Journal of Materials Processing Technology 172 (2006) 139–145

Experimental evaluation and finite element simulation of explosive

forming of a square cup from a brass plate assisted by a lead plug
V.N. Wijayathunga ∗ , D.C. Webb
School of Engineering, Leeds Metropolitan University, Calverley Street, Leeds LS1 3HE, UK
Received 2 September 2002; received in revised form 20 March 2003; accepted 5 September 2005

Abstract
A FE model is developed to simulate the experimental tests for the impulsive deep drawing of a brass square cup with the presence of a lead
plug. The loading of the assembly is achieved by the detonation of a high explosive, underwater. The formability of the brass plate in the absence
of the lead plug gives rise to material instability and uneven material thickness over the formed region causing rupture. The presence of lead plug
initially slows down the forming process by acting as a reservoir for the kinetic energy. As the deformation progresses, kinetic energy in the plug is
transferred to the plate and the increased time scale of the operation enables the system to be treated as a quasi static forming process. In addition to
postponing the onset of material instability, the presence of the lead plug enables a higher ratio of drawability and a better uniformity in thickness
of the formed dome. The experimental investigations have indicated that the optimum shape of the plate specimen is a square blank with circular
cut-off segments. The effects of the medium impedance, wave reflection and refraction are considered to be negligible in order to improve the
simplicity of the modelling procedure. Finally, the computed results are compared with those obtained from experimental tests.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Explosive forming; Square cups; Impulse loading; Stand-off distance; Plug cushion

1. Introduction formability limit an alternative technique referred to as ‘plug

cushion forming’ has been introduced [9,10]. A plug of soft
In recent years, the application of sheet metal parts in material is inserted in contact with plate, in between the source
production has become more frequent. Scientific, numerical of explosion and the plate. Thus, the pressure wave instead
and engineering advances in sheet forming operations are made reaching the plate directly exerts itself on the plate via the plug.
at a rapid rate. Hence, accurate and efficient simulations of Depending on the thickness of plate and plug as well as the
these operations have become significantly important. There angle of the plug chamfer, the deformation process is slowed
are many investigations pertaining to axi-symmetric and square down. Since the plug is made of a relatively soft material, its
cup drawing operation with reference to formability limit. A relative energy absorption capacity is low. This allows it to act
representative example could be found in references [1–5]. In as a conduit to transfer the energy at a relatively slower rate
all these cases the sheet metal is stretched and formed in to to the plate. Experiments conducted for brass plate specimen
required shape. Explosive forming is a technique that is gen- and lead plug are specified in reference [10]. The measured
erally adopted for sheet metal forming operations that involve thickness strain is compared with numerically simulated results
very large sheets and usually axi-symmetric. Description of the produced using ABAQUS FE Explicit code version 6.2 [11] and
method and its applications could be found in references [6–8]. the compatibility between the two is found to be acceptable.
The method is essentially dynamic and the forming is induced
by exposing sheet metal surface to an incoming pressure wave
2. Materials, procedure and equipment
generated by underwater explosion. But the direct application
of the pressure wave on the metal surface gives rise to a 2.1. Blank material
stress–strain field distribution of unacceptable profile, which
may lead to premature failure of material. Hence, to improve the 18 SWG (1.2192 mm) brass squares of 203.2 mm (8 in.) sides with cor-
ners machined circular to 215.9 mm (8.5 in.) diameter (see Fig. 1). The static
stress–strain curve for this material obtained for the ‘as received’ condition
∗ Corresponding author. employing the bulge test is given in Fig. 5.

0924-0136/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2005.09.014
140 V.N. Wijayathunga, D.C. Webb / Journal of Materials Processing Technology 172 (2006) 139–145

3. Experimental procedure

A grid of measured concentric circles is printed on the blank,

which is positioned and clamped as shown in Fig. 2. The
explosive charge is placed centrally at a ‘stand-off’ distance of
300 mm. With the plug in position, the detonator and shot firing
cable are attached and the whole assembly is lowered into the
tank to the required hydrostatic head (610 mm). After obtaining
the desired vacuum and the circuit checked with an ohmmeter,
Fig. 1. Blank. the exploder is connected and the charge detonated. Measure-
ments of the grid positions are then retaken.
2.2. Explosive materials
4. Experimental results
The explosive used in the experimental investigations is a plastic sheet explo-
sive known commercially as ‘Metabel’. This explosive material is supplied in
the form of sheets 250 mm × 125 mm and 3.175 mm thick, and it is mouldable (a) The natural hoop strain, εr = ln(r/r0 ) and natural thickness
and has good water resistance. The calorific value of the explosive is not known strain, εt = ln(t/t0 ) are evaluated from the measurements
but its detonation velocity is 6.7 km/s and its density 1.53 g/cm3 . taken with respect to the grid printed on the brass plate.
Graphs of the principal natural hoop and thickness strain
2.3. Die details versus particle original radius are presented in Fig. 4.
(b) Photographs of some of the deformed profiles are shown in
The free forming of dome-shaped square cups is achieved using a die with
square ducts of 101.6 and 88.9 mm square, a corner radius of 3.175 mm and an Fig. 3.
entrance profile radius of 12.7 mm (see Fig. 2).
5. Finite element modelling procedure
2.4. Lead plug manufacture
Essentially, the modelling procedure addresses the issue of
A two-piece heated steel mould is used to produce 101.6 mm square castings,
provided with pipes to ensure freedom from cavities. These castings are then
multiple contacts associated with rigid (die and blank holder)
machined into square plugs of 82.55 mm side (minimum) with up to a 45◦ and deformable (plate and plug) parts. The general procedure
chamfer angle and thickness of 12.7 mm. adopted for the FE simulation consists of the following steps.

