparison of the simulated results and experimental data.............................................................................................................. 165
3.2.2. The case with a free surface ................................................................................................................................................................... 165
3.2.3. The case with bubble under cylinder .................................................................................................................................................... 166
3.3. The interaction between a bubble and a complex elasticplastic structure (e.g. a surface ship) .................................................................... 167
3.3.1. Explosion in middle and far field........................................................................................................................................................... 167
3.3.2. Explosion in near field............................................................................................................................................................................ 169
4. Conclusions......................................................................................................................................................................................................................... 169
Acknowledgments ............................................................................................................................................................................................................. 170
References........................................................................................................................................................................................................................... 170

Corresponding author. Fax: +86 0451 82518296.


E-mail address: amanzhang@gmail.com (A.M. Zhang).
0141-1187/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apor.2008.11.003
160 A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171
1. Introduction
During an underwater explosion, there will be an initial shock
wave propagating radially outwards to be followed by a high-
pressure bubble containing hot gaseous products of the explosion.
Under the effects of hydrostatic pressure, gravity and inertia, the
so-called after flow, jet and pulsating pressure are developed
owing to the motion of the bubble. Not only the shock wave but
also the bubble load can have great damages on the underwater
structures. For instance, the pulsating pressure and after flow can
cause global damage on the structure (e.g. a surface ship), while
the high-speed re-entrant water jet will cause local damage to
the structure. Nowadays, researches are mainly focused on the
interaction between a bubble and a rigid wall (e.g. [17]), but
fewer published literatures are about the interaction between
an underwater explosion bubble and an elasticplastic structure.
Kalumuck and Chahine et al. [8] calculated the interaction between
a bubble and an elasticplastic structure with the combination
of the finite element method (FEM) and boundary element
method (BEM), and also developed the 2DYNAFS, 3DYNAFS and
other codes. Klaseboerk [9] studied the underwater explosion
bubble dynamics and the interaction between a bubble and a
simple flat plate numerically and experimentally. Based on their
achievements, this paper discusses the interaction between a
bubble and a complex elasticplastic structure (e.g. a surface
ship), taking the free surface into account simultaneously. The
damages caused by the after flow, pulsating pressure, jetting load
are investigated, aiming at revealing the power of the bubble load.
2. Theory and numerical model
2.1. Introduction
The numerical calculations can be split into two parts: the fluid
part and the structural part. The fluid part is carried out using
the boundary integral method. Special care must be taken after
the jet impact induced by the collapse of the bubble, since the
fluid domain then becomes doubly connected; a vortex ring is
placed inside the bubble to account for this phenomenon. The
structural part is solved using the finite-element solver ABAQUS. A
full coupling between the two codes has been made; information
concerning displacements of the structure is passed on from the
structural code to the fluid code and forces (deduced from the
pressure loading onthe structure) are passedonfromthe fluidcode
to the structural code. A controlling interface code decides which
code has to be called to do the appropriate calculations, making the
relevant results available as input to the other code. Time steps and
meshes can be different for the two codes.
2.2. Boundary-element method (BEM) for the fluid part
To investigate the dynamics of the underwater explosion
bubble, it is assumed that the surrounding fluid field is full
of ideal fluid which is inviscid, irrotational and incompressible.
Therefore the velocity potential is introduced and the velocity
vector can be derived from this potential as: u = . When it
is coupled with the continuity equation u = 0, the velocity
potential is governed by the Laplace equation:

2
= 0. (1)
As the Laplace equation (1) is an elliptic equation, the solution can
always be computedeverywhere inthe fluiddomain, providedthat
either the potential (Dirichlet condition) or the normal velocity
/n (Neumann condition) is given on the boundaries of the
problem. Here /n = n is the normal inward derivative of
the boundary S and n is directed out of the fluid. According to
Greens formula, the velocity potential of any point in the fluid
can be obtained by the velocity potential on the boundary and its
normal derivative, in other words, the function in the fluid field
can be described by laying distributing sources on the boundary
and distributing dipoles along the normal direction. The boundary
condition at infinity is of the form:
r =
_
x
2
+ y
2
+ z
2
, 0 (2)
where r = (x, y, z) is the position vector. Then Eq. (1) can be
written as:
(p) =
__
S
_
(q)
n
G(p, q) (q)

n
G(p, q)
_
dS. (3)
Eq. (3) is Greens integral formula, where S is the boundary of the
fluid domain including bubble surface S
b
, free surface S
f
and wall
surface S
w
; p and q are a fixed point and the integration variable
situated on S, respectively; is the solid angle viewed from the
point p : = 4 for an interior fluid point, c(p) = 2 for point p
on a smooth surface and c(p) < 4 for point p at the corner.
The solid angle subtended at the control point p by surface S can
be obtained through integral as follows:
=
__
S
G
n
(p, q) dS
q
, p S. (4)
The three-dimensional field Green function is givens:
G(p, q) = |p q|
1
. (5)
Ignoring the motion of the gas inside the bubble and its
corresponding influence on gas pressure, we suppose that the
gas pressure is only relative to the initial state and volume of
the bubble, and then the gas pressure inside the bubble may be
expressed in term of its volume V as:
p
b
= p
c
+ p
0
_
V
0
V
_

(6)
where P
c
is saturatedvapor pressure of noncondensable gas; P
0
and
V
0
are the initial pressure and volume of the bubble respectively,
referring to literature [6]; and is the ratio of specific heat of
the gas, which is equal to 1.25 for the gaseous explosion products
resulting from an TNT explosion [10] and 1.4 for ideal gas [11].
To generalize our research, a systemof non-dimensionalisation
is adopted here in which all the parameters are scaled: R
m
for
length, P = P

P
c
for pressure, R
m
(/P)
1/2
for time,
(P/)
1/2
for velocity and R
m
(P/)
1/2
for velocity potential.
Here P

