Professional Documents
Culture Documents
Abstract
A polymer coating applied to the wet surface of a ship’s hull can reduce some types of damage to a ship subjected to a
far-field underwater explosion. Results from numerical simulations are compared with full-scale underwater explosion
test data for a ship with and without polymer coating to study the effectiveness of such coatings in blast-resistant design.
The polymer coating is a thin wall structure applied to the wet surface of the ship hull, while the ship is a relatively large
structure. Hence, an effective approximate simulation method for the full-scale ship with polymer coating is developed.
Based on the homogenization theory of polymer coating, a continuum model that takes into account the strain rate
dependence is introduced. The shock response subjected to underwater explosion of the finite element ship model is
investigated. The results obtained from underwater explosion tests and numerical simulation are compared and dis-
cussed. The blast-resistant property of polymer coating is elaborated. Mesh sensitivity and the validity of the polymer
coating equivalent continuum model are studied.
Keywords
Polymer coating, underwater explosion, homogenization theory, strain rate dependence, shock response spectrum
combatant ships. All-metal sandwich plate or polymer is located on the right side at the ship mid-section. The
coating applied to the ship wet surface can buffer and charge location is listed in Table 1. The keel shock fac-
absorb significant energy in crushing under intense tor (KSF) can be expressed as
impulsive load. Xue had verified that square honeycombs pffiffiffiffiffi
as cores for all-metal sandwich plate were effective in W 1 þ sin b
KSF ¼ 3 ð1Þ
enhancing the blast-resistant property in that they com- D 2
bine excellent crushing strength and energy absorption.
where W is the weight of trinitrotoluene (TNT) charge
A continuum model of square honeycomb core has been
(kg); D is the distance between the charge and keel (m)
introduced that can be used to simulate core behavior in
and b is the angle between the keel–charge straight line
large structural calculations, but not feasible to mesh the
and the horizontal line (rad). A charge of 5 kg TNT is
detailed core geometry.10
used in the test.
This study aims to develop an effective simulation
A photograph of the full-scale UNDEX test is
method, which is based on the homogenization theory
shown in Figure 2. The water depth of the test area is
of polymer coating,10,11 to investigate the shock
10–15 m.
response of ship subjected to far-field UNDEX and
verify the coating’s effect in blast-resistant design. Six
UNDEX tests for full-scale surface ship with and with- Equivalent continuum model of polymer
out polymer coating are preformed. The nonlinear coating
finite element tool ABAQUS is employed to investigate
the ship with polymer coating. The numerical results The polymer coating is a thin wall structure, which is
are compared with the test data. composed of a top plate, bottom plate and core structure.
A unit cell of the coating is presented in Figure 3(a). The
inner part of the core is a tube, and the outer is a thin
Experiment wall hexagon. The core is 50 mm in height, while the tube
is 40 mm in outer diameter and 2.20 mm in thickness and
Six UNDEX full-scale ship tests were performed to the hexagon is 46.19 mm in outer edge length and 4.00
study the shock response of surface ship with and with- mm in thickness. The coating is formed by unit cell’s lat-
out polymer coating. As shown in Figure 1, the charge eral and longitudinal extension. Its material is a type of
neoprene with shore hardness of 65.
Xue had investigated the continuum model of square
honeycomb cores, which can be used to simulate core
behavior in large structural calculations when it is not
feasible to mesh the detailed core geometry.10 Similarly,
the equivalent continuum constitutive model of the
coating is an approximate simulation method to replace
detailed mesh of core members in coating structure in
large finite element model based on homogenization
theory, thereby reducing the size of the numerical simu-
lation. The equivalent continuum constitutive model of
the coating is shown in Figure 3(b).
Figure 4(a) shows the coating’s effective stress–strain
curve of the detailed model and equivalent continuum
model at strain rate of 100 s21. Overall strain e is
defined as the coating’s macro-characteristic, which
neglects the local distortion of the core. Overall strain
Figure 1. Schematic representation of the UNDEX tests. is expressed as
Polymer coating Charge depth, H (m) Horizontal distance from keel, L (m) KSF (kg1/2 m21)
Figure 3. Unit cell of the coating: (a) detailed model and (b) equivalent continuum model.