Fig. 2. Die details.

 V.N. Wijayathunga, D.C. Webb / Journal of Materials Processing Technology 172 (2006) 139–145 141

Fig. 3. Samples of impulsively drawn cups.

5.1. Model discretization In the FE simulation, the motion of a rigid body is com-
pletely governed by the motion of a single node, known as the
Eight-noded continuum elements with reduced integration rigid body reference node. The relative position of the nodes and
and hourglass control were used for the development of the plate therefore the elements that are part of the rigid body remains
and the plug. These elements have a better convergence rate fixed throughout the simulation. The motion of a rigid body
than tetrahedral elements. However, their better performance is can be prescribed by applying boundary conditions at the rigid
achieved only when they are approximately rectangular and the body reference node. Rigid bodies interact with the remainder
minimum angle at the intersection of faces is not less than 45◦ . of the model via the contact pair(s) constituting surfaces of the
The stiffness of the die and the blank holder is several orders deformable elements and rigid body surfaces. In this model four
of magnitude higher than that of the deformable plate and the different contact pairs (blank holder and plate, plate and die, plug
plug. The stress, strain, etc., response of the die and the blank base and plate, plug sides and plate) were defined to simulate
holder are not of interest to this study. Stable time increment the contact between blank holder, die and plate.
estimates for elements in stiff regions could result in very small
global time increments. Nodes and elements that are part of a 5.2. Material property definition
rigid body do not affect the global time increment. Using a rigid
body instead of a deformable element representation in a stiff Brass is defined as a linear elastic and non-linear isotropic
region can result in a much larger global time increment, with- hardening material. Data from Fig. 5 is utilised for the material
out significantly affecting the overall accuracy of the solution. property definition.
Therefore, in order to achieve better computational efficiency,
the die and the blank holder are modelled as rigid bodies using σ̄ = σyield + K(ε̄)n (1)
four-noded bilinear rigid elements. where σ̄ and ε̄ are stress and strain in the plastic region, respec-
tively. Values of 648 MPa and 0.551 are selected for the strength

Fig. 4. Natural loop strain and natural thickness strain derived from experiments. Fig. 5. Stress-strain data for brass obtained from tests.
142 V.N. Wijayathunga, D.C. Webb / Journal of Materials Processing Technology 172 (2006) 139–145

coefficient K and hardening exponent n. Density is taken as ule. In the absence of such a module, the alternative procedure
8.23 g/cm3 , the elastic modulus as 1.06 × 1011 N/m2 and the described below was adopted.
Poisson’s ratio as 0.34. For the plug material lead, density is Upon the detonation of the explosive material, it is immedi-
taken as 11.34 g/cm3 , the elastic modulus as 0.365 × 1011 N/m2 ately converted to gaseous products under very high pressures
and the Poisson’s ratio as 0.425. and temperatures. The high internal gas pressure is at a state
of dis-equilibrium with prevailing hydrostatic head and atmo-
5.3. Boundary conditions and friction spheric pressure and subsequent expansion occurs. The inertia
of the water impedes the radial expansion of the gas globe and
The model is assumed to have a double symmetry, hence only as a result a high intensity shock wave is propagated through
one quarter of the complete geometry needs to be discretized for the medium. For a constant shock wave speed the point on the
the purpose of the current analysis. As a matter of fact another plug closest to the explosive charge centre is loaded first and the
axis of symmetry exists and can be visualised by bisecting the remaining exposed surface is loaded according to:
existing system in half. This third axis of symmetry has not been
 
taken in to account in order to avoid using triangular section or 1 Q
p(t) = p0 exp − t− (2)
collapsed elements. The omission of the third axis of symmetry θ vw
does give rise to a more CPU intensive solution, but there is
a significant gain in the accuracy of the results. The die and where Q is the difference between the shortest distance and the
the blank holder are constrained against translation by applying distance to the point in consideration on the loaded surface, vw
fixed boundary conditions to the respective reference nodes. The the speed of pressure wave propagation and t is the time. The
friction coefficient employed for the relative motion of the blank values of p0 and θ are dependant on the charge weight w and the
against the die, is 0.05. stand-off distance h. The empirical formulae for the values of θ
and p0 for underwater explosion are given [9] by:
5.4. Underwater explosive loading  1
Mθ
1 w3
θ= kθ w 3   (3)
The general FE procedure for underwater pressure generation h
and simulation requires an underwater shock analysis mod-

Fig. 6. Deformed shape of the plate at various stages of the process.

 V.N. Wijayathunga, D.C. Webb / Journal of Materials Processing Technology 172 (2006) 139–145 143

Table 1
Values of constants for calculation of time constant and peak pressure

Explosive kθ × 106 Mθ A × 10−3 α

TNT 92.269 −0.18 13.826 1.13

Pentolite 92.269 −0.18 13.826 1.13
HBX-3 95.188 −218 13.247 1.14



1 α
w 3
p0 = A  (4)
h

The chosen constants of kθ , Mθ , A and α in SI units for different

Fig. 8. Energy plot for the plug.
charge material are given in Table 1 [9].
Upon the arrival of the pressure wave at the deforming target
material, its intensity is doubled by reflection. But, as the defor-
mation commences, the magnitude of the pressure is reduced
by ρc vp , where ρc is the impedance of the pressure transmitting
medium (water) and vp is the velocity of the deforming material
interface. Therefore, the actual pressure acting on the deforming
surface is given by:

 
1 Q
p(t) = 2p0 exp − t− − ρ c vp (5)
θ vw

Although it is possible to specify the pressure loading in the FE

coding exactly by Eq. (5), the relative magnitudes of various
quantities allows the use of an approximate form as: Fig. 9. Energy balance for the model.
t
p(t) = 2p0 exp − (6)
θ

5.5. Output from FE simulation

The deformed shape of the plate at various stages of the pro-

cess is shown in Fig. 6. The energy plot for the plate and the plug
are shown in Figs. 7 and 8, respectively, whereas energy balance
for the model is shown in Fig. 9. Fig. 10a shows the variation of
hoop strain, εr = ln(r/r0 ) and thickness strain, εt = ln(t/t0 ) along
a direction 45◦ to the diagonal of the plate for plug thickness
of 12.7 mm. Similarly, Fig. 10b shows the strain variation along
the diagonal direction of the plate.

Fig. 10. (a) Hoop strain (ln r/r0 ) and thickness strain (ln t/t0 ) variation with
distance along a direction 45◦ to the die diagonal. (b) Hoop strain (ln r/r0 ) and
Fig. 7. Energy plot for the plate. thickness strain (ln t/t0 ) variation with distance along the die diagonal.
144 V.N. Wijayathunga, D.C. Webb / Journal of Materials Processing Technology 172 (2006) 139–145

6. Discussion and moves away from the centre towards the sides of the cup.
Similarly, material in the outer region moves in and stretches
In underwater explosive forming of square cups, it is found over the die radius and forms the side of the cup. The plot ln(t/t0 )
that the insertion of a lead plug in contact with the plate, in is an indication of the thickness variation of the drawn cup. It is
between the plate and the source of explosion would result an evident that after the drawing operation thickness reduces at the
improvement in the strain field compared to the case in which pole relative to the initial thickness, but the distribution of the
the specimen is directly exposed to the pressure wave. If the thickness in this region is almost uniform. Thickness gradually
plate had been directly exposed to the explosion, the rapid increases when moving towards the upper end of the cup.
energy transfer would either cause tearing or an excessively non- The comparison of experimentally measured strains given
uniform thickness distribution in the formed product. in Fig. 4 and the values from the simulation given in Fig. 10a
Although in both cases the momentum delivered to a unit sur- and b shows a certain degree of incompatibility between the
face area remains the same, incorporation of a lead plug would experiments and FE simulation. This could be attributed to the
increase the mass per unit surface area. Hence, the process is following:
slowed down raising the formability limit. The initial velocity
in absence of the lead plug is, (a) The lack of perfect co-axiality in positioning the blank and
θ the plugs with the die assembly.
p(t) dt
v= 0
(7) (b) A layer of water with variable thickness could get trapped
ρbrass between the contacting surfaces, i.e. plug lower surface and
whereas when the plug is inserted the velocity reduces to, plate upper surface. As deformation initiates at the blank
periphery and moves towards the centre, the trapped water
θ
p(t) dt is pushed towards the centre. As a consequence of this, a rel-
v= 0
(8)
ρbrass + ρplug atively higher hydrostatic pressure (greater than the contact
pressure) is generated and this could give rise to a greater
where ρ is density per unit surface area and θ is the explosive strain than what is predicted by the simulation.
time constant given by Eq. (3). In slowing down the forming
process the dynamic operation is shifted towards quasi-static
response operation and hence it produces uniform thinning in 7. Conclusions
the blank, giving rise to a better formability limit. The essence
and workability of the operation are related to the relative energy Implementation of FE simulation and the comparison of sim-
absorbing capacity of the plug and the plate. The energy required ulated results with the experimental evidence are established.
by the plug for deformation is far less than the amount required The presented FE procedure is easy to implement and less expen-
by the plate due to the soft nature of the plug material and the sive than conducting experimental investigation. Because of the
absence of constraints upon it. Hence, the plug acts as a reser- establishment of reasonable parity between experiments and FE
voir, which stores energy during the initial stage of the process it is perhaps prudent and more economical for a complex forming
and subsequently transmits the surplus energy to the plate. The process with multiple and continuