Escolar Documentos
Profissional Documentos
Cultura Documentos
th
Australasian Congress on Applied Mechanics, ACAM 6
12-15 December 2010, Perth, Australia
E
C =
0
is the elastic stress wave speed in pressure bars, E and are Youngs modulus
and the density of pressure bars, respectively,
0
L and
0
A (
4
2
0
0
d
A
t
= ; where
0
d is the diameter of
the specimen) are the original length and cross-section area of the specimen and A is the cross-
section area of pressure bars, ) (t
I
c ; ) (t
R
c and ) (t
T
c are the recorded incident, reflected and
transmitted strain pulses with the time being shifted from the strain gage locations to the interfaces
between pressure bars and specimen according to the elastic wave speed in pressure bars.
Figure 2 shows a schematic of the dynamic bulge testing set-up. We considered a movable bulge
cell with natural rubber as pressure carrying medium which can be used to perform dynamic bulge
tests in a conventional SHPB system. When bulging with rubber, sealing problems and the possibility
of leakage of the high-pressure liquid employed in hydraulic bulging are eliminated. The need for the
filling and removal of fluid or the cleaning of the bulged specimen after forming is eliminated. The
insertion of the rubber is quick and convenient and the rubber can be re-used [7].
Figure 2: Schematic of the dynamic bulge testing set-up.
3 Theoretical analysis
3.1 Membrane theory
For a thin spherical shell expanded uniformly by internal pressure, the membrane stress is given very
closely by the approximation
d
d
t
pR
2
= o (4)
where p is the bulge pressure and
d
t ,
d
R are the thickness and radius at the top of the dome,
respectively. The equivalent strain can be calculated using the sheet thickness:
|
|
.
|
\
|
=
0
ln
t
t
d
c (5)
The radius at the top of the dome can be calculated by
d
d c d c
c
d
h
h R h R
d
R
2
2
2
2
2
+
|
|
.
|
\
|
+ |
.
|
\
|
= (6)
where
c
R is the radius of the fillet of the cavity,
c
d is the diameter of the cavity and
d
h is the dome
height.
Chakrabarty and Alexander [8] developed analytical methods to describe the deformation in the
hydraulic bulge test. They assumed that the shape of the bulge is spherical. With this assumption, the
thickness at the top of the dome can be calculated by the following equation:
n
c d
d
d h
t t
|
|
.
|
\
|
+
=
2
2
0
) / 2 ( 1
1
(7)
where n , is the strain hardening coefficient.
3.2 Dynamic bulge theory
The input bar of SHPB is used to measure the bulging pressure. The rubber pressure in the bulge cell
is determined directly from the incident strain pulse ) (t
I
c and the reflected strain pulse ) (t
R
c at the
input bar/rubber interface:
)] ( ) ( [ ) ( t t E t p
R I
c c + = (8)
The corresponding input bar/rubber interface velocity is:
)] ( ) ( [ ) (
0
t t C t u
R I in
c c + = (9)
The velocity of the bulge cell ) (t u
out
is determined from the transmitted wave ) (t
T
c
at the output
bar/die interface.
) ( ) (
0
t C t u
T out
c = (10)
The effective bulge velocity is the difference of interface velocities,
) ( ) ( ) ( t
out
u t
in
u t u = A (11)
The effective bulge displacement can be determined by integration of (11) as a function of the
measured strain histories:
( ) t t c t t c t c d d C t u
t
T R
t
I
) ( )] ( ) ( [ ) (
0
} }
+ + = A (12)
3.3 Mechanical behavior of sheet metal
The J ohnson-Cook [9] material model is used to analyze the mechanical behavior of OFHC copper.
This model is particularly suited to model high strain rate deformation of metals in adiabatic transient
dynamic analysis. It expresses the equivalent von-Mises flow stress as a function of the equivalent
plastic strain, strain rate, and temperature. In quasi-static conditions, metals work harden along the
well-known relationship which is known as parabolic hardening rule:
n
kc o o + =
0
(13)
where
0
o is the yield stress of the metal, n is work hardening exponent and k is the exponential
factor. Dynamic events often involve increases in temperature due to adiabatic heating and so the
thermal softening must be included in the constitutive model. The effect of temperature on the flow
stress can be described by following relation:
(
(
|
|
.
|
\
|
=
m
r m
r
r
T T
T T
1 o o (14)
where
m
T is the melting point,
r
T is a reference temperature at which reference stress,
r
o is
measured and m is material dependant constant. The strain rate effect can be simply expressed with
following relationship, which is very often observed at strain rates that are not too high.
c o ln (15)
Based on the above relations, J ohnson and Cook [9] presented the following equation for strength
model, where the von-Mises flow stress is given as:
] ) ( 1 )][ ln( 1 ][ ) ( [
*
0
m n
T C B A + + =
c
c
c o
(16)
where m n C B A , , , , are material constants which are experimentally determined; c is the effective
plastic strain rate and
0
c is the reference strain rate which can for convenience be made equal to 1
(
1
0
1
= s c ). The expression in the first set of brackets gives the stress as a function of strain for
1
0
=
|
|
.
|
\
|
c
c
and 0
*
= T . The expressions in the second and third sets of brackets represent the effects
of strain rate and temperature. The homologous temperature
*
T , is the ratio of current temperature T
to the melting temperature
m
T .
r m
r
T T
T T
T
=
*
(17)
where
r
T is the reference temperature at which
0
o is measured. The J ohnson-Cook constants for
OFHC copper are listed in Table 1.
Table 1: Material constants for OFHC copper sheet.
) (
3
m
kg
) (MPa A ) (MPa B C
n m
8960 90 292 0.025 0.31 1.09
3.4 Hyper-elasticity
The uniaxial compression stress-strain diagrams of natural rubber measured by SHPB apparatus at
different strain rates are illustrated in Figure 3. As can be seen from Figure 3, as the strain rate
increases, the strength of the material increases. Rubber-like materials have very little compressibility
compared to their shear flexibility and have nonlinear stressstrain characteristics for relatively large
deformations. Under such conditions, they are generally assumed as nearly incompressible. To model
these hyper-elastic materials through FEM, a constitutive law based on total strain energy density W
has to be adopted [10]. ABAQUS allows the automatically evaluation of hyper-elastic material behavior
by creating response curves using selected strain energy potentials. Among several approaches,
Ogden [11] theory is used based on the polynomial development of total strain energy. The Ogden
material model has previously been used with success to predict the behavior of rubber materials at
high strain rates (see e.g., Ramezani et al. [10]). The form of the Ogden strain energy potential is:
ij
ij
W
c
o
c
c
= (18)
i el
i
N
i i
i
N
i
J
D
W
i i i
2
1
3 2 1
2
1
) 1 (
1
) 3 (
2
+ + + =
= =
o o o
o
(19)
where W is the strain energy per unit of reference volume;
i
are the deviatoric principal stretches
which can be defined by
i i
J
3
1
= ;
i
are the principal stretches;
el
J is the elastic volume ratio;
and
i
,
i
o and
i
D
are temperature-dependent Ogden constants. Compressibility can be defined by
specifying nonzero values for
i
D , by setting the Poisson's ratio to a value less than 0.5, or by
providing test data that characterize the compressibility. We assumed a fully incompressible behavior
for natural rubber with 4997 . 0 = v and
i
D
equal to zero and so the second expression in (19) can be
eliminated. To determine the strain energy density W, ABAQUS uses a least-squares fitting algorithm
to evaluate the Ogden constants automatically from experimental data.
Figure 3: Stress-strain behavior of natural rubber under compression at different strain rates.
4 Experiments and finite element simulations
The bulge cell is composed of a thick-walled rubber chamber and a die (Figure 2). For experiments,
the round OFHC copper sheet specimen of thickness 1mm is clamped between the chamber and the
die. The input bar is inserted into the chamber which is filled with natural rubber to transmit the
pressure from the input bar to the sheet surface. The outer diameter of the SHPB pressure bars match
the inner diameter of the container and die. The pressure bars are made of silver steel and have
1500mm length and 12mm diameter. Several collars support the pressure bars, allowing it to slide
freely and to remove any bending waves due to an impact. When the striker bar impacts the input bar
at a defined velocity, a compressive stress wave is generated propagating towards the input
bar/rubber interface. This stress wave is transmitted through the rubber and ultimately causes the
bulging of the sheet specimen while both the bulge cell and the output bar are accelerated [12].
Throughout each experiment, the incident, reflected and transmitted waves are measured at the
center of the pressure bars by strain gages attached to the bars. The strain gage signals are recorded
using a HYLDE FE-H359-TA amplifier and a PICOSCOPE 3206, 200MHz digital oscilloscope. The
experimental set-up is shown in Figure 4.
Figure 4: Dynamic bulge cell set-up mounted on SHPB.
In order to simulate dynamic bulge forming, a finite element model is built in commercial software
ABAQUS. An explicit nonlinear approach with negligible temperature effects is assumed for
simulations. By taking advantage of axisymmetry, it is possible to simulate the process as 2-D
axisymmetric model. The die and rubber container are made of mild steel and are modeled as linear
elastic bodies with the Youngs modulus GPa E 206 = , Poissons ratio 3 . 0 = v , and mass density
3
m
kg
7800 = . Natural rubber is used to bulge an OFHC copper blank with diameter 60mm and
thickness of 1mm. One of the major requirements for computer simulations is the incorporation of
material properties through realistic models. Rubber is modeled as a hyper-elastic material, as
explained in Section 3.4. The J ohnson-Cook material model is used to simulate the behavior of metal
at high strain rates. The experiments were performed using two different striker bar velocities; i.e.,
s m V
st
/ 11 =
and s m V
st
/ 14 = . An empirical relationship was found between the striker bar velocity
and natural rubber strain rate at SHPB system, which is
0
2L
V
st
~ c (20)
where
st
V is the striker bar velocity and
0
L is the length of rubber. The initial length of the natural
rubber in the dynamic bulge test is 30mm and according to (20), the approximate strain rate of natural
rubber during the test will be
1
180
s and
1
230
s .
Since the model is developed by taking advantage of axisymmetry, the component nodes at the
symmetry edges are restrained in the appropriate directions. The end of output bar is also constrained
in order to model the momentum trap. The Coulomb friction model with the coefficient of friction of
0.25 is used to model the interface between rubber and sheet [7]. All other contact surfaces are
modeled as friction-free interfaces. The interactions between all components are modeled using an
automatic surface to surface contact algorithm. All geometric entities are modeled using CAX4R
elements. CAX4R is a 4-node bilinear axisymmetric quadrilateral, reduced integration, hourglass
control element. The pressure bars mesh comprises only one element row in the radial direction. The
pressure bars are made of silver steel and are modeled as linear elastic with the Youngs modulus of
GPa E 214 = and the mass density of
3
/ 7830 m kg = . The Poisson ratios are set to a non-
physical value of 0 = v . Thus, uniaxial waves in the computational model are not altered when
traveling along the bar axis [10].
The initial velocity is applied to the striker bar to impact the input bar. The simulation begins with the
input bar in contact with natural rubber. The copper blank is then introduced between the die and the
flexible rubber. After the striker bar hits the input bar, the impact stress pulse travels throughout the
input bar and upon arrival to input bar/rubber interface, a part of that reflects back to the input bar and
the rest of the stress pulse travels toward the rubber and makes the rubber to move down and bulge
the sheet. The axisymmetric finite element model of the high strain-rate rubber-pad forming at the end
of the process with s m V
st
/ 14 = is shown in Figure 5.
Figure 5: Axisymmetric finite element model of the bulge cell at the end of the process.
5 Results and discussions
Figure 6 shows the strain signals at the middle of pressure bars for experiments performed at striker
bar velocity of 14m/s. The incident strain pulse rises to its plateau strain level of about
5
10 194
during a time interval of about 19s. The time profile of the incident wave is of rectangular shape and
its amplitude remains quite constant during the simulation. The pulse shape of reflected wave is
triangular and reaches the maximum strain of
5
10 95
which happens at 15s. The reflected strain
signal becomes compressive after time duration of 34s. The transmitted strain pulse arrives at the
die/output bar interface after 39s and reaches the maximum amplitude of
5
10 120
at the end of
the simulation.
Figure 6: Representation of the incident, reflected and transmitted strain signals at striker velocity of
14m/s.
Figure 7 shows the impact pressure at input bar/rubber interface with impact velocity of 14m/s
evaluated according to (8). As can be seen from the figure, the pressure-time history shows an initial
peak at about 17s and then the pressure level increases monotonically until it reaches its maximum
level at 96s. The maximum pressure is about 142 and 173MPa respectively for the experiments
carried out at the striker velocities of 11 and 14m/s. subsequently, the pressure amplitude decreases
because of the end of the incident pulse. To compare the results of experiments and finite element
simulations, the pressure-time history at the input bar/rubber interface is monitored directly throughout
the simulations. The comparison of the curves demonstrates the good correlation between
experimental and simulation results. In general the simulation tends to predict slightly lower peak
pressure than the experimental and theoretical analysis. The maximum errors at the peak pressures
are 4.1% and 4.3% respectively at striker velocities of 11 and 14m/s.
