You are on page 1of 13

Journal of Coastal Research 00 0 000–000 Coconut Creek, Florida Month 0000

TECHNICAL COMMUNICATIONS
www:cerf -jcr:org

Dynamics of an Underwater Explosion Bubble near a Rigid Wall: Effect of


Slenderness Ratio, Installation, and Distance Parameter
Zhifan Zhang, Longkan Wang, Xiongliang Yao, and Jicai Lang*
College of Shipbuilding Engineering
Harbin Engineering University
Harbin, Heilongjiang, China

ABSTRACT
Zhang, Z.; Wang, L.; Yao, X., and Lang, J., 0000. Dynamics of an underwater explosion bubble near a rigid wall: Effect of
slenderness ratio, installation, and distance parameter. Journal of Coastal Research, 00(0), 000–000. Coconut Creek
(Florida), ISSN 0749-0208.

Structural damage of maritime construction (e.g., dams, warships, etc.) has received considerable international attention
in recent years because of underwater explosions from accidental events and terrorist bombing attacks. Therefore,
research studies on underwater explosion load characteristics will have a great influence on the future of coastal and
maritime engineering. Here, level set–direct ghost fluid–Runge Kutta discontinuous Galerkin method and boundary
element method are combined to establish a model of underwater explosion near a rigid wall. First, the hybrid algorithm
is used to simulate the process of underwater explosion in free field; the results agree well with experimental data,
proving the effectiveness of the algorithm. Second, the process of underwater explosion near a rigid wall is simulated by
the presented method. Finally, effects of parameters – slenderness ratio, installation (horizontal and vertical), and
distance from the center of the explosive to the rigid wall – are investigated on shock waves and bubble dynamics. It is
found that during the detonation process, the ellipsoidal bubble gradually turns into a spherical one. The radial pressure
peak value is higher than the axial one. During the collapse phase, the slenderness ratio and the installation have little
effect on bubble shape, maximum radius, and pulsation cycle. During the bubble-jet process, a high-speed jet penetrates
the bubble toward the rigid wall and generates a high-pressure region on the bubble wall. For the charge placed
vertically, the jet velocity rises while the jet width decreases as the slenderness ratio increases; results for horizontal
cases are opposite. The jet velocity in the vertical case is lower than that in the horizontal case; on the contrary, the jet
width is larger. With the increase of the distance parameter, the pressure on the upper surface of the bubble and the jet
velocity are higher, whereas the jet width is smaller.

ADDITIONAL INDEX WORDS: Maritime engineering, column charge, bubble jet, load characteristic.

INTRODUCTION fore, a numerical method (Kim and Gitterman, 2013; Klaseboer


Structures such as warships and high dams, etc. may be et al., 2005; Ming et al., 2016; Zhang, Li, and Cui, 2015; Zhang
severely damaged by underwater explosions because of the and Liu, 2015) is usually combined with experiments to study
effects of the strong discontinuous shock wave, the high- an underwater explosion bubble. The boundary element
pressure bubble impulse, and the high-speed jet during the method (BEM) was used to simulate the growth and collapse
bubble collapse phase (Cole, 1948; Wang and Zhang, 2014; of transient vapour cavities near a rigid boundary and a free
Zhang et al., 2014). Research on these loads has a significant surface by Blake, Taib, and Doherty (1986, 1987), the results of
influence on protective design and antiknock performance; which were compared with the experimental results to verify
they have attracted wide attention and were studied by various the validity of the numerical method. Results using a coupled
researchers using experiments, theoretical analysis, and BEM–finite element method method and from experiments
numerical modelling. Although experimental research (Hung were compared to analyze the dynamics of an underwater
and Hwangfu, 2010; Klaseboer et al., 2005; Rajendran and explosion bubble near a submerged structure (Klaseboer et al.,
Narasimhan, 2001; Zhang et al., 2015) is the most effective and 2005). An indirect BEM (IBEM) simulation of three-dimen-
direct method to investigate the near-field underwater explo- sional explosion bubbles near a rigid wall was presented by
sion, it has disadvantages of less safety, bad operability, Wang and Khoo (2004) and the results agreed well with the
unrepeatability, and limited information acquisition. There- experimental ones. Bubble dynamics in different boundary
conditions, such as free surface, rigid boundary, etc., were
DOI: 10.2112/JCOASTRES-D-16-00094.1 received 20 May 2016; discussed by means of BEM and were studied experimentally
accepted in revision 30 June 2016; corrected proofs received
with a high-speed camera by Zhang et al. (Zhang et al., 2015;
30 September 2016; published pre-print online 03 January 2016.
*Corresponding author: heulangjicai@126.com Zhang, Li, and Cui, 2015; Zhang and Liu, 2015). The BEM was
Ó
Coastal Education and Research Foundation, Inc. 2016 used by Wang et al. (1996a,b) to discuss the influences of a free
0 Zhang et al.