= gH + P
atm
is defined as the hydrostatic pressure at
infinity on the O
xy
coordinate plane located at the initial bubble
center, where is the density of the fluid; g is the gravity
acceleration; H is the initial depth of the bubble center and
P
atm
is the normal atmospheric pressure; R
m
is the maximum
radius that the bubble would attain in an infinite fluid domain
under the pressure of P

.The surface tension is neglected in


this investigation, since it is insufficient to cause appreciable
effect for most of the lifetime of a cavitation or an underwater
explosion bubble [1113]. Based on the potential flow theory and
published literature, the non-dimensional Bernoulli equation on
the boundaries of the bubble, structure and free surface are as
follows (all variables are non-dimensionalized in the equations):
d
dt
= 1
_
V
0
V
_


2
z +
1
2
|u|
2
(On a bubble surface), (7a)
d
dt
= 1 P
2
z +
1
2
|u|
2
(On a structural surface), (7b)
d
dt
=
1
2
|u|
2

2
_
z
f
_
(On the free surface), (7c)
A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171 161
where and u are the velocity potential and velocity vector on
the boundary respectively; z is the displacement along the gravity
direction on the boundary, P is the pressure on the structure
surface, = P
0
/P is the dimensionless initial pressure parameter
inside the bubble and = (gR
m
/P)
1/2
is the dimensionless
buoyancy parameter,
f
= H/R
m
is the dimensionless initial depth
parameter. Providing that r is the spatial vector coordinate of a
fluid particle on the boundary, and then the motion equation of
the fluid particle on the boundary could be given as:
dr
dt
= . (8)
If the initial conditions are given, Eqs. (3), (7) and (8) forms a
completed set of equations for the motion of bubble and other
boundaries. Inorder to attain the initial conditions, a high-pressure
spherical bubble with radius R
0
, whose initial expanding velocity is
zero, is assumed to be formed at the initial stage of an underwater
explosion. During the early phase of the bubble motion, the effects
of the buoyancy and boundary can be ignored due to the bubbles
small size and the high-pressure gas inside, therefore the motion
of the bubble can be described by the Rayleigh equation [14]:

RR +
3
2

R
2
=
_
R
0
R
_
3
1. (9)
As long as an arbitrary initial velocity is given, the new initial
radius andpressure of the bubble canbe obtainedby integrating (9)
backwards in time. The method to determine the initial pressure
and radial velocity is presented in [9].
2.3. Toroidal bubble model
The simulation of the dynamics of the bubble consists of two
main stages, namely the pre-jet stage (singly-connected region)
and post-jet stage (doubly-connected region, Toroidal Bubble). For
the former stage, it can be solved with the method described
in Section 2.2; while for the latter stage, the bubble evolves
into a toroidal shape after jetting and the fluid becomes doubly-
connected, at which time the velocity potential on the surface of
the bubble may be a multiform function. So far, there have been
several axis-symmetrical models that can simulate the dynamics
of a toroidal bubble. Lundgren and Mansour [15] divided the
bubbles collapse into two phases. They dealt with the first phase
which lasted from the initiation to the moment of jetting with
a simple boundary element method. In the other phase, they
introduced a vortex line into their work. As a result, it not only
made the calculation of the bubble collapse to be prolonged, but
also the numerical simulation of the phenomenon that vortex was
produced in the anaphase of bubble collapse to be carried out;
while this method is limited to the simulation of the bubble having
a constant volume. Best [16] introduced a branch cut approach
and formulated a boundary integral equation valid for both on the
bubble surface and the branch cut. However, the deficiency of this
method is that the cutting bubble surface needs special disposal, so
it is hard to be popularized. In the process of simulation, Zhang &
Duncan et al. [17,18] defined a layer to separate the water jet and
the surrounding domain during the toroidal bubble phase, which
acts as a vorticity sheet and moves with the flow, and employed
modified boundary element method to calculate the whole process
of bubble collapse. But the deformation of the layer cant exceed
the bubble surface, which makes the tracking very challenging,
especially to the simulation of three-dimensional models. To solve
this problem, Wang et al. [11] employed a so-called surgical-cut
to convert the originally singly-connected bubble to a multiply-
connected toroidal bubble after jet impact. Instead of adding a
vortex sheet at the impact area, a vortex ring was placed inside the
Fig. 1. A toroidal bubble with the technique of surgical-cut: the identification of
the jet top node I and the impact node J ; I and J will be joined when the distance d
between them is less than some constant.
Fig. 2. Motion of a typical toroidal bubble: (a) shows the state when the bubble
begins to collapse and a jet just appears; (b) shows the state when the jet penetrates
the bubble and a doubly-connected bubble has been formed; (c) shows the state
when bubble rebounds and goes into the second period; the brown bar in the (b)
and (c) represents the vortex ring.
bubble to account for the double connectivity of the bubble. There
is no longer the need for meticulously tracking the vorticity sheet
as in[17]; we just ensure that the vortex ring stays inside the toroid
as the bubble evolves. The models above are all axis-symmetrical.
As for three-dimensional models, Zhang et al. [6] extended the
research of Wang et al. [11], putting the vortex ring into the
simulation of the three-dimensional toroidal bubble; based on this
model, the whole process of expansion, collapsing, jet formation
and rebounding of a three-dimensional bubble can be simulated.
Until now, most of the numerical simulations of 3D toroidal
bubble [19,13] are mainly based on the vortex-ring model of Zhang
et al. [6]. The 3D surgical-cut procedure is showed in Fig. 1.
This paper adopts the vortex-ring model posed by Zhang
et al. [6] too, in which the total potential is decomposed into
two parts: one is the potential associated with the circulation
generated by the impact, termed the ring potential; the other is
the remnant energy which is uniformly distributed in the entire
fluid domain. Then:
(r, t) = (r) + (r, t). (10)
Now the total potential consists of the ring potential and remnant
potential, and Eqs. (7) and (8) should be transformed into
corresponding forms, referring to Zhang et al. [6]. Because of
the severe instability of the bubble after instantaneous impact,
should be smoothed every several time steps to guarantee the
convergence of the model. Furthermore, the location of the vortex
ring should be updated continuously with the change of the bubble
shape so that the vortex ring can be inside the bubble all the time.
Fig. 2 shows the typical dynamic behavior of a toroidal bubble,
i.e. the transition of an underwater explosion bubble froma singly-
connected to multiply-connected domain.
Whats worse, the distortion of the mesh may occur in the
simulation of the three-dimensional bubble dynamics, therefore
a smoothing scheme is needed on the bubble surface and other
boundaries. Especially, after jet impact most elements concentrate
in the region where the jet is formed, which makes the meshes
close to it too dense, resulting in the abortion of the calculation. In
order to avoid this problem, an elastic mesh technology (EMT) [20]
is employed in this paper.
162 A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171
Fig. 3. Flow chart of the numerical algorithm (E. and B.C. are short for equation and boundary condition respectively in the chart).
2.4. Time-step size control and numerical procedures
To update the position and velocity potential of the boundaries,
the time-advancing scheme is adopted in current study. If the
velocity potential (t) and position r(t) of the boundaries for
time t (step i) are given, then the normal velocity component
/t(t) and material velocity u(t) can be obtained through
integral equation (3) and the finite difference method respectively.
Then the velocity potential (t + t) on the boundaries for next
time step (step i + 1) can be attained by Bernoulli equation (7).
Combining the finite difference method, the update of the velocity
potential is prescribed as:

i+1
=
i
+ (

i
+

i+1
)t/2. (11)
The position vector r(t + t) of the boundaries for next time step
(step i + 1) can be acquired by Eq. (8), and the update position of
the bubble boundaries would thus be of the form:
r
i+1
= r
i
+ ( r
i
+ r
i+1
)t/2. (12)
In Eqs. (11) and (12) t is the time-step size, which must be
controlled carefully to maintain the stability of the solution. In the
present paper the time-step size t is chosen as:
t = min {t
1
, t
2
} (13a)
t
1
=

max

1 +
1
2
||
2

2
z
_
V
0
V
_

(13b)
t
2
=

max

1
2
||
2

2
(z
f
)

(13c)
where is some constant andthe change of the velocity potential
on every node in each step should be not more than . A value of
0.02 is set for in the computation of current study, with which
the calculation is stable throughout. The code is implemented as
follows:
(a) Initialize the routine, i.e. read the initial information of the
nodes and elements on the structure and initial parameters of
the bubble such as R
0
, , ;
(b) Begin time stepping (assume the total time T
total
= 0,constant
= 0.02, step counter i = 1,and get
0
and t
0
through
Eq. (13) with the initial data);
(c) Solve the boundary integral equation (3); obtain the normal
velocity (/n)
i
on the node of the bubble and free surface
and also the distribution of the velocity potential
m
i
on the
structural surface; combine the infinite difference method to
solve the velocity vector u
i
on the node of the boundaries.
(d) Calculate the pressure P
m
i
on the fluid-structural interface(the
so-called wetted surface) with unstable Bernoulli equation
(7b),and impose the pressure load P
m
i
on the structure through
current solver ABAQUS with an interface routine;
(e) Solve the displacement r
m
i
and velocity r
m
i
of the structure
under new pressure load P
m
i
with finite element method in
solver ABAQUS;
(f) Update the boundary condition(B.C.) with the new velocity r
m
i
and displacement r
m
i
and apply new B.C. to the fluid boundary
integral code, perform step (c) to (f) until (/n)
i
= r
m
i
n
is satisfied on the wetted surface, then obtain the new normal
velocity (/n)
i
and new position r
i
of the boundaries;
(g) Get the stress, strain and other dynamic variables on the
structure to check the damages of bubble load (if necessary);
(h) Add t
i
to T
total
, update the location r
i
and velocity potential

i
of the bubble surface with Eqs. (7) and (8) and acquire next
time step t
i+1
with Eq. (13);
(i) Return to (c) and carry out next step i + 1 until T
total
is more
than certain time

T, and then the whole calculation ends.
To illustrate the process more clearly, the flow chart is shown as
Fig. 3:
2.5. Finite-element method (FEM) solver for the structural part and
coupling with fluid part
The fluidstructure interaction is usually calculated with two
methods [9,21]. One is the so-called full coupling method and the
other method is termed the loose coupling method. Full coupling
method solves all the unknowns simultaneously: (i) compute
the hydrodynamic loads applied on the structure from the fluid
and calculate the structural response; (ii) update the boundary
conditions based on the calculated displacement and velocity;
(iii) calculate the response of the fluid and the structure until the
velocity and displacement on the structural wetted surface match
the fluid motion. The loose coupling method is similar to the full
coupling one: the fluid and the structure are solved in stagger
A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171 163
iteration process, but the iteration calculations are completed in
two adjacent time steps instead of the same step. Both these two
methods are available in solving the fluidstructure interaction.
However, the full coupling method gets the full coupling solution
with huge calculation work; the loose coupling method is actually
an approximate method with incomplete solution, the solution of
which is usually not convergent. So it is not suitable for the long-
time numerical simulation.
The full coupling method is used in this paper incorporating the
BEM with FEM. The BEM is used to solve the bubble dynamics and
the pressure caused by the bubble motion, while the FEMis used to
solve the response of the structure (e.g. ship) under the pressure.
Several nonlinear FEM solvers such as PAM-CRASH, ABAQUS, LS-
DYNA, and MSC-DYTRAN can be used to calculate the structural
part, which have good precision and utility routines available. In
this paper, the nonlinear FEM solver ABAQUS is adopted to solve
the dynamic response of the structure according to the following
Eq. (14):
_
V
: dV +
_
V
r
m
r
m
dV +
_
V
r
m
r
m
dV
+
_
S
fs
Pr
m
ndS
_
S
f
r
m
fdS = 0 (14)
where and n are the stress and unit outward normal of the
structure respectively, the density of the material, the mass
proportional damping factor, r
m
the acceleration of the structure,
r
m
the velocity of the structure, f the surface pressure applied to
the structure, r
m
a variational displacement field, the strain
variation that is compatible with r
m
, S
fs
the structural wetted
surface, S
f
the surface which f acting on, and P the pressure acting
on the wetted surface which can be calculated through Eqs. (7b) or
(15),
P(t) = ((t) (t t))/t
2
z(t) +
1
2
|u(t)|
2
. (15)
As described above, the response of structure under bubble load
can be solved through combining Eqs. (14) and (15), and the
calculations canbe carriedout inthe way describedinthe step(f) in
last section, which indicates that the velocity on the wetted surface
coincides with that on the fluid boundary Eq. (16). Details can refer
to flow chart (3) in the Section 2.4.
/n = r
m
t
n. (16)
An interface routine alternately calls the BEM and FEM solvers.
The time steps of the two codes can be, and are usually, different.
Therefore, a leapfrog time-advancing scheme is selected in the
present work to cope with the different time steps of the fluid and
solid codes.
3. Results and discussions
Underwater explosion can be divided into two stages, namely
the stage of the shock wave and bubble pulsation. Generally the
load caused by the shock wave is very high but the corresponding
duration is very short (several milliseconds); while the load caused
by bubble is low but the duration is longer (several seconds) [10],
showed as Fig. 4. Although both the loads will inflict severe
damages on the adjacent structures, their damage mechanisms
are different: the shock wave will result in great damages on the
structure whose natural period is in the order of milliseconds, and
this kind of structures is generally local structures on the ship such
as hull plates andgrillages, therefore the shock wave usually causes
local damages on ships;
Fig. 4. Phenomenon of the underwater explosion: shock wave and a high-pressure
bubble appear after explosion; the shock wave generates high-pressure load but
sustains short while pulsating bubble induces low-pressure load but sustains long
and the pressure caused by bubble collapse is getting lower along with time.
Fig. 5. The damage of an underwater explosion bubble on a ship: (a) shock wave
moves at a very high speed and generates the first shock on the ship; (b) gas
bubble is expanding and the ship is raising; (c) bubble is collapsing with a jet being
developed and the ship is pulled downward; (d) jet penetrates the bubble and
impacts on the ship, breaking the ship off.
However, the bubble pulsation drives the fluid around moving
in a big area, which induces the after flow thus, and the collapse
of the bubble results in pulsating pressure. Both the after flow
and pulsating pressure will cause global damage on ships. If the
frequency of the bubble pulsation approaches to eigenfrequency
of the ship, it will cause whipping motion of the ship which
endangers the total longitudinal strength and even breaks off the
ship from its middle part. At the same time, the high-speed jet
formed in bubble collapse aggravates the damages on ships so
that the destroyed ship may submerge. The whole process of the
dynamic response of a surface ship under the effect of bubble
pulsation is showed as Fig. 5.
The underwater explosion shock wave and its damage on
structures have been the core of underwater explosion problems
studied before the middle 1990s. There have been many literatures
and numerical algorithms (DAA2, ALE and so on) to simulate the
explosion shock wave from non-contact explosion in far field
to contact explosion in near field. For example, the commercial
solvers like ABAQUS/UNDEX, LS-DYNA/USA can simulate the
damage of the underwater explosion shock wave on structures in
good precision. Therefore ABAQUS/UNDEX is adopted in present
paper to analyze the underwater explosion, and the shock wave
load refers to the model of Geers and Hunter [22]. The pressure of
164 A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171
the shock wave versus time is:
P(t) =
1
R

f
4
_
a
c
R
_
A

V(t) (17)
where R is the distance from arbitrary measuring point to the
charge center,
f
is the mass density of the fluid (i.e. water), a
c
is
the radius of the explosive charge, A is a constant concerning about
the material property of the charge (for TNT A = 0.18),

V(t) can
be obtained through Eq. (18):

V(t) =
4a
c

f
P
c
[0.8251 exp (1.338t/T
c
)
+0.1749 exp (0.1805t/T
c
)] (18)
where P
c
= K
_
m
1/3
c
/a
c
_
1+A
, T
c
= km
1/3
c
_
m
1/3
c
/a
c
_
B
, m
c
is the
mass of the explosive charge. K, k, and B are also constants for
charge material (for TNT K, k, and B are taken as 5.21 10
7
,
9.0 10
5
and 0.185 respectively).
Although current FEM solvers have good precision when
simulating the damage of the shock wave on underwater
structures, there are many limitations to simulate the interaction
between the underwater explosion bubble and structures, which is
because that the bubble pulsating period is usually very long and
the meshes of the bubble will distort during evolution, resulting
in huge calculation for explicit finite element scheme. Therefore
based on the potential flow theory, boundary element code is
exploited in this paper to study the interaction between the
underwater explosion bubble and structures.
3.1. Circular plate model
The interaction between a bubble and an elasticplastic circular
plate is studied in this section; the experimental data are from
Reference [9]. The experiment was carried out in a pool: an
explosive charge of 55 g exploded under the circular steel plate
of thickness 2 mm. The steel plate was fixed on the surface, one
side encountered the blasting load and the other side was exposed
in the air. The mass density of the plate is 7800 kg/m
3
; the yield
stress is 240 MPa; the shear modulus is 80.7 GPa while the Youngs
modulus is 210 GPa; and the Poissons ratio is 0.3. Assumed the
model was ideal elasticplastic and the dimensionless damping
ratio was taken as 0.05. The steel plate had been fixed at its
boundaries for all six degrees of freedom and it was thick enough
to allow relatively large deformation of the plate to occur. The
standoff distance had been chosen to be 1.2 m below the plate
(or 2.2 times of the maximum bubble radius). The distance
between the explosive charge and the floor of the pond was 3.5 m
(the charge depth was also 3.5 m). In the experiment, several
displacement sensors were used to record the deformation of the
steel plate. Since the standoff distance was far larger than the
bubble radius, the bubble behaved, to some extent, like a free-field
bubble. The numerical results are shown in the Fig. 6; the vertical
displacement of the plate at 89 ms is shown in Fig. 7.
Fig. 6(a) shows the initial states of the charge and plate; at
around 46 ms, the centre of the plate moves towards the bubble
because of the negative pressure, while the bubble remains roughly
spherical (Fig. 6(b); at the time of 89 ms, it can be seen that the
centre of plate moves away from the bubble resulting from the
collapse of the bubble, and the bubble evolves into a toroidal form
(Fig. 6(c)). The time history of the displacement of the plate center
is given in Fig. 8, both the experimental and numerical results are
shown.
As can be seen from Fig. 8, both the experiment and simulation
have a deformation of zero initially. After the explosion, the
displacement of the plate center approaches nearly 30 mm in
several milliseconds. Then, the plate center changes its direction
(a) t = 0 ms. (b) t = 46 ms. (c) t = 89 ms.
Fig. 6. The coupling effect between bubble and plate at t = 0, 46 and 89 ms; bubble
first expands and then collapses to minimum; a toroidal bubble forms finally (the
brown bar represents the vortex ring). The centre of plate moves up and down with
the expansion and collapse of the bubble.
Fig. 7. Deformation of a circular plate at t = 89 ms, the color contour represents
the magnitude of the vertical displacement. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 8. Comparison of the displacement of the plate center along z-axis: a positive
displacement means that the plate is moving away from the bubble.
and moves towards the bubble rapidly. Bubble pressure turns to
be lower than its ambient pressure, so a suction effect exerts
on the plate. The plate then stays a relatively long time, nearly
80 ms, when the plate is pulled below the water surface. When
finally the bubble collapses, the pressure caused by the ensuing jet
and the compressed gas inside the bubble eventually leads to the
permanent (and plastic) deformation of the plate. The deformation
tested in the experiment is a little larger than the numerical result;
the ultimate simulated plastic deformation of the plate is 37 mm
while the experimental result is 40 mm, with an error of about
8%. There are many reasons accounting for the difference such
as the potential flow assumption and also error of the numerical
simulation. To illustrate the power of bubble load further, the time
history of the pressure and stress of the heading-on element (this
element is the nearest to the charge center) on the plate are shown
in Fig. 9.
Fig. 9 shows that when the bubble collapses, the plate suffers
the maximum pressure load and its stress reaches its maximum
value correspondingly, which indicates that the pressure resulted
A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171 165
(a) Pressure. (b) Stress.
Fig. 9. Time history of the pressure and stress of the heading-on element (this element is the nearest one to the charge center) at the plate center.
Fig. 10. Comparison of the displacement of the bubble center: the positive
displacement means the bubble is rising.
Fig. 11. The sketch map of a cylinder model: blast direction is showed in arrow.
frombubble collapse and ensuing jet can cause severe damages on
the structure. During the whole dynamic process, the time history
of the displacement of the bubble center is shown in Fig. 10. The
bubble rises slowly in the expansion phase while it rises fast in
the collapse phase. The numerical results compare well with the
experimental results.
3.2. Cylinder model
The interaction between a bubble and an elasticplastic
cylinder is studied in this section; the experimental data are from
Reference [23]). The model is shown in Fig. 11. The outer diameter
of the cylinder is 1 m; the thickness of the shell is 5 mm and the
length is 1.8 m. The experiment was carried out in a pool, with an
explosive charge of 1 kg. The cylinder was made in plain steel with
the density 7800 kg/m
3
, yielding stress 230 MPa, Youngs modulus
210 GPa, and Poissons ratio 0.3. The elasticplastic model was
isotropic hardening and the non-dimensional damping ratio was
taken as 0.05.
3.2.1. Comparison of the simulated results and experimental data
The experimental case is simulated with the developed code
first. The center of the cylinder is located 5 m below the water
surface; the charge is parallel to the cylinder, with the blast
direction is shown in Fig. 11. The charge center is located 5 maway
fromthe cylinder in the horizontal direction, and the distance from
Fig. 12. Time history of the displacement of the heading-on node (this node
encounters the shock wave first): the positive displacement means the cylinder is
moving away from the bubble.
the charge center to the free surface is far larger thanthe maximum
radius of the bubble. Thereby the free surface effects on the freely
constrained cylinder can be ignored. However, the effects induced
by the bubble pulsation pressure, after flow and jet are taken into
account. Fig. 12 shows the time history of the displacement of the
heading-on node (this node encounters the shock wave first) on
the cylinder.
Fig. 12 shows that the cylinder moves against and towards the
bubble with the bubbles expansion and collapse under the effects
of pulsating pressure and after flow. The maximum displacements
of the experimental and numerical results are 13.51 mm and
12.3 mm respectively, with an error of about 10%. Generally,
the numerical simulation compares well to the experiment, and
the dynamic process of the interaction between the bubble and
cylinder is shown in Fig. 13.
Fig. 13(a) shows the states of the bubble and cylinder at the time
of 1 ms, when the pressure inside the bubble is very high and the
bubble expands rapidly, driving the ambient flow to move, which
forms the after flow and impels the cylinder to move. The motion
direction of the cylinder is shown in the figure. At around 112 ms,
the bubble reaches its maximum volume and the pressure inside
the bubble is lower than its ambient pressure, when the bubble
begins to collapse. The cylinders moving direction is shown in
Fig. 13(b). At the time of 218 ms (Fig. 13(c)), the bubble collapses
to its minimum volume and a toroidal bubble has been formed.
The brown bar in the figure represents the vortex ring. Then the
bubble begins to rebound and the motion direction of the cylinder
is shown in the figure.
3.2.2. The case with a free surface
Based on the model in last section, the case with a free surface
is studied by changing the boundary conditions. An explosive
package charge of 1 kg is located 1.5 m below the free surface and
the cylinder is placed on the right of the charge, 3 m away in the
same horizontal plane. The dynamic coupling process among the
bubble, free surface and cylinder is shown in Fig. 14.
In Fig. 14 the color contour represents the magnitude of the
velocity potential of the bubble, free surface and cylinder. Fig. 14(a)
166 A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171
(a) t = 1 ms. (b) t = 112 ms.
(c) t = 218 ms.
Fig. 13. The dynamic process of the interaction between bubble and cylinder at t = 1, 112 and 218 ms.
(a) t = 0 ms. (b) t = 18 ms.
(c) t = 96 ms. (d) t = 184 ms.
(e) t = 204 ms.
Fig. 14. The interaction process among bubble, free surface and cylinder at t = 0, 18, 96, 184 and 204 ms, the color contour represents the magnitude of the velocity
potential. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
shows the boundary states at the time of 0 ms, when the pressure
inside the bubble is very high and the bubble expands fast. At
around 18 ms (Fig. 14(b)), the bubble expands and drives the
ambient flow to move, forming the after flow and impelling the
cylinder to move towards left. Besides, the free surface is raised a
little as well. At the time of 96 ms (Fig. 14 (c)), the bubble reaches
its maximum volume and the pressure inside it is lower than its
ambient pressure. The free surface is raised further and the bubble
will begin to collapse. At around 184 ms (Fig. 14(d)), the bubble
begins to collapse and the free surface keeps on rising. The cylinder
moves towards the bubble overall. And finally, at the time of 204
ms, the bubble has collapsed and the jet penetrates the opposite
side of the bubble, so the bubble evolves into a toroidal shape
(Fig. 14(e)). The cylinder moves towards the bubble even more.
Fromthe process above it canbe seenthat the cylinder moves away
from and towards the bubble with the expansion and collapse of
the bubble respectively.
3.2.3. The case with bubble under cylinder
This section focuses on the case that the charge bursts right
under the cylinder. Similar to the cases above, a charge weighted
1 kg blasts 5 m below the water surface. The cylinder is freely
suspended above the charge with its center 3 m above the initial
bubble center. The mass density of the cylinder is 7800 kg/m
3
; the
yielding stress is 230 MPa; the Youngs modulus is 210 GPa and the
Poissons ratio is 0.3. Here the free surface effects arent taken into
account. The dynamic process of the bubble and cylinder is shown
in Fig. 15.
Fig. 15(a) shows the states at the time of 1 ms, when the
pressure inside the bubble is very high and the bubble expands
A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171 167
(a) t = 1 ms. (b) t = 112 ms. (c) t = 218 ms.
Fig. 15. The dynamic process of the interaction between bubble and cylinder at
t = 1112 and 218 ms.
Fig. 16. Time history of the vertical displacement of cylinder center: the positive
displacement means the cylinder is moving away from the bubble.
(a) t = 1 ms. (b) t = 112 ms. (c) t = 218 ms.
Fig. 17. The dynamic process of the interaction between bubble and rigid fixed
cylinder at t = 1112 and 218 ms.
rapidly, driving the ambient flow to move and forming an after
flow. The after flow impels the cylinder to move upwards fast.
At around 112 ms (Fig. 15(b)), the bubble reaches its maximum
volume and the pressure inside the bubble is lower than its
ambient pressure and then it begins to collapse. The cylinder
is raised greatly because of its inertia. At the time of 218 ms
(Fig. 15(c)), the bubble has collapsed and the jet is formed
while the bubble becomes toroidal, with the brown bar in the
figure representing the vortex ring. Then the bubble rebounds
and the cylinder is pulled back. The time history of the vertical
displacement of the cylinder center is shown in Fig. 16, fromwhich
it can be seen that the cylinder moves away from and towards
the bubble along with the expansion and collapse of the bubble.
This coincides with the conclusion which has been drawn in the
Reference [23].
Change the boundary conditions of the cylinder by fixing its
bottom rigidly, and then calculate the cases above. The dynamic
coupling process of the bubble and cylinder is shown in Fig. 17.
Fig. 17(a) shows the states of the bubble and cylinder at the
time of 1 ms, when the pressure inside the bubble is very high
and the bubble expands outwards quickly but the cylinder deforms
a little. At the time of 112 ms (Fig. 17(b)), the bubble reaches its
Fig. 18. Permanent deformation of the middle part of the cylinder: the
color contour represents the magnitude of the equivalent plastic strain. (For
interpretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)
maximumvolume. Aconcave appears at the bottomof the cylinder
since its edges are fixed rigidly. At around 218 ms (Fig. 17(c)), the
bubble collapses and develops a jet. The bubble becomes toroidal
(the brown bar in the figure still represents the vortex ring) and
then begins to rebound. The middle part of the cylinder deforms
greater and a clear concave is formed at last. The contour of its
plastic deformation is shown in Fig. 18.
The pressure caused by the ensuing jet acts upon the cylinder,
generating the permanent deformation (plastic deformation) on it.
The cylinder deforms badly, which indicates the great power of
the bubble load. The time history of the pressure and stress of the
heading-onelement (this element encounters the shock wave first)
on the cylinder is shown in Fig. 19.
Fig. 19 shows that when the bubble collapses, the pressure
reaches the maximum value and then the structure suffers the
maximum stress, which indicates that the pressure caused by
bubble collapse and jet can inflict severe damage upon structure.
3.3. The interaction between a bubble and a complex elasticplastic
structure (e.g. a surface ship)
The interaction between a bubble and a simple elasticplastic
structure is studied above. In this section the three-dimensional
code is converted to settle complex elasticplastic structure
problem in project application field. Finally, a full coupling
dynamic analysis code is exploited, which is fit for solving the
interaction between a three-dimensional bubble and a complex
elasticplastic structure.
3.3.1. Explosion in middle and far field
The length of the ship adopted in the simulation is L, the width
is B and the draft is T. The mass density of the ship is 7800 kg/m
3
,
the yielding stress is 350 MPa, the Youngs modulus is 210 GPa
and the Poissons ratio is 0.3. The ideal elasticplastic model is
isotropic hardening and the dimensionless damping factor is taken
as 0.05. The charge weights N kg and is located 0.24 L under the
midship section. The maximum diameter of the bubble generated
by the explosion is about 0.08 L. The dimensionless distance from
the bubble center to the free surface is greater than 3, so the free
surface effects arent taken into account. The interaction between
the underwater explosion bubble and surface ship is simulated by
the three-dimensional code in this paper and the dynamic process
is shown in Fig. 20.
The color contour represents the magnitude of the stress of
the surface ship. Fig. 20(a) shows the initial size and mesh of
the bubble and surface ship, when the bubble expands outwards
rapidly under the effect of the high-pressure gas inside. At around
0.049 s, the bubble moves and drives the ambient flow to move,
forming an after flow. The after flow generates a low-frequency
pushing force on the ship and raises it (Fig. 20(b)). At the time of
168 A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171
(a) Pressure. (b) Stress.
Fig. 19. Time history of the pressure and stress of the heading-on element (this element encounters the shock wave first) on the cylinder.
(a) t = 0. (b) t = 0.049.
(c) t = 0.48. (d) t = 0.84.
(e) t = 0.95. (f) t = 1.12.
Fig. 20. The interaction between underwater explosion bubble and ship at dimensionless t = 0, 0.049, 0.48, 0.84, 0.95 and 1.12; the color contour represents the magnitude
of the Mises stress. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
0.48 s, the bubble reaches its maximum volume, and the pressure
inside is less than the surrounding pressure, so the ship is pulled
downwards by the bubble. The bubble basically keeps spherical,
and then begins to collapse (Fig. 20(c)). As can be seen from
Fig. 20(d), at the time of 0.84 s, the bubble begins to collapse
and the ship moves downwards wholly. The bubble is no longer
A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171 169
Fig. 21. Whipping deflection of hull: the time changes from 0.1 s to 1.1 s.
Fig. 22. Distribution of the stress along the ship: the stress in middle ship is bigger
than that in the stem and stern.
Fig. 23. Time history of the displacement of the typical part of hull: a positive
displacement means that the ship is moving away from the bubble.
spherical. In Fig. 20(e) the bubble collapses and the jet penetrates
the bubble, impacting on the opposite side of the bubble. And then
the bubble becomes toroidal, continuing collapsing. The brown bar
in the figure represents the vortex ring. The ship is in hogging
condition because of the bubble collapse and jet impact. If the
midship structure isnt strong enough, the ship will probably be
broken off. Fig. 20(f) shows the states at the time of 1.12 s, when
the bubble rebounds and gets into the next period.
It can be clearly observed that the ship takes on its first-order
vertical mode and does whipping motion, because the pulsating
frequency of the bubble approaches the low-order eigenfrequency
of the ship. The movement of the ship is shown in Fig. 21 and the
stress distribution along its length direction is shown in Fig. 22.
As Fig. 22 shows, the stress in the middle part of the ship is much
greater than that in the stem and stern. The vertical displacement
of the heading-on node (this node encounters the shock wave first)
in the midship section is shown in Fig. 23, from which we can see
that the ship moves up and down with the expansion and collapse
of the bubble.
Along the length direction of the ship, three points are taken in
the middle sheer plate to check the pressure on the ships outside
plate. The three points labeled with P1, P2 and P3 are 0, 0.5 L
Fig. 24. Contrast of the pressure in different positions: pressure at P1 is the biggest
and pressures at P2 and P3 equals to each other.
Fig. 25. Time history of the stress of the typical parts(outside plate, transverse
bulkhead and deck) near the midship section.
and 0.5 L from the midship section respectively. Fig. 24 shows the
pressure on these points: the peak pressure of P1 is the biggest;
the pressures of P2 and P3 are equal, much lower than that of P1.
This also indicates that the pressure along the motion direction
of the bubble and attack direction of the jet is the highest. The
time history of the stress of the typical parts amidships is shown
in Fig. 25. The pressure reaches its maximum when the bubble
collapses. This once again shows that the pressure induced by the
bubble collapse can cause serious damage on the ship structures.
3.3.2. Explosion in near field
According to the studies, when the charge bursts under the
bottom of the bull, the generated bubble load can have a great
power. Moreover, the shorter the standoff distance is, the greater
the power is. So the near-field explosion (the charge bursts near
the structure) is studied in this section: The charge (TNT) weights
N kg and locates 0.04 L below the midship section. The dynamic
response of the surface ship is shown in Fig. 26.
Fig. 26 shows the whole dynamic response process of the ship.
As Fig. 26(a) shows, the shock wave load only causes local damage
on the ship. FromFig. 26(b)(e), it can be seen that the bubble load
causes global damage and leads to a large range of plastic strain on
the ship. Time history of the equivalent plastic strain of an element
selectedfromthe deck inmidshipsectionis showninFig. 27, which
indicates that the plastic strain of the midship section is mainly
caused by the bubble load (the time when the bubble collapses is
about 0.9 s).
In the end, it has to be noted that cavitation is a significant
phenomenon during an underwater explosion and also for the
surviving environment of surface ships [24,25]. Unfortunately, the
boundary-element method is poorly suited to the investigation
of cavitation. So cavitation isnt taken into account in this paper;
however, it should be paid much attention to in further study.
4. Conclusions
With the conjunction of BEMand FEM, the interaction between
the underwater explosion bubble and elasticplastic structure is
studied in this paper. Firstly, a numerical algorithm to calculate
170 A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171
(a) t = 0.12. (b) t = 0.75.
(c) t = 0.90. (d) t = 1.11.
(e) t = 1.26.
Fig. 26. The dynamic response of ship to near-field explosion at dimensionless t = 0.12, 0.75, 0.90, 1.11 and 1.26; the color contour represents the magnitude of the
equivalent plastic strain. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 27. Time history of the equivalent plastic strain of the deck element in midship
section.
the coupling dynamics of a bubble, an elasticplastic structure and
a free surface is showed. Then, a complete 3D code based on the
algorithm is exploited, whose computational error is within 10%
compared with the experimental results. Meanwhile, the code is
used in the project application field, simulating the interaction
between a surface warship and an underwater explosion bubble.
Founded on the works above, some conclusions are got as follows:
(1) From the time history of the pressure and stress on the
structure, it can be seen that both of them reach their
maximumvalues when the bubble collapses, which shows that
the pressure caused by the bubble collapse and ensuing jet can
cause severe damages on the ship.
(2) Because the bubble pulsating frequency approaches to ships
low-order eigenfrequency, the low-order vertical mode of the
warship is aroused and the ship presents whipping motion.
(3) In an underwater explosion, the surface ships and underwater
structures (e.g. cylinder model in this paper) all move up and
down with the expansion and collapse of the bubble.
(4) The explosion bubble load has a great power when the charge
bursts under the bottom of the hull. Besides, the shorter the
standoff distance is, the greater the power is. Generally, the
shock wave only cause local damage on the ship, while the
bubble load will inflict total damage on it, endangering its total
longitudinal strength and even breaking it off.
Acknowledgments
Thanks to Prof. K.S. Yeo and Prof. B.C. Khoo at National
University of Singapore and Prof. E.Klaseboer at Institute of High
Performance Computing, who supported and assisted in many
valuable ways during the course of this research. This study
is supported by a Grant (50779007) from the National Science
Foundationof China and also a Grant (50809018) fromthe National
Science Foundation for Young Scientists of China.
References
[1] Naude CF, Ellis AT. On the mechanism of cavitations damage by non-
hemispherical cavities in contact with a solid boundary. ASME Journal of Basic
Engineering 1961;83(2):64856.
A.M. Zhang et al. / Applied Ocean Research 30 (2008) 159171 171
[2] Benjamin TB, Ellis AT. Cavitation: The collapse of cavitation bubbles and
the pressures thereby produced against solid boundaries. Philosophical
Transactions of the Royal Society of London 1966;260(1):22140.
[3] Plesset MS, Chapman RB. Collapse of an initially spherical vapor cavity in the
neighborhood of a solid boundary. Journal of Fluid Mechanics 1971;47(3):
28390.
[4] Blake JR, Taib BB, Doherty G. Transient cavities near boundaries part I, rigid
boundary. Journal of Fluid Mechanics 1986;17(3):47997.
[5] Chan PC, Kan KK, Stuhmiller JMA. Computational study of bubble-structure
interaction. Journal of Fluids Engineering 2000;122(1):78390.
[6] Zhang YL, Yeo KS, Khoo BC, Wang C. 3Djet impact and toroidal bubbles. Journal
of Computational Physics 2001;16(6):33660.
[7] Zhang AM, Yao XL, Yu XB. The dynamics of three-dimension underwater
explosion bubble. Journal of Sound and Vibration 2008;311(4):1196212.
[8] Kalumuck KM, Chahine GL, Duraiswami R. Bubble dynamics fluidstructure
interactionsimulationby coupling fluidBEMandstructural FEMcodes. Journal
of Fluids and Structures 1995;9(4):86183.
[9] Klaseboer E, Hung KC, Wang CW, Khoo BC, et al. Experimental and numerical
investigation of the dynamics of an underwater explosion bubble near a
resilient/rigid structure. Journal of Fluid Mechanics 2005;53(7):387413.
[10] Cole RH. Underwater explosion. Princeton (NJ, USA): Princeton University
Press; 1948.
[11] Wang QX, Yeo KS, Khoo BC, Lam KY. Strong interaction between a buoyancy
bubble and a free surface. Theoretical and Computational Fluid Dynamics
1996;8(1):7388.
[12] Wang QX, Yeo KS, Khoo BC, LamKY. Nonlinear interaction between gas bubble
and free surface. Computers & Fluids 1996;25(7):60728.
[13] Wang C, Khoo BC. An indirect boundary element method for three-
dimensional explosion bubbles. Journal of Computational Physics 2004;19(4):
45180.
[14] Rayleigh JW. On the pressure developed in a liquid during the collapse of a
spherical cavity. Philos Magazine 1917;34(3):948.
[15] Lundgren TS, Mansour NN. Vortex ring bubbles. Journal of Fluid Mechanics
1991;22(4):17796.
[16] Best JP. The formulation of toroidal bubbles upon collapse of transient cavities.
Journal of Fluid Mechanics 1993;251:79107.
[17] Zhang S, Duncan JH, Chahine GL. The final stage of the collapse of a cavitation
bubble near a rigid wall. Journal of Fluid Mechanics 1993;25(7):14781.
[18] Zhang S, Duncan JH. On the nonspherical collapse and rebound of a cavitation
bubble. Physics of Fluids 1994;6(7):235262.
[19] Klaseboer E, Khoo BC, Hung KC. Dynamics of an oscillating bubble near a
floating structure. Journal of Fluids and Structures 2005;10(2):110.
[20] Wang C, Khoo BC, Yeo KS. Elastic mesh technique for 3D BIM simulation with
an application to underwater explosion bubbles. Computers & Fluids 2003;
32(9):1195212.
[21] Belytschko T, Mullen R. Two-dimensional fluidstructure impact computation
with regulation. Computer Methods on Applied Mechanics and Engineering
1981;27(1):13954.
[22] Geers TL, Hunter KS. An integrated wave-effects model for an underwater
explosion bubble. Journal of the Acoustical Society of America 2002;111(4):
1584601.
[23] Li GH, Li YJ, Zhang XC, Yang YC, et al. Verification of the relationship between
the bubble motion and the warship equipment vibration by test. Journal of
Ship Mechanics 2005;9(1):98105.
[24] Sprague MA, Geers TL. A spectral/finite-element analysis of a frigate-like
structure subjected to an underwater explosion. Computer Methods in
Applied Mechanics and Engineering 2006;195:214967.
[25] Sprague MA, Geers TL. A spectral-element method for modeling cavitation
in transient fluidstructure interaction. International Journal for Numerical
Methods in Engineering 2004;60(15):246799.

You might also like