Figure 7: Pressure-time history at input bar/rubber interface.
The strain history of the sheet during dynamic bulging process calculated using (5) is shown in Figure
8. According to Figure 8, the sheet deformation starts at about 40s after the input bar impacts the
rubber and reaches the maximum strain of 0.26 and 0.38 at striker velocities of 11 and 14m/s.
Combining Figures 7 and 8, we arrive at Figure 9 which shows the pressurestrain curve of OFHC
copper sheet at two different impact velocities during high strain-rate bulge test. As depicted in Figure
9, as the strain rate (impact velocity) increases, the formability of the material increases. In general, an
increase in the strain rate also increases the flow stress. However, when the strain rate is increased
significantly, only a small fraction of the work required to deform the material is stored; the remainder
is converted into heat. This results in adiabatic conditions, which lead to work softening [10].
Figure 8: Strain history of copper sheet during bulge forming.
Figure 9: Pressurestrain diagram of OFHC copper at different impact velocities.
6 Conclusions
In this paper, dynamic bulge test was investigated experimentally and numerically. The main
conclusions of this research are summarized below:
- The incident strain pulse is compressive and rectangular shape and it remains quite constant
during the simulation. The pulse shape of reflected wave is triangular in the tensile range.
- The pressure-time history at input bar/rubber interface shows an initial peak at about 17s and
then the pressure level increases monotonically until it reaches its maximum level of about
142 and 173MPa respectively for the striker velocities of 11 and 14m/s.
- Comparison of pressure-time curves obtained by experiments and finite element simulations
show good correlation between experimental and simulation results.
- As the strain rate increases, the strength of OFHC copper sheet increases.
References
1. P. Broomhead and R.J . Grieve, 1982, Effect of strain rate on the strain to fracture of a sheet steel
under biaxial tensile stress conditions, J Eng Mater Technol, 104(2), pp. 102106.
2. M. Atkinson, 1996, Accurate determination of biaxial stress-strain relationships from hydraulic bulging
tests of sheet metals, Int J Mech Sci, 39(7), pp. 761769.
3. G. Gutscher, H.C. Wu, G. Ngaile and T. Altan, 2004, Determination of flow stress for sheet metal
forming using the viscous pressure bulge (VPB) test, J Mater Process Technol, 146, pp. 17.
4. V. Grolleau, G. Gary and D. Mohr, 2008, Biaxial Testing of Sheet Materials at High Strain Rates Using
Viscoelastic Bars, Exp Mech, 48, pp. 293306.
5. M. Ramezani, Z.M. Ripin and R. Ahmad, 2010, Plastic bulging of sheet metals at high strain rates, Int
J Adv Manuf Technol, 48(9-12), pp. 847858.
6. H. Kolsky, 1949, An Investigation of the Mechanical Properties of Materials at Very High Rates of
Strain, Proc Roy Phys Soc, B 62, pp. 676700.
7. M. Ramezani, Z.M. Ripin and R. Ahmad, 2009, Computer aided modelling of friction in rubber-pad
forming process, J Mater Process Technol, 209, pp. 49254934.
8. J . Chakrabarty and J .M. Alexander, 1970, Hydrostatic bulging of circular diaphragms J Strain Anal
5(3), pp.155161.
9. G.R. J ohnson and W.H. Cook, 1983, A constitutive model and data for metals subjected to large
strains, high strain rates and high temperatures, Proceedings of the seventh international symposium
on ballistic, The Hague, The Netherlands, pp. 541547.
10. M. Ramezani, Z.M. Ripin, R. Ahmad, H.M. Akil and M. Damghani, 2010, High strain rate bulge forming
of sheet metals using a solid bulging medium, Proc IMechE B: J Eng Manuf, 224(2), pp. 257270.
11. R.W. Ogden, 1972, Large deformation isotropic elasticity on the correlation of theory and experiment
for incompressible rubberlike solids, Proc R Soc Lond, A 326, pp. 56584.
12. M. Ramezani and Z.M. Ripin, 2010, Combined experimental and numerical analysis of bulge test at
high strain rates using split Hopkinson pressure bar apparatus, J Mater Process Technol, 210, pp.
10611069.