surface and the buoyance parameter on the jet characteristics. METHODS


In all these numerical research studies mentioned above, the A combined LS-DGF-RKDG-BEM algorithm is used to build
fluid was assumed to be inviscid, irrotational, and incompress- the LS-DGF-RKDG explosion model and bubble dynamics
ible. However, the compressibility of the fluid has a significant model on the basis of the wave motion equation. LS-modified
influence on bubble motion when the Mach number at the top of GF-DG method is used to simulate the process of the column
bubble jet reaches 0.3 or even higher during the collapse phase charge detonation and bubble-jet formation, whereas the BEM
(Wang and Blake, 2010). Therefore, to improve the accuracy of is applied to simulate the process of bubble expansion and
numerical calculation, differential equations for the motion of a contraction.
spherical bubble in a compressible liquid were obtained by
Prosperetti and Lezzi (1986) on the basis of a simplified LS-DGF-RKDG Method in Compressible Fluid
singular-perturbation method. An axisymmetric bubble model During the jet phase of an underwater explosion bubble,
was built by Wang and Blake (2010) to simulate the motion of because of a large Mach number in flow field (caused by the
nonlinear and nonspherical bubbles in a compressible liquid by high velocity on the bubble surface) and the pressure wave, the
using the boundary integral method. With the idea from the fluid compressibility should be considered. The basic equations
doubly asymptotic approximation method (Geers, 1971, 1978; for a two-dimensional axisymmetric fluid field (Fedkiw et al.,
Geers and Felippa, 1983), Zhang, Wang, and Wu (2013) and Ye, 1999; Park, 2008) can be expressed as
Sun, and Pang (2015) adopted the boundary integral equation
for compressible fluids to get the motion law of the bubble in a Ut þ Ñ  FðUÞ ¼ SðUÞ; ð1Þ
compressible fluid. where
Column charge is commonly used in underwater missiles 8 9 2 3 8 9
>
> q > qu qv > qu >
(e.g., torpedo). However, in most cases, spherical bubble models < > = 6 qu2 þ p 7
>
< 2 >
=
qu 6 qvu 7 1 qu
were developed to study the underwater explosion bubble, and U¼ , FðUÞ¼4 2 5, SðUÞ ¼  x
> >
> qv > quv qv þ p >
> quv >
>
the initial velocity of the bubble significantly affecting bubble : ; : ;
qe uðqe þ pÞ vðqe þ pÞ uðqe þ pÞ
dynamics is ignored (Zhang et al., 2013). As a result, it is
necessary to take nonspherical detonation and initial velocity and where t, q and e denote time, density, and energy; x and y
into consideration. In this paper, a hybrid algorithm of level are the spatial dimensions; u and v are velocity components in
set–direct ghost fluid–Runge Kutta discontinuous Galerkin the x and y directions, respectively; and p is the pressure. For
method (LS-DGF-RKDG) (Cockburn and Shu, 1998; Park, detonation products, the pressure can be obtained by the
2008; Qiu, Liu, and Khoo, 2007, 2008; Wang et al., 2015; Wang, Jones–Wilkins–Lee equation of state (Fedkiw et al., 1999;
Zhang, and Wang, 2016) and BEM is used to simulate the Park, 2008); for the gas inside the bubble, pressure can be
entire process of underwater explosion from detonation to expressed as the function of the bubble volume on the basis of
bubble jet. First of all, a LS-DGF-RKDG model is developed to adiabatic assumption (Fedkiw et al., 1999; Park, 2008; Qiu,
analyze the process from column charge initiation to ellipsoi- Liu, and Khoo, 2007); as for the water, pressure can be gained
dal/spherical bubble formation. This is due to the fact that LS- by the Tait state equation (Fedkiw et al., 1999; Park, 2008; Qiu,
DGF-RKDG not only has an advantage to capture the shock Liu and Khoo, 2007).
wave generated by bubble motion, but also eliminates the The fluid motion control equations can be solved by RKDG
nonphysical oscillation efficiently while guaranteeing the method and the moving interfaces of multimaterial flow can be
accuracy. Then, considering that the BEM has advantages of easily captured by the LS method (Osher and Fedkiw, 2003).
interface capturing, higher computational efficiency, and The DGF method (Park, 2008) is used to combine the above two
accuracy, it is adopted to simulate the process of bubble methods. The detailed description of the coupling of these
pulsing. During this process, the previous results calculated methods can be seen in Park (2008) and Cockburn and Shu
with LS-DGF-RKDG are used as initial conditions for the BEM (1998). Taking the two-dimensional Euler equations as an
calculation. Finally, because some special numerical tech-
example to explain the spatially discrete RKDG method,. the
niques, such as vortex ring/sheet model, regulating grids, or
computational domain of fluid is denoted below. First, multiply
boundary nodes, etc. should be imported into BEM during the
both sides of Equation (1) with the test function /(x, y). Then,
jet phase, a combined method of RKDG and LS method, which
integrate over a cell Xj. Finally, integrate by parts and the
has an advantage of interface treatment, is used in the
semidiscrete equation can be obtained:
simulation of the bubble jet, taking the previous results
calculated with BEM as the initial condition. Consequently,
Z XZ
the hybrid algorithm, which fully utilizes the advantages of d
both the LS-DGF-RKDG method and BEM, is used in this U/ðx; yÞdX þ FðUÞ  n/ðx; yÞdC
dt e]Xj
paper. The detonation LS-DGF-RKDG model and the bubble Xj e
Z
dynamics model based on wave equation are developed to  FðUÞÑ/ðx; yÞdX
investigate the load characteristics of a column charge Xj
underwater explosion near a rigid wall in compressible fluid. ¼0 ð2Þ
The influences of different parameters, namely slenderness
ratio, installation, as well as distance from the center of the
explosive to the rigid wall, etc., on shock wave and bubble where n is the unit normal of the cell boundary pointing
dynamics are also discussed and analyzed. outward.

Journal of Coastal Research, Vol. 00, No. 0, 0000


Underwater Explosion Bubble near a Rigid Wall 0

Line integral and surface integral are obtained using BEM in Compressible Fluid
Gaussian quadrature formula (Cockburn and Shu, 1998) as On the basis of the compressible fluid mechanics theory, the
Z boundary integral equation in compressible fluid is applied in
X
L
FðUÞ  n/ðx; yÞdC ’ xl FðUðxel ; yel ; tÞÞ  n/ðxel ; yel Þjej ð3Þ this numerical model. During the bubble pulsing phase, the
e l¼1 state equation of the fluid around the bubble can be expressed as
Z X
M dP
c2 ¼ ð10Þ
FðUÞ  Ñ/ðx; yÞdX ’ xj FðUðxXj ; yXj ; tÞÞ  Ñ/ðxXj ; yXj ÞjXj dq
j¼1
Xj
where sound speed c is set as 1500 m/s for the small-amplitude
ð4Þ perturbation. Assuming that the fluid field is irrotational, the
Owing to the discontinuity of the flux at the boundary of the relation between the velocity u and the velocity potential / can
cell, the flux F(U[xel, yel, t])n, the approximate solution U, and / be written as u ¼ Ñ/. Thus the continuity equation can be given
can be replaced by the numerical flux F(Û[xl, yl, t])n, the by
approximate solution Û, and /̂, respectively. According to these 1 Dq
approximations, substituting Equation (3) and Equation (4) Ñ2 / ¼  ð11Þ
q Dt
into Equation (2), we can get
Z The relation between the velocity potential / and the
d XX
L
pressure p can be expressed as
Û /̂ðx; yÞdX þ xl FðÛðxel ; yel ; tÞÞ  n/ðxel ; yel Þjej
dt e]Xj l¼1
Xj ]/
pþq ¼0 ð12Þ
X
M ]t
 xj FðÛðxXj ; yXj ; tÞÞ  Ñ/̂ðxXj ; yXj ÞjXj ¼ 0 ð5Þ
j¼1
Substituting Equation (10) and Equation (12) into Equation
(11) and omitting the second-order term, the linear wave
where the numerical flux F(Û[xl, yl, t])n can be solved by the equation is obtained:
Lax–Friedrichs flux, written as
1
1 Ñ2 / ¼ / ð13Þ
 þ
FðÛðxel ; yel ; tÞÞ  n ¼ ðFðÛ ðxel ; yel ; tÞÞ þ FðÛ ðxel ; yel ; tÞÞÞ  n c2 tt
2
On the basis of Green’s second identity and Equation (13), the
a þ 
 ðÛ ðxel ; yel ; tÞ  Û ðxel ; yel ; tÞÞ Kirchhoff retarded potential equation (Zhang, Wang, and Wu,
2
2013) is given by
ð6Þ Z Z
1 ]/
b/ðrp ; tÞ ¼  ðrq ; t  rpq =cÞds
where a is the maximum eigenvalue of the Jacobian matrix rpq ]nq
S
]F(U)/]U. Z Z
Inside the quadrilateral element (xi½, xiþ½)3(yj½, yjþ½), the rpq  nq
 /ðrq ; t  rpq =cÞds
approximate solution Û(x, y, t) can be given by r3pq
S
Z Z 2
Ûðx; y; tÞ ¼ ŪðtÞ þ Ux ðtÞni ðxÞ þ Uy ðtÞgj ðyÞ þ Uxy ðtÞni ðxÞgj ðyÞ rpq  nq ]/ðrq ; t  rpq =cÞ
0 1 0 1  ds ð14Þ
cr3pq ]t
1 1
þUxx ðtÞ@ni ðxÞ  A þ Uyy ðtÞ@gj ðyÞ  A
2 2 S
3 3
where b is the solid angle, rpq is the distance between the field
ð7Þ point p and source point q, nq is the unit outward normal vector
xxi yy at the source point, and t – rpq/c is the source time.
where ni ðxÞ ¼ Dxi =2
, gj ðyÞ ¼ Dyj =2j , Dxi ¼ xiþ½  xi½, Dyj ¼ yjþ½ 
Then, the local approximation and the global approximation
yj½.
of Equation (14) can be written as (Zhang, Wang, and Wu,
Substituting Equation (6) and Equation (7) into Equation (5),
2013)
the semidiscrete formulation can be expressed as

Ut ¼ RðUÞ ð8Þ /̇ðtÞ þ ck/ðtÞ ¼ cu̇ n ðtÞ ð15Þ

where R(U) is the discrete operator of spatial derivatives. The Z Z


1  
discrete difference scheme can be gained using a third-order b/ðrp ; tÞ ¼  un;t ðrq ; tÞ  sun;tt ðrq ; tÞ dS
rpq
total variation diminishing Runge–Kutta time discretization Z ZS
method, which is written as rpq  nq
 /ðrq ; tÞdS ð16Þ
r3pq
ð1Þ ðnÞ ðnÞ S
U ¼ U þ DtRðU Þ
3 1 according to the Laplace conversion, using the perturbation
U ð2Þ
¼ U ðnÞ þ ðU ð1Þ þ DtRðU ð1Þ ÞÞ
4 4 analysis to match globally and locally and keeping only the
ð9Þ
1 2 first- and second-order terms. The relation between potential
U nþ1 ¼ U ðnÞ þ ðU ð2Þ þ DtRðU ð2Þ ÞÞ
3 3 and normal velocity can be given by

Journal of Coastal Research, Vol. 00, No. 0, 0000


0 Zhang et al.

/̈ðr; tÞ þcðK þ kÞ/̇ðr; tÞ þ KB1 Cc2 /ðr; tÞ of the bubble surface, i.e. Equation (19), is adopted to calculate
¼ c ü n ðr; tÞ þ cK u̇ n ðr; tÞ ð17Þ the bubble position. Second, the normal and tangential velocity
components can be obtained from the boundary integration
The kinematic boundary condition and the dynamic bound- Equation (17) and the differential processing of the velocity
ary condition of the bubble surface can be expressed as (Wang, potential, respectively. Third, the pressure inside the bubble is
2013) updated by using the state equation of ideal gas.

d/ u2 P‘ P (2) Information-delivery from BEM to LS-DGF-RKDG


¼ þ   gz ð18Þ
dt 2 q q method

The results from the calculation using BEM before the


dr
¼ Ñ/ðx; y; zÞ ð19Þ process of bubble jet are taken as the initial condition for the
dt calculation using the LS-DGF-RKDG method, such as the
The pressure P at any point in the flow domain is obtained initial velocity vw and the initial pressure pw of the water. The
through the Bernoulli equation, in which the velocity Ñ/ and initial density qw of water can be obtained by the state equation
]//]t are evaluated using the IBEM (Wang and Khoo, 2004) of water, given as
and the auxiliary function method (Wu and Hu, 2004),
respectively. qw ¼ qr ðpw =B0 þ 1Þ1=N ð20Þ
3
A Hybrid Algorithm Using LS-DGF-RKDG Method and where qr is the reference density, which is set as 1000 kg/m ; pw
BEM denotes the initial pressure of water; B0 is used to limit the
The fluid-governing equations for RKDG method are the maximum change of density; the initial maximum pressure is set
Euler equations, which can be used to simulate the flow field of as 2 3 107 Pa; and N is the density ratio index, which is set as 7.
the large-amplitude perturbation. The wave equation obtained According to the relation between the pressure inside the
by simplifying several equations, namely continuity equation, bubble p and the bubble volume V, the initial pressure Pa of the
momentum equation, mass conservation equation, as well as gas can be obtained
state equation, is used as the fluid-governing equation for
V0 c
BEM. It is only applicable to the case of a small-amplitude Pa ¼ Pc þ P0 ð Þ ð21Þ
Va
perturbation. The mean Mach number on the bubble surface is
about 0.1 (Wang, 2013; Wang and Blake, 2010) during the where Pc, P0, and V0 denote the pressure of saturated
bubble pulsing and collapse phases, which means that in this condensable vapor, the initial pressure, and the density of the
case the fluids are weakly compressible, i.e. a small-amplitude bubble, respectively; c is the specific heat capacity, which is set
perturbation. as 1.25 for the bubble generated by trinitrotoluene (TNT)
As a result, the hybrid algorithm based on LS-DGF-RKDG explosion.
and BEM can be applied to simulate the whole process of The initial density qa of the gas is obtained by the equation qa
underwater explosion, including the detonation, bubble puls- ¼ W/V and the initial velocity vector va is obtained by using the
ing, and jet development. For the LS-DGF-RKDG method, velocity vector linear interpolation within the fluid field around
parameters such as node coordinate, velocity, density, pres- the bubble.
sure, and energy, etc. are included in the calculation. As for
BEM, parameters including node coordinates and velocity
Parameter Definition
Some parameters need to be defined and are listed as follows:
vectors at the boundaries, velocity and pressure of the fluid
(1) The slenderness ratio k is defined as the division between
around, etc. can be obtained from the calculation using LS-
the diameter of the cross-section R and the height of the
DGF-RKDG, but no parameters inside the bubble can be
cylindrical charge l, i.e. k ¼ R/l. (2) The distance parameter ch is
obtained. Considering the advantages and drawbacks of both of
defined as ch ¼ d/Rm, where d is the distance between the center
the methods, a hybrid algorithm that can be divided into two
of the initial bubble and the rigid wall and Rm is the maximum
steps is proposed here.
radius during the expansion phase. (3) As for ‘‘jet width,’’ the
(1) Information delivery from LS-DGF-RKDG method to maximum height of the involution part of the bubble surface is
BEM ‘‘L,’’ at the moment just before jet penetration. The diameter D
of the inner ring at the position of ‘‘L/2’’ on the bubble surface is
Parameters of the air–water interface obtained from the LS- defined as jet width Djet in this paper.
DGF-RKDG method can be used as the initial condition for the
BEM calculation. For BEM, the initial node coordinates and RESULTS
velocity vectors at the boundaries are obtained from that at the
In this part, results for bubble dynamics of a column charge
air–water interface using the LS-DGF-RKDG method; the
subjected to underwater explosion and effects of different
initial pressure of the bubble is set as the average pressure of
parameters – slenderness ratio, installation, and distance
gas using LS-DGF-RKDG.
parameter – on load characteristics are included.
Subsequently, on the basis of the initial conditions above and
the BEM bubble model presented above, three parameters, Numerical Verification
namely, bubble position, velocity vectors, and pressure inside, The numerical results (e.g., bubble position, size, and shape,
can be further updated. First, the dynamic boundary condition etc.) agree well with the experimental results, which conform to

Journal of Coastal Research, Vol. 00, No. 0, 0000


Underwater Explosion Bubble near a Rigid Wall 0

the basic laws of bubble movement throughout the whole When kis fixed, compared with vertically installed charge, the
process of underwater explosion; during the charge detonating horizontally installed one has smaller jet velocity, wider jet
stage, the pressure attenuation tendency of numerical calcu- width, and smaller distance from lower surface of the bubble to
lation agrees well with that of the experiment; in the pulsing the rigid wall.
and jetting phase, numerical results (e.g., displacements of
upper surface, lower surface, and center position) are consis-
Influence of Distance arameter on Bubble Dynamic
For all the cases of ch ranging from 1.0 to 1.6, during the
tent with the experimental results. All of the above results
process of bubble jet, with the increase of the distance
sufficiently prove the effectiveness and accuracy of the hybrid
parameter ch, the fluid pressure acted on the bubble surface
algorithm to solve the problem of column charge underwater
and the jet velocity becomes larger, but the cycle of the bubble
explosion.
and the displacement of the bubble center become smaller.
Underwater Explosion Simulation of Column Charge
near a Rigid Wall DISCUSSION
Because the downward Bjerknes force, which makes the Load characteristics associated with underwater explosions
bubble attracted by the rigid boundary, is larger than the have gradually been moved to the forefront of attention in
upward buoyancy force, the bubble presents nonspherical coastal and maritime engineering. Parameters such as slen-
collapse at the time t ’ 59.1 ms. At the same time, a jet with derness ratio, installation, and distance from the center of
high speed is formed toward the rigid wall. Also, a high- explosive to the rigid wall have significant effects on it.
pressure region is induced in the jet area of the bubble, and the Therefore, these effects of parameters on loading characteris-
pressure peak value is about 40 MPa. The jet has an impact tics are discussed in detail in this section. The slenderness ratio
directly on the wall and a high-pressure region is developed ranges from 1 to 3 for a typical torpedo and there is a larger
between the wall and the bubble during the later stage of the range for mines. Thus, the studied cases are focused on the
collapse. The bubble continues to shrink and radiate a pressure middle and small slenderness ratio charge explosion in this
wave outward. Finally, the bubble expands again because the paper.
pressure inside the bubble is higher than that in the fluid field. To validate the presented algorithm, a model of a column
In addition, the pressure on the wall drops because of the charge underwater explosion near a rigid wall is developed and
attenuation of the jet. numerical results are compared with experimental data.
Convergence tests of the LS-DGF-RKDG and BEM methods
Influence of Slenderness Ratio on Load Characteristics
are taken as well. After that, influences of different parameters
Slenderness ratio has great influence on the initial shape and
are studied with 9.5 g of TNT charge and at a depth of 1 m in the
velocity of the bubble; with the increase of the slenderness ratio
water. Cases are listed as follows: for the slenderness ratio, 1 
k, the shape of the bubble changes from cylindrical, which is
k  6; for the installation, the charge is installed horizontally
identical to the column, to ellipsoidal, and finally becomes
and vertically respectively; for the distance parameter, 1.0  ch
spherical. Generally, the radial pressure peak value of the
 1.6.
shock wave is larger than the axial pressure peak value. The
larger k is, the bigger the radial pressure is and the smaller the Numerical Verification
axial pressure is. To verify the effectiveness of LS-DGF-RKDG-BEM in the
simulation of column charge underwater explosion, the
Influence of Installation (Horizontal and Vertical) on
numerical results are compared with the experimental results.
Bubble Dynamic
The numerical model is developed according to the experimen-
The charge position has an effect on the initial shape. The jet
tal model, in which a 9.5-g column of TNT charge is located at a
width of the horizontal case is larger than that of the vertical
water (80 cm 3 120 cm) depth of 1 m. In the LS-DGF-RKDG
case when k¼ 3.35 and 5.62.
method, there are 345,600 grid cells and the time-step size is
For a vertically installed column charge, the slenderness CFL 3 Dx
defined as Dt¼ maxðuÞþc , where CFL is set as 0.2; Dx, u, and c are
ratio k only has a little influence on the pulsation cycle, the
the grid size, velocity, and sound speed, respectively. In the
maximum radius, and the vertical displacement of the bubble
BEM, the bubble surface is discretized into 64 nodes and the
center. However, with the increase of k, the jet velocity becomes D/
time-step size is chosen as Dt¼  V c , where D/
bigger while the jet width becomes smaller. When the jet  2 2
max 1þ0:5jD/j v ze 0 
V
penetrates the bubble surface toward the rigid wall, a high- is chosen as 0.01; v and e are buoyancy and strength
pressure area is induced on the wall. Subsequently, the inside parameters; c is the specific heat capacity and set as 1.5 in
pressure of the bubble decreases after the jet, which leads to the this paper.
re-expansion of the bubble. Comparison between numerical and experimental results of
The installation of the column charge has nearly no influence the bubble motion is shown in Figure 1. It can be seen that the
on the pulsation cycle and the maximum radius of the bubble by shape and the first cycle of the bubble in the numerical
comparing these two cases, i.e. vertically installed case and simulation are all in good agreement with the experimental
horizontally installed case. For horizontally installed charge, results. At t ’ 0.294 ms, the initial bubble keeps the cylindrical
with the increase of the slenderness ratio k, the jet velocity feature of the column charge and series of shock waves with
decreases and the jet width increases. In contrast, for the exceedingly high pressure are formed in the early expansion
vertically installed case, with the increase of the slenderness phase. The shape of the bubble develops from an ellipsoid into a
ratio k, the jet velocity increases and the jet width decreases. sphere at t ’ 5.56 ms. After the radius of the bubble reaches the

Journal of Coastal Research, Vol. 00, No. 0, 0000


0 Zhang et al.

Figure 1. Comparison between numerical and experimental results of bubble shapes; maximum radius Rm ’ 0.327 m. Dimensional times in the experiment: (a) t
’ 0.446, 5.58, and 28.6 ms, (c) t ’ 53.1, 54.0, and 54.7 ms. Dimensional times in the simulation: (b) t ’ 0.294, 5.56, and 28.6 ms, (d) t ’ 53.2, 54.1, and 54.9 ms.
(Color for this figure is available in the online version of this paper.)

maximum value at t ’ 28.6 ms, the bubble converts into the result the lower surface of bubble will collapse at t ’ 53.2 ms.
contraction state because of the lower pressure inside the As the collapse continues, the bubble volume reaches the
bubble than that of the fluid around. The shape of a bubble minimum size and the bubble is in calyptriform shape at t ’
cannot remain spherical under the effect of gravity and as a 54.1 ms. Subsequently, the pressure inside the bubble drops

Journal of Coastal Research, Vol. 00, No. 0, 0000


Underwater Explosion Bubble near a Rigid Wall 0

Figure 4. Pressure peak values vs. grid cells, presenting a good convergence
of the LS-DGF-RKDG method.

