FINITE ELEMENT ANALYSIS OF
RESPONSE OF A FLOATING STRUCTURE
TO AN UNDERWATER EXPLOSION (UNDEX)
M.Sc. Thesis by
Fatih ARUK
Supervisors: Prof. Dr. Tuncer
TOPRAK
Dr. Ergün
BOZDAĞ
What is the importance of UNDEX resistance?
•
Warships should last destructive effects of any near
underwater explosion
.
•
Also shipboard systems must be shock hardened to a
certain level to ensure combat survivability of both
personnel and equipment
.
So,
shock resistance
is a major issue that should be
considered during early design phase of...
warships
, radars, weapons, torpedos or any
other shipboard equipment
Shock Trials (Physical Tests);
•
Extremely Expensive
•
Dangerous
•
Harmful to the surrounding Environment
•
Require years of planning and preparation
•
Due to safety risk, shock trials do
not test up to the ship’s design
limits or even the true wartime
shock environment.
•
These tests are performed after the
first ship is already built.
Shock trials of USS WINSTON S. CHURCHILL
(DDG 81) in 2001
•United States Navy spent tens of millions of dollars
•
Years of planning and preparation
Shipboard eqipment testing; MIL-S 901D
(Military Specification for shock testing of ship
board equipment)
Shock Test
Platform
Impact Tests;
expensive experimental tests on simple cylindrical
shells and plate structures.
Computational modeling and response
,
if perfected, can effectively and
accurately replace the experimental
procedures used to obtain the
UNDEX
UNDEX Simulation;
The shock response of an immersed or floating structure is
obtained when subjected to a near UNDEX loading
•
Kwon and Cunningham;
dynamic responses of stiffened cylinder and
beam elements
DYNA3D [Explicit FE Code]+
Underwater Shock Analysis (USA) [BE Code
based on DAA]
•
90s Kwon and Fox;
the nonlinear dynamic response of a cylinder
subjected to side-on underwater explosion
•
Sun and McCoy
UNDEX analysis of a composite cylinder
ABAQUS + a fluid-structure interaction code
•Cichocki, Adamczyk, and Ruchwa
implemented fluid-structure interaction
phenomenon, pressure wave distribution, and
the radiation boundary conditions into
Three dimensional ship shock trial simulation of a
UNDEX PHENOMENA;
Similitude Relations (Pressure versus Time)
1( )
( , )
A c ca
P R t
P
R
f
τ
+
=
B c c ca
v t
R
a
τ
=
( )
f
τ
=
e
−
τ
τ
≤
1
1.338
1.805
( ) 0.8251
0.1749
f
τ
=
e
−
τ
+
e
−
τ
τ
≤
7
;
P
;
R
;
c
a
;
t
Pressure
, , , ;
c
c
P v A B
Distance to charge (stand-off distance)
Charge radius
Time
0 1 2 3 4 5 6 x 10-3 0 2 4 6 8 10 12 14 P (M p a ) t(s)
Pressure vs. time history for 25kg of HBX-1 charge, standoff distance of 10m according to Swisdak according to Price
UNDEX PHENOMENA;
Similitude Relations (Pressure versus Time)
Material
Source
TNT (1.52 g/cc)
Coles (1946)
1.42
992
0.13
0.18
TNT (1.60 g/cc)
Farley and Snay (1978)
1.45
1240
0.13
0.23
TNT (1.60 g/cc)
Price (1979)
1.67
1010
0.18
0.18
HBX-1 (1.72 g/cc)
Swisdak (1978)
1.71
1470
0.15
0.29
HBX-1 (1.72 g/cc)
Price (1979)
1.58
1170
0.144
0.24
Explosive Gas Bubble
1 3 5 5 6
(
10)
cm
T
K
D
=
+
1 3 max 6 1 3(
10)
cm
a
K
D
=
+
Explosive Gas Bubble
UNDEX PHENOMENA;
Gas Bubble
Period
Max. Gas Bubble
Radius
For our case
( MIL-S-901 D; 27.3 kg HBX-1
at 7.3 m depth)
T=0.64 s
amax =4.4 m
•
Bubble pulses are a strong source of excitation
for ships whose bending vibration mode is near
to the bubble pulse frequency
•
It is especially important for the
late time
response
of the ship
Explosive Gas Bubble,
Geers-Hunter Bubble Model (2002)
UNDEX PHENOMENA;
UNDEX
=
SHOCK WAVE PHASE
+
BUBBLE OSCILLATION PHASE
SHOCK WAVE PHASE
provides initial
conditions for
BUBBLE OSCILLATION
PHASE
ABAQUS includes Geers-Hunter (2002) model for UNDEX loading.
(a fourth-order Runge-Kutta integrator to prescribe the pressure variation)
MIL-S-901 D; 27.3 kg HBX-1 at
7.3 m depth
27.3 kg HBX-1 at
65 m depth
UNDEX PHENOMENA;
Cavitation
Cavitation takes place in water when there is
area of near-zero absolute pressure (about 206.8
Pa)
Two types of cavitation occur in an UNDEX event;
‘bulk’
cavitation;
a large volume of low
pressure due to reflections
from sea surface
‘local’
cavitation;
a small zone of low
pressure at
fluid-structure interaction
surface.
The effect of cavitation on the response of the floating
structures is important and must be properly modeled
to obtain physically meaningful results.
UNDEX PHENOMENA;
Cavitation
‘bulk’
cavitation
0
i
atm
stc
R
P P
+
+
P
+
P
=
Cavitation condition:
(
j
2
j
1
)
c
f
R
R
t
c
−
=
( , ) 0
F x y
=
( , ) 0
G x y
=
Equations of lower and upper
cavitation boundaries:
EXPLOSION VARIATIONS ACCORDING TO
UNDEX PHENOMENA;
Cavitation
‘local’ cavitation
Taylor plate theory
Assumptions;
•
The plate is rigid
•
The shock wave is planar
p
i
r
v
= −
v
v
i
f
f i
r
f
f r
P
c v
P
c v
ρ
ρ
=
=
max2
2
p t p f f p idv
m
c v
P
P e
dt
θρ
−+
=
=
max
2
( )
(1
)
t
t
p
p
P
v t
e
e
m
β θ
θ
θ
β
−
−
=
−
−
max
2
( )
1
t
t
p
P
P t
e
θ
β
e
β θ
β
−
−
=
−
−
As becomes large (a
lightweight plate), cavitation
occurs faster.
UNDEX PHENOMENA;
Cavitation
‘local’ cavitation
max
2
( )
(1
)
t
t
p
p
P
v t
e
e
m
β θ
θ
θ
β
−
−
=
−
−
max
2
( )
1
t
t
p
P
P t
e
θ
β
e
β θ
β
−
−
=
−
−
cavt t
≤
cavt t
≤
Incident and total pressures behind, and velocity of shock platform
Elements of UNDEX Simulation;
Acoustic Equations
Equilibrum equation for small motions of a compressible, adiabatic fluid
with velocity-dependent momentum loses;
The slow flow assumption is
accurate for
steady fluid velocities up to
Mach 0.1
Acoustic Constitutive Equation;
Acoustic Constitutive Equation
for
cavitating fluid
;
(
)
{
}
max
v
,
c
o
p
=
p
p
−
p
linear
nonlinear
Elements of UNDEX Simulation;
Formulation of Direct Integration, Coupled Acoustic-Structural Analysis
Introducing a variational field δp, integrating over
entire body and applying Green’s theorem
yields;
( )
f1
fp
T
ρ
− −
∂
=
⋅
= − ⋅
∂
x
n u
n
x
&&
Boundary traction
term
Elements of UNDEX Simulation;
Formulation of Direct Integration, Coupled Acoustic-Structural Analysis
fp
S
the value of the acoustic
pressure is
prescribed;
ft
S
prescribes motion of the
fluid
particles, modeling
pressure wave
( )
0
f
ft
T
x
=
T
=
n u
−
⋅
&&
fi
S
the radiating acoustic boundary,
waves passing exclusively outward
fs
S
acoustic-structural
interaction
Elements of UNDEX Simulation;
Formulation of Direct Integration, Coupled Acoustic-Structural Analysis
0 1 1
1
1
1
1
0
f ft fi fs f f f f V S m S Sp
p
p
p
p
dV
pT dS
K
K
p
p
p dS
p
dS
c
a
γ
δ
δ
δ
ρ
ρ
δ
δ
−
∂
∂
+
+
⋅
−
+
∂
∂
+
+
−
⋅
=
∫
∫
∫
∫
x
x
n u
&&
&
&
&&
:
0
fs tm
m
m
m
c
V
V
V
m
m
S
S
dV
dV
dV
p
dS
dS
δε
α ρδ
ρδ
δ
δ
+
⋅
+
⋅
+
+
⋅
−
⋅
=
∫
∫
∫
∫
∫
σ
u u
u u
u n
u t
&
&&
the final
variational
statement for
the acoustic
medium
the virtual work
statement for a
structure.
Elements of UNDEX Simulation;
Formulation of Direct Integration, Coupled Acoustic-Structural Analysis
●
The Discretized Finite Element Equations
interpolation functions
up to the number of displacement degrees of freedom.
up to the number of pressure nodes
[ ]
M
f
{ }
p
&&
+
[ ]
C
f
{ }
p
&
+
[
K
]
f
{ }
p
=
[
S
f
s
]
{ } { }
u
&&
+
P
f
[ ]
M
s
{ }
u
&&
+
[ ]
C
s
{ }
u
&
+
[ ]
K
s
{ }
u
=
[
S
f
s
]
T
{ } { }
p
−
P
s
Surface-Based Acoustic-Structural Interaction Procedure
Elements of UNDEX Simulation;
Note that the source point
should be located out of the fluid domain.
Incident Wave Loading
Elements of UNDEX Simulation;
Elements of UNDEX Simulation;
Mesh Refinement
For reasonable accuracy,
at least six representative
internodal intervals of
the acoustic mesh should fit into the shortest acoustic
wavelength
present
in the analysis.
Eight or more will be better.
; maximum linear element length
; the speed of sound
;the number of linear elements per acoustic
wavelength
; max. frequency of excitation which can be simulated accurately
max
max
1500
8 0.05
3750
m s
f
m
f
Hz
≤
⋅
≤
We used an element size of about 50 mm around the
acoustic-structural interface. The element size
increases up to 150 mm at outer fluid regions.
Meshing whole fluid
medium with 50 mm
elements would result in
about 16 million
Explicit Time Integration
Elements of UNDEX Simulation;
( 1) ( ) 1 1 ( ) ( ) 2 2 ( )2
i i N N i i i Nt
t
u
u
+u
+ −∆
+ ∆
=
+
&&
&
&
•
An explicit central-difference time integration rule is used;
•
Each increment is relatively inexpensive because there is
no solution for a set of simultaneous equations.
•
The time increments must be quite small so that the
accelerations are nearly constant during an increment.
Advantages of the explicit time integration method;
Explicit Time Integration
Elements of UNDEX Simulation;
•
Well-suited to solving high-speed dynamic events
•
No global tangent stiffness matrix. Iterations and
tolerances are not required.
●
Stability of Explicit Integration;
●
The Stable Time Increment Estimation;
If damping
included;
Structural Damping
Explicit Time Integration
Elements of UNDEX Simulation;
Mass
Proportion
al Damping
Stiffness
Proportion
al Damping
In names of
natural freq.;
damps
lower
frequenci
es
damps
higher
frequencie
s
Effects of damping on the stable time
increment in Explicit Analysis
R
β
has greater effect on
stable time increment
UNDEX ANALYSIS METHODOLOGY
UNDEX TEST PARAMETERS FROM MIL-S-901D
MIL-S-901D
specification which covers shock testing requirements for
ship board machinery, equipment, system and structures.
heavyweight shock
testing platform
UNDEX TEST PARAMETERS FROM MIL-S-901D UNDEX FE MODEL GENERATION FLUID-STRUCTURE INTERACTION FE ANALYSIS CORRELATION OF UNDEX RESPONSES AND VALIDATION
OF NUMERICAL CODE CONDUCT UNDEX TEST SHORT DURATION DYNAMIC RESPONSE Equivalen t? SHORT DURATION DYNAMIC RESPONSE N O MODIFY UNDEX MODEL/ANALYSES PARAMETERS Y E S
UNDEX ANALYSIS METHODOLOGY
UNDEX CORRELATION METHODOLOGY
UNDEX ANALYSIS METHODOLOGY
SUBMODELING ANALYSIS
MODELING AND ANALYSIS
Weight; about 39
tones
CAD MODELING;
CATIA
V15
MODELING AND ANALYSIS
MESHING;
ABAQUS CAE
Number of nodes:
140316
Number of elements:
142327
Linear
quadrilateral
elements of type S4R
Connections were imposed
by means of kinematic
MODELING AND ANALYSIS
GENERATING A REDUCED (COARSE) MODEL for
tryouts and acoustic mesh convergence studies;
HYPERMESH
Number of nodes: 7858
Number of elements:
8229
Linear quadrilateral
elements of type S4R
538
Linear triangular
UNDEX ANALYSIS WITH REDUCED (COARSE) MODEL
FLUID MESH CONVERGENCE ANALYSIS
Boundary Conditions
11
f ff
c
=
ρ
K
11
2
f f f ff
a
K
β
γ
ρ
ρ ρ
= ⋅
+
Plane type radiating
surfaces;
f
=
1
0
UNDEX ANALYSIS WITH REDUCED (COARSE) MODEL
FLUID MESH CONVERGENCE ANALYSIS
Acoustic-Structural
Interaction
Initial Static Pressure
Incident Wave (UNDEX)
Loading
UNDEX ANALYSIS WITH REDUCED (COARSE) MODEL
FLUID MESH CONVERGENCE ANALYSIS
UNDEX ANALYSIS WITH REDUCED (COARSE) MODEL
FLUID MESH CONVERGENCE ANALYSIS
UNDEX ANALYSIS WITH REDUCED (COARSE) MODEL
ANALYSIS WITH DEFORMABLE PLATFORM
UNDEX ANALYSIS WITH REDUCED (COARSE) MODEL
ANALYSIS WITH DEFORMABLE PLATFORM;
EFFECT OF DAMPING
1.5
R
α
=
0.5
6
R
E
β
=
−
For first two modes; % 0.4
For the first torsional and
bending modes ; 0.25 %
and 0.15 %
UNDEX ANALYSIS WITH REDUCED (COARSE) MODEL
ANALYSIS WITH DEFORMABLE PLATFORM;
UNDEX ANALYSIS WITH MAIN (FINE) MODEL
EFFECT OF MESH REFINEMENT AROUND
ACOUSTIC-STRUCTURAL INTERACTION REGION
Mesh
Convergence
Analysis;
Element size;
150 mm
Number of
DFT Analysis of
Incident Pressure
Waves;
Element size;
50
mm
Number of nodes;
UNDEX ANALYSIS WITH MAIN (FINE) MODEL
EFFECT OF MESH REFINEMENT AROUND
UNDEX ANALYSIS WITH MAIN (FINE) MODEL
EFFECT OF CAVITATION
UNDEX ANALYSIS WITH MAIN (FINE) MODEL
EFFECT OF STRUCTURAL DAMPING
1.5
R
CONCLUSION AND OUTLOOK
•
Fluid mesh size has important effect
on the structural
response and should be selected carefully for
accurate results;
§
Mesh Convergence Study
§
DFT analysis of loadings
•
Though it requires a nonlinear fluid behavior which
adds to the cost of the analyses,
including cavitation
is a must
to obtain physically meaningful results.
•
The effect of damping was also shown to be
important for peak acceleration estimation in the
late time response of the platform.
•
Submodeling analysis can bu run to obtain
converged stress-strain results at some sub-region
of the structure.
•