Figure 4. Stress–strain curve of the coating: (a) data of detailed model and continuum model at strain rate of 100 s21 and (b) data
of equivalent continuum model at different strain rates.
Strain rate of 1 s21 Strain rate of 10 s21 Strain rate of 100 s21
Stress Strain Stress Strain Stress Strain
0 0 0 0 0 0
320,000 0.1 375,000 0.04 380,000 0.04
320,000 0.7 510,000 0.07 800,000 0.07
1,440,000 0.8 600,000 0.1 810,000 0.1
1,700,000 0.874 700,000 0.65 840,000 0.64
1,283,000 0.74 2,000,000 0.74
3,065,000 0.8 4,200,000 0.8
5,230,000 0.85 14,500,000 0.865
7,300,000 0.874
Figure 6. Schematic representation of the ship right part and measurement locations.
horizontal distance of 10.2 m, generating a peak pres- Experiment and simulation comparison
sure of 7.19e6 Pa with a time constant of 0.000215 s. The shock responses of the ship are measured with 14
The UNDEX simulation model in ABAQUS/
accelerometers. The ship right part and measurement
Explicit includes the following:
locations are shown in Figure 6. The accelerometers
A5, A6, A7 and A11 measure the vertical and trans-
1. The total wave formulation is used in simulation, verse shock responses, else measure only the vertical
and the incident wave loading is applied on the responses. A1, A2, A7, A10 and A14 are located on the
fluid surface that wetting the ship. keel. A8 is located on the first longitudinal. A13 is
2. The zero acoustic pressure boundary condition is located on the hull above the waterline. A4, A9 and
satisfied at the nodes on the free surface, and the A12 are located on the lower deck. A3, A5 and A11 are
initial acoustic static pressures are specified on the located on the main deck. A6 is located on the
acoustic medium nodes. superstructure.
3. Cavitation effect is taken into account in the The acceleration-time test data are filtered using a 5-
UNDEX simulation, and the fluid cavitation limit Hz Butterworth high-pass filter to eliminate zero drift
is defined as 0 Pa. and a 5-kHz low-pass filter. The velocity–time data are
4. The reflection plane over seabed is specified to cal- acquired from the integration of the acceleration his-
culate additional incident wave load due to reflec- tory. The peak velocities are presented in Table 3.
tion. The seabed plane is located 12 m below the Corresponding to the six tests presented in Table 1,
fluid-free surface, and the reflection coefficient six numerical simulations with the same charge location
value is 1.2. have been investigated. Simulations 1–3 are about the
5. The fluid outer surface is specified as a non- ship without coating. Simulations 4–6 are about the
reflecting spherical boundary. The spherical non-
ship with coating. The locations recorded that the simu-
reflecting condition provides an approximation:
lation results are as the same as that in the tests.
acoustic wave is transmitted across such a bound-
Table 4 shows the peak velocities at different situa-
ary with little reflection of energy back into the
tions. The hull peak velocity of the ship with coating
acoustic medium.
decreases about 1.0%–41.2% compared to that without
coating. The coating could reduce the hull dynamic
Define the parameters responses subjected to UNDEX. Most structure loca-
1 1 1 1 g tions’ peak velocity of ship with coating is smaller than
¼ pffiffiffiffiffiffiffiffiffi and ¼ þ pffiffiffiffiffiffiffiffiffi that of ship without coating, including vertical and
c1 rf Kf a1 rf r1 2rf rf Kf
transverse responses. However, the vertical peak velo-
The transient sphere radiation boundary condition city of ship with coating at A3 and A6 is not attenu-
can be written as ated, and even a little larger than that of ship without
coating.