Figure 2. Time history curves of shock wave pressure generated by


experimental and LS-DGF-RKDG methods at the test point, showing good center position between results from calculation and experi-
agreement of numerical and experimental results for pressure values vs. ment. The trends of these two curves are similar and the bubble
time. radius calculated at any specific time is nearly the same as the
experimental measurement. The maximum radius in calcula-
below the pressure in the fluid field. The bubble will expand tion is 0.327, which is similar to the value of 0.322 obtained
again because of the pressure difference at t ’ 54.9 ms. from the experiment. The error between them is less than 5%.
Consequently, the numerical results agree well with the Also, the numerical results, including the summit, lowest
experimental results in describing the bubble motion, which point, and the displacement of center position, share the same
verifies the validity of the hybrid algorithm presented in this trend with the experimental results. However, the lower
paper. surface and center position curves during the last stage of
Pressure curves obtained from calculation results and collapse are steeper. For another, the bubble surface collapses
experimental measurement at the test point located 0.5 m with increasing speed and quickly floats upward.
horizontally from the center of the column charge are compared In conclusion, the results can be obtained through the
in Figure 2. Solid line and dashed line denote numerical and comparison in bubble dynamic behavior between calculation
experimental results respectively. According to Figure 2, it can and experiment, with the bubble undergoing the process of
be seen that the shock wave propagates rapidly in water after expansion, contraction, and finally developing a jet. The
initiation and reaches the test point at t ’ 0.30 ms, with an numerical results (e.g., bubble position, size, and shape, etc.)
exceedingly high pressure up to 19.8 MPa. The error of agree well with the experimental results, which conforms to the
pressure peak value between experiment and calculation is basic laws of bubble movement throughout the whole process of
less than 5%. Afterward, the shock wave decays exponentially underwater explosion; during the charge detonating stage, the
in the water. It is obvious that the tendency of the numerical pressure attenuation tendency of numerical calculation agrees
results is in good agreement with that of the experimental well with that of the experiment; in the pulsing and jetting
results. As a result, the validity of using the axisymmetric LS- phases, numerical results (e.g., displacements of upper surface,
DGF-RKDG model to simulate the column charge detonation is lower surface, and center position) are consistent with the
verified. experimental results. All of the above results sufficiently prove
Figure 3 illustrates the comparison of bubble radius and the the effectiveness and accuracy of the hybrid algorithm to solve
displacement curves of the upper surface, lower surface, and the problem of column charge underwater explosion.
The convergence tests of the LS-DGF-RKDG method and
BEM are taken in this section. In the LS-DGF-RKDG method,
the pressure peak values vs. the grid cells at the test point
located 0.3 m horizontally from the center of the column charge
are shown in Figure 4. We note that the grid cell is N and
discuss the different cases (N¼ 86,400, 117,600, 153,600,
194,400, 240,000, 290,400, 345,600, 470,400 and 614,400). As
this figure shows, the pressure peak values in cases where N¼
345,600, 470,400, and 614,400 are basically consistent, but the
cases where N¼ 470,400 and 614,400 require a significantly
larger amount of computation than the case where N¼ 345,600.
However, the accuracy is low in cases where N , 345,600.
Therefore, Nis chosen as 345,600 in the LS-DGF-RKDG
method in this paper to ensure accuracy.
Figure 3. Comparison curves between numerical (BEM) and experimental
The bubble maximum radius vs. the node numbers at the
results of bubble radius, upper surface, lower surface, and center position.
Comparison results for bubble radius and displacement vs. time proves the bubble boundary (it is set as 8, 12, 16, 24, 32, 48, and 64
validity of our presented hybrid algorithm. respectively) are compared in Figure 5. It is obvious that the
calculated value gradually converges with the increase of the

Journal of Coastal Research, Vol. 00, No. 0, 0000


0 Zhang et al.

Figure 5. Bubble maximum radius vs. node numbers, showing a good


convergence of BEM.

node numbers. When the bubble is discretized to 64 nodes, the


accuracy can be guaranteed. Hence, we choose 64 as the node
number at the boundary in the BEM.

Underwater Explosion Simulation of Column Charge Figure 6. Model of column charge underwater explosion near a rigid wall,
near a Rigid Wall showing the detailed model of an underwater explosion.
The numerical model of column charge underwater explosion
near a rigid wall is shown in Figure 6, where a ¼ 80 cm, b ¼ 120
40 MPa. As Figure 7(h) shows, the jet has an impact directly on
cm, r ¼ 0.65 cm, l ¼ 4.37 cm, and d ¼ 39.2 cm. The coordinate
the wall and a high-pressure region is developed between the
system is marked in the figure as well as four test points,
wall and the bubble during the later stage of the collapse. The
namely A (0.5, 0), B (0, 0.5), C (0.39, 0), and D (0, 0.39). The
bubble continues to shrink and radiate the pressure wave
quadrilateral meshes are evenly distributed and there are
outward. Finally, the bubble expands again because the
345,600 grid cells in total.
pressure inside the bubble is higher than that in the fluid field.
Using the numerical model presented above, the character-
In addition, the pressure on the wall drops because of the
istic of bubble motion near a rigid wall caused by the
attenuation of the jet, as shown in Figure 7(i).
underwater column charge explosion was studied. Figure 7
illustrates the bubble shapes and flow characteristics in the Influence of Slenderness Ratio on Load Characteristics
case of k ¼ 3.35, ch ¼ 1.2 with the charge installed vertically. It In this section, the influence of the slenderness ratio on the
can be seen from Figure 7(a) that the initial shape stays shock wave is discussed. Five cases are selected to study the
cylindrical and the pressure peak value in the radial direction is vertical column charge explosion when ch ¼ 1.2 using LS-DGF-
higher than that in the axial direction. Then, the ellipsoid- RKDG method, i.e. k ¼ 1.05, 2.20, 3.35, 4.49, and 5.62
shaped bubble gradually turns into a spherical one. Also, owing respectively.
to the higher pressure inside, the bubble pressure wave is Figure 8 shows the process from column charge initiation to
generated. It is obvious that the impedance of the rigid wall is ellipsoidal bubble formation when k ¼ 1.05, 3.35, and 5.62
higher than that of the fluids. As a result, a compression wave is respectively. It is obvious that an ellipsoidal bubble is formed
reflected when the shock wave transmits from the fluid of low during this process. It can be seen from Figure 8(a) that shock
impedance to the rigid wall of high impedance, as shown in waves with extremely high pressure are generated and that
Figure 7(b). The bubble keeps expanding until it reaches its they rapidly propagate into the water at the time of 0.03 ms.
maximum size at 35 8 ms in Figure 7(c). The lower surface of the The pressure peak value of these three cases can be up to
bubble exhibits a flat feature since the lower part of the bubble about 220, 230, and 250 MPa respectively from left to right. It
undergoes a great repulsion with the decrease of the distance indicates that the pressure peak value rises with the increase
between the lower surface of the bubble and the rigid wall. of the slenderness ratio. Besides, when k ¼ 1.05, the bubble
Subsequently, the bubble begins to shrink because of the lower retains a spherical shape and the distribution of shock wave
pressure inside the bubble than that in the fluid field. Owing to pressure in the radial and axial directions is almost the same.
the attraction of the wall and the buoyancy effect, the bubble is This phenomenon is similar to that of a spherical charge
stretched in the vertical direction and transforms from a sphere detonation. With the propagating of the shock wave, a kind of
into an ellipsoid as shown in Figure 7(d) and Figure 7(e). spherical wave, the pressure peak value decays exponentially
Because the downward Bjerknes force, which makes the bubble with time in the water as shown in Figure 8(b). At t ’ 0.15 ms
attracted by the rigid boundary, is larger than the upward the peak value falls to about 40 MPa, which is an order of
buoyancy force, the bubble presents nonspherical collapse at t ’ magnitude smaller than that at 0.03 ms. At the same time,
59.1 ms. At the same time, Figure 7(f) shows that a jet with high comparing the shapes in different cases, it is obvious that for
speed is formed toward the rigid wall. As the jet penetrates the k ¼ 1.05, the bubble stays spherical; for k ¼ 3.35, the initial
bubble, the bubble transforms from a singly-connected domain bubble develops from an ellipsoid into a sphere; as for k ¼
into a doubly-connected domain and becomes a toroidal shape 5.62, the bubble keeps an ellipsoid shape. As Figure 8(c)
as shown in Figure 7(g). A high-pressure region is induced in shows, the shock wave reaches the rigid wall and a
the jet area of the bubble and the pressure peak value is about compression wave is produced at t ’ 0.35 ms. At this time,

Journal of Coastal Research, Vol. 00, No. 0, 0000


Underwater Explosion Bubble near a Rigid Wall 0

Figure 7. Process from detonation to bubble jet; from (a) to (c) the times are: t ’ 0.15, 12.4, and 35.8 ms; from (d) to (f) the times are: t ’ 56.1, 57.8, and 59.1 ms;
from (g) to (i) the times are: t ’ 59.4, 60.6, and 61.3 ms. This figure presents the entire process of a column charge subjected to underwater explosion and bubble
motion near a rigid wall. (Color for this figure is available in the online version of this paper.)

the wave pressure drops down to about 15 MPa. Here, one Figure 9 shows the pressure peak value with slenderness
point is that the pressure calculated around the symmetry ratio as argument at different test points, namely A, B, C, and
axis has relatively high precision, without any nonphysical D, which are marked in Figure 6. Test points A and C are
condition such as the oscillation. located 0.39 m and 0.5 m horizontally away from the center of

Journal of Coastal Research, Vol. 00, No. 0, 0000


0 Zhang et al.

Figure 8. Pressure nephograms after column charge detonation for different


Figure 10. Bubble shapes at different times. (a) and (b) present the case of k
slenderness ratios; from left to right the slenderness ratios are: 1.05, 3.35,
¼ 1.05; (c) and (d) present the case of k ¼ 5.62. (a) and (c) are during the
and 5.62. Pressure nephograms at (a) t ¼ 0.03 ms, (b) t ¼ 0.15 ms, (c) t ¼ 0.35
expansion phase at 0.2, 3.04, and 30.1 ms; (b) and (d) are during the collapse
ms.
phase at 30.1, 55.5, and 59.1 ms. This figure shows bubble shapes for
different slenderness ratios.
the column charge. It can be seen that there is a slight increase
in peak value with the increase of slenderness ratio at these The bubble shapes during expansion phase are shown in
two points. As for test points B and D, they are located 0.39 m Figures 10(a) and 10(c) for k ¼ 1.05and 5.62 respectively. It
and 0.5 m vertically away from the center of the column charge. can be seen that in the case of k ¼ 1.05, the bubble grows
The peak value drops obviously with the increase of k. The peak almost spherically, whereas the bubble transforms from
values with the charge placed horizontally and vertically are initial ellipsoid into a sphere in the case of k ¼ 5.62. Owing to
nearly the same in the case of k ’ 1.05. For all other cases, the the influence of the rigid wall, the bubble side near the wall is
peak value in the vertical direction is higher than that of the slightly flattened for both cases. At t ’ 30.1 ms, the bubble
horizontal direction. reached its maximum size after expansion in the two cases.
According to the results presented above, slenderness ratio After that, the bubble began to collapse. Because of the
has great influence on the initial shape and velocity of the conjunction of the Bjerknes force and the buoyancy force, the
bubble, which further affects the bubble dynamic characteris- bubble would gradually present an ellipsoidal shape. Subse-
tics. Therefore, the influence of slenderness ratio on the bubble quently, a high-pressure region develops between the rigid
dynamics is discussed in this section. wall and the bubble, and as a result a high-speed jet that

Figure 9. Comparison of the pressure peak at different test points. (a) Test points A and B, located 0.39 m away from the center of the column charge; (b) test
points C and D, located 0.5 m away from the center of the column charge. The predicted results for pressure peaks with the increase of slenderness radios at
different test points are shown.

Journal of Coastal Research, Vol. 00, No. 0, 0000


Underwater Explosion Bubble near a Rigid Wall 0

Table 1. Parameters of bubble motion for different slenderness ratios.

Jet Jet
Slenderness Velocity Width
Case Ratio k Installation T (ms) Rm (m) (m/s) (m)
1 1.05 Vertical 65.64 0.3251 155.46 0.1281
2 2.20 Vertical 65.58 0.3247 169.15 0.1255
3 3.35 Vertical 65.61 0.3248 189.55 0.1210
4 4.49 Vertical 65.59 0.3247 196.85 0.1175
5 5.62 Vertical 65.60 0.3252 219.95 0.1149

penetrates the surface of the bubble is induced toward the


rigid wall. Comparing the two cases of k ¼ 1.05 and k ¼ 5.62,
although the initial bubble shapes are different, other Figure 12. Jet velocities for different slenderness ratios.
results, such as maximum radius, pulsation cycle, and
vertical displacement of bubble center, etc., are almost the maximum radius. However, the bubble lower surface of the
same. Owing to the different initial shapes and velocities of horizontal case is closer to the rigid wall than that of the
the bubble, the bubble shape during the collapse phase is vertical case when the bubble volume reaches the maximum
different. For k ¼ 1.05, the jet is wider and bubble volume is size. This is probably because for the horizontal case, larger
smaller than at k ¼ 5.62.
velocity at the top and bottom is obtained when the bubble
The slenderness ratio is carefully chosen to efficiently study
reaches its maximum volume, and the stronger attraction
its influence on bubble dynamics and the results are listed in
effect happens during the collapse phase. As Figure 11(a)
Table 1. It can be observed that the slenderness ratio has little
shows, the initial shape and jet width are almost the same for
effect on the maximum radius (about 0.325 m) and the first
both cases when k ¼ 1.05. As shown in Figures 11(b) and 11(c),
pulsation cycle (about 65.6 ms). However in contrast, both jet
the charge position has an effect on the initial shape. The jet
velocity and jet width are affected by the slenderness ratio,
width of the horizontal case is larger than that of the vertical
with a rise from 155 m/s to 220 m/s and a decline from 0.1281 m
case when k ¼ 3.35 and 5.62.
to 0.1149 m respectively.
Figures 12 and 13 illustrate the jet velocity and jet width as a
Influence of Installation (Horizontal and Vertical) on function of slenderness ratio in both the horizontal and vertical
Bubble Dynamics cases. Obviously, the jet velocity is smaller and the jet width is
The installation of the charge also has an important influence larger when the charge is placed vertically. The jet velocity
on the initial bubble shape and the distribution of velocity. diminishes and the jet width increases with the rise of
Therefore, comparisons are also made between the vertical case slenderness ratio k when the charge is installed horizontally.
and the horizontal case. To be specific, the cylindrical charges However, the trend is the opposite in the vertical case.
are horizontally installed for cases 6–8 and the results are Influence of Distance Parameter on Bubble Dynamics
compared with those in cases 1, 3, and 5 presented in the To study the nonspherical bubble dynamic features near a
previous section (i.e. vertically installed). The compressible rigid wall with the increase of distance parameter ch, the
BEM is adopted to conduct the numerical simulation of bubble compressible BEM is used to simulate the bubble of vertical
motion with other parameters fixed as ch ¼ 1.2, k ¼ 1.05, 3.35, column charge, choosing ch ranging from 1.0 to 1.6, i.e. ch ¼
and 5.62 in cases 6–8 respectively. 1.0, 1.2, 1.4, and 1.6 as well as charge slenderness ratio k ¼
Figure 11 gives bubble shapes during different phases in 3.35.
cases of k ¼ 1.05, 3.35, and 5.62 (from left to right) for both The bubble maximum shape and pressure contouring for
installations (horizontal and vertical). It can be seen that cases of ch ¼ 1.0, 1.2, 1.4, and 1.6 are shown in Figure 14. When
installation has little influence on the pulsing cycle and the the bubble reaches its maximum radius in Figure 14(a), it can

Figure 11. Bubble shapes for different installations (horizontal and vertical) during different phases, with k¼1.05, 3.35, and 5.62 corresponding to (a), (b), and (c);
0.15 ms, 30.1 ms, and 59.1 ms correspond to the times of initial bubble shape, maximum volume, and minimum volume.

Journal of Coastal Research, Vol. 00, No. 0, 0000


0 Zhang et al.

Figure 13. Jet widths for different slenderness ratios.


Figure 15. Comparison of vertical migration time curves of bubble center.
The predicted results for vertical migration with the increase of the distance
be observed that the repulsive force diminishes as the distance parameter are shown.
parameter ch decreases. In other words, the closer the bubble
lower surface is to the rigid wall, the larger the repulsive force collapse stage, but not before collapse that is before 0.062
on the bubble is. Besides, the bubble shapes gradually change ms. It can be seen from the figure that the jet velocity
from flat to spherical with the increase of the distance gradually increases with the rise of the distance parameter
parameter ch. During the jet phase, the attracting force on ch. In addition, for all the cases of ch ranging from 1.0 to 1.6,
the bubble toward the rigid wall (the Bjerknes effect), the the numerical results are all in good agreement with the
volume, and the width decrease with the increase of the
experimental results presented in the reference (Philipp
distance parameter ch as shown in Figure 14(b). On the
and Lauterborn, 1998), including bubble shape, cycle,
contrary, the pressure at the upper surface of the bubble
displacement, and jet velocity, etc.
increases and the bubble center rises with the increase of the
distance parameter ch.
CONCLUSIONS
Figures 15 and 16 show the vertical displacement of the
Underwater explosion can cause undesirable damage to
bubble center and the jet velocity curves as a function of the
coastal structures that contributed to the development of
distance parameter ch. k is fixed as 3.55 and ch is chosen as
civilization. It clearly indicates that the damage response of
1.0, 1.2, 1.4, and 1.6, respectively. As Figure 15 shows, the
structures to underwater explosion should be seriously
bubble is pushed away from the rigid wall a little during
investigated. A combined LS-DGF-RKDG-BEM algorithm is
expansion phase. To be specific, the closer the bubble is to
used to build the LS-DGF-RKDG explosion model and
the rigid wall, the larger the repulsive force is and the bigger
bubble dynamics model based on the wave motion equation.
the upward vertical displacement of the bubble center is.
The load characteristics of an underwater column charge
However, during the collapse phase, the bubble is pulled
explosion near a rigid wall in compressible fluid is simulated
back rapidly to the rigid wall. With the decrease of the
and analyzed. In the simulation, the shock wave propaga-
distance parameter ch, the attraction effect becomes stron-
tion rule and the bubble dynamics characteristics in
ger, and the upward vertical displacement of the bubble
different cases are calculated and studied. These results
center diminishes. In addition, the first cycler appears to
can provide preference for the safety assessment and the
drop down with the increase of distance parameter ch.
protective design of structures, which has a certain
Figure 16 gives the bubble jet velocity as a function of
application value in military affairs and civilian engineer-
different distance parameters ch. Obvious differences be-
ing.
tween results with different ch can be observed in the final

Figure 14. Pressure nephograms in the cases of (a) maximum volume and
(b) bubble jet; from left to right the distances are: ch ¼ 1.0, 1.2, 1.4, and 1.6.
This figureshows bubble shapes when the bubble volume reaches maximum Figure 16. Comparison of jet velocity time curves. The predicted results for
and bubble jets for different distance parameters. jet velocities with the increase of the distance parameter are shown.

Journal of Coastal Research, Vol. 00, No. 0, 0000


Underwater Explosion Bubble near a Rigid Wall 0

ACKNOWLEDGMENTS one-dimensional case. Journal of Computational Physics, 222(1),


This research is sponsored by the National Natural Science 353–373.
Qiu, J.; Liu, T.G., and Khoo, B.C., 2008. Simulations of compressible
Foundation of China (grant no. U1430236, 51479041, two-medium flow by Runge–Kutta discontinuous Galerkin methods
51279038). with the Ghost Fluid method. Communications in Computational
Physics, 3(2), 479–504.
LITERATURE CITED Rajendran, R. and Narasimhan, K., 2001. Damage prediction of
clamped circular plates subjected to contact underwater explosion.
Blake, J.R.; Taib, B.B., and Doherty, G., 1986. Transient cavities near
International Journal of Impact Engineering, 25(4), 373–386.
boundaries. Part 1. Rigid boundary. Journal of Fluid Mechanics,
Wang, C. and Khoo, B.C., 2004. An indirect boundary element method
170, 479–497.
for three-dimensional explosion bubbles. Journal of Computational
Blake, J.R.; Taib, B.B., and Doherty, G., 1987. Transient cavities near
Physics, 194, 451–480.
boundaries. Part 2. Free surface. Journal of Fluid Mechanics, 181,
Wang, G.H. and Zhang, S.R., 2014. Damage prediction of concrete
197–212.
gravity dams subjected to underwater explosion shock loading.
Cockburn, B. and Shu, C.W., 1998. The Runge–Kutta discontinuous
Engineering Failure Analysis, 39, 72–91.
Galerkin method for conservation laws. V. Multidimensional
Wang, L.K.; Chen, H.L.; Ye X., and Yao, X.L., 2015. Study on load
systems. Journal of Computational Physics, 141, 199–224.
characteristics of underwater explosion using RKDG-LS-DGF and
Cole, R.H., 1948. Underwater Explosion. Princeton, NewJersey:
BEM. Shock and Vibration, 2015, 165252.
Princeton University Press, 162p.
Wang, L.K.; Zhang, Z.F., and Wang, S.P., 2016. Pressure character-
Fedkiw, R.P.; Aslam, T.; Merriman, B., and Osher, S., 1999. A non-
istics of bubble collapse near a rigid wall incompressible fluid.
oscillatory eulerian approach to interfaces in multimaterial flows
Applied Ocean Research, 59, 183–192.
(the Ghost Fluid Method). Journal of Computational Physics, 152,
457–492. Wang, Q.X., 2013. Non-spherical bubble dynamics of underwater
Geers, T.L., 1971. Residual potential and approximation methods for explosions in a compressible fluid. Physics of Fluids, 25, 1–21.
three dimensional fluid-structure interaction problems. Journal of Wang, Q.X., and Blake, J.R., 2010. Non-spherical bubble dynamics in
the Acoustical Society of America, 49, 1505–1510. a compressible liquid. Part 1. Travelling acoustic wave. Journal of
Geers, T.L., 1978. Doubly asymptotic approximations for transient Fluid Mechanics, 659, 191–224.
motions of submerged structures. Journal of the Acoustical Society Wang, Q.X.; Yeo, K.S.; Khoo, B.C., and Lam, K.Y., 1996a. Nonlinear
of America, 64(5), 1500–1508. interaction between gas bubble and free surface. Computers &
Geers, T.L., and Felippa, C.A., 1983. Doubly asymptotic approxima- Fluids, 25, 607–628.
tions for vibration analysis of submerged structures. Journal of the Wang, Q.X.; Yeo, K.S.; Khoo, B.C., and Lam, K.Y., 1996b. Strong
Acoustical Society of America, 73(4), 1152–1159. interaction between a buoyancy bubble and a free surface.
Hung, C.F. and Hwangfu, J.J., 2010. Experimental study of the Theoretical and Computational Fluid Dynamics, 8, 73–88.
behaviour of mini-charge underwater explosion bubbles near Wu, G.X. and Hu, Z.Z., 2004. Simulation of non-linear interactions
different boundaries. Journal of Fluid Mechanics, 651, 55–80. between waves and floating bodies through a finite-element-based
Kim, S.G. and Gitterman, Y., 2013. Underwater explosion (UWE) numerical tank. Proceeding of Royal Society A, 460, 2797–2817.
analysis of the ROKS Cheonan incident. Pure and Applied Ye, X; Sun, L.Q., and Pang, F.Z., 2015. Dynamics and noise radiation
Geophysics, 170, 547–560. of multiple bubbles in compressible fluid using boundary integral
Klaseboer, E.; Hung, K.C.; Wang, C.; Khoo, B.C.; Boyce, P.; Debono, equation. Engineering Computations, 32(3), 885–913.
S., and Charlier, H., 2005. Experimental and numerical investiga- Zhang, A.M.; Cui, P.; Cui, J., and Wang, Q.X., 2015. Experimental
tion of the dynamics of an underwater explosion bubble near a study on bubble dynamics subject to buoyancy. Journal of Fluid
resilient/rigid structure. Journal of Fluid Mechanics, 537, 387–413. Mechanics, 776, 137–160.
Ming, F.R.; Zhang, A.M.; Xue, Y.Z., and Wang, S.P., 2016. Damage Zhang, A.M.; Li, S., and Cui, J., 2015. Study on splitting of a toroidal
characteristics of ship structures subjected to shockwaves of bubble near a rigid boundary. Physics of Fluids, 27, 062102.
underwater contact explosions. Ocean Engineering, 117, 359–382. Zhang, A.M. and Liu, Y.L., 2015. Improved three-dimensional bubble
Osher, S. and Fedkiw, R.P., 2003. Level Set Methods and Dynamic dynamics model based on boundary element method. Journal of
Implicit Surfaces. New York: Springer. Computational Physics, 294, 208–223.
Park, J., 2008. A Coupled Runge Kutta Discontinuous Galerkin– Zhang, A.M.; Wang, S.P.; Huang, C., and Wang, B., 2013. Influences
Direct Ghost Fluid (RKDG-DGF) Method to Near-Field Early-Time of initial and boundary conditions on underwater explosion bubble
Underwater Explosion (UNDEX) Simulations. Blacksburg, Virgin- dynamics. European Journal of Mechanics B/Fluids, 42, 69–91.
ia: Virginia Polytechnic Institute and State University. Zhang, A.M.; Wang, S.P., and Wu, G.X., 2013. Simulation of bubble
Philipp, A. and Lauterborn, W., 1998. Cavitation erosion by single motion in a compressible liquid based on the three dimensional
laser-produced bubbles. Journal of Fluid Mechanics, 361, 75–116. wave equation. Engineering Analysis with Boundary Elements, 37,
Prosperetti, A. and Lezzi, A., 1986. Bubble dynamics in a compress- (9), 1179–1188.
ible liquid. Part 1. First-order theory. Journal of Fluid Mechanics, Zhang, S.R.; Wang, G.H.; Wang, C.; Pang, B.H., and Du, C.B., 2014.
168, 457–478. Numerical simulation of failure modes of concrete gravity dams
Qiu, J.; Liu, T.G., and Khoo, B.C., 2007. Runge–Kutta discontinuous subjected to underwater explosion. Engineering Failure Analysis,
Galerkin methods for compressible two-medium flow simulations: 36, 49–64.

Journal of Coastal Research, Vol. 00, No. 0, 0000

You might also like