∂p 1 1 1 Compared to the data in Table 3, the simulation
n ¼ p_ þ p ð6Þ results have the same regulation as the test data. The
∂X rf c1 a1
coating could attenuate the keel peak velocity greatly
where rf is the density of the fluid, Kf is the bulk modu- but have less influence on the other locations at the
lus of the fluid, r1 refers to the radius of the sphere, g is ship structure and even enhance the shock response at
the ‘‘volumetric drag’’ (force per unit volume per velo- a few locations such as superstructure top middle point.
city), n is the normal vector pointing into the fluid, p is However, the reduction of test data is more than that
the excess pressure in the fluid (the pressure in excess of in simulation. The blast-resistant property of the coat-
any static pressure) and x is the spatial position of the ing is underestimated in simulation. Figures 7–10 show
fluid particle. the shock response spectrum comparison at points A7
/: no data measured.
(vertical), A14, A12 and A6 (vertical) in test 5 and simulation 2, the rigid body vertical displacement of the
simulation 5, respectively. ship is 82 mm at 0.25 s; the vertical deformation peaks of
Simulation results and experimental data agree well A7, A11 and A6 are 12.4, 22.5 and 27.0 mm, respectively.
with Figures 7–10, which further validated the dynamic While, in simulation 5, the rigid body displacement
simulation analysis of the ship model subjected to far- of the ship is 88 mm at 0.25 s; the vertical deformation
field UNDEX. Simulation results in the shock response peaks of A7, A11 and A6 are 7.9, 13.2 and 22.7 mm,
spectrum are a little different from the experimental respectively, which indicates that the polymer coating
data in a wide range of frequencies, which can be could decrease the vertical deformation of the hull and
explained, as follows: (a) the model is not precise ship structures greatly.
enough; (b) parameters such as the damping are not The compared peak velocities of ship keel between
accurate enough and (c) deviation introduced by the simulations 2 and 5 are presented in Figure 12. The
continuum constitutive model of polymer coating. shock response of the ship with polymer coating is
much less than that without polymer coating, and the
reduction is from 6.2% to 38.5%. The effect to attenu-
Simulation results and discussion ate the shock response is more prominent in the ship
Figure 11 shows the vertical deformation of the ship in mid-section than that in the prow and stern sections.
simulations 2 and 5 at locations A7, A11 and A6. In The other locations on the hull have the same rule as
Figure 8. Shock response spectrum comparison at point A14. Figure 10. Shock response spectrum comparison at point A6
Solid black line for test 5; dashed red line for simulation 5. (vertical).
Solid black line for test 5; dashed red line for simulation 5.
that on the keel in simulation. Overall, it proves that milliseconds and then decays with exponential mode.
the polymer coating is effective to protect the hull sub- However, the dynamic responses of ship structures are
jected to the UNDEX. much different from that on the hull. Compared to
The hull maximum vertical velocity appears at the point A8, low-frequency component of the response at
right side under the waterline, which is the minimum A4 is more dominant.
distance of hull location from the charge. The vertical In simulation 2, the vertical peak velocity at A7,
peak velocity of the right side of the hull is a bit larger A14 and A1 appears at 1, 6 and 10 ms, while in simula-
and commences earlier than that of the left side. tion 5, the vertical peak velocity at A7, A11 and A1
Figure 13 shows the velocity–time history at points appears at 3, 10 and 17 ms. The shock wave reaches at
A8 and A4 in simulation 5. They represent the typical the mid-section first and then transmits to the stern
velocity–time response of hull and other structures, and prow section. The coating makes the keel peak
respectively. A8 is located on the first longitudinal. As velocity commences a little later.
a point on the hull, dynamic response at A8 subjected Physically, in simulations 1–3, the incident shock
to UNDEX increases to a peak value in about several wave impinges on the hull directly, and evidently, the
Figure 11. Vertical deformation of the ship: (a) simulation 2 and (b) simulation 5.
Figure 12. Vertical peak velocity of ship keel in simulations 2 Figure 14. Vertical peak velocity of ship keel in simulations 1–3.
and 5.
Figure 15. Vertical peak velocity of ship keel in simulations Figure 17. Mesh sensitivity study.
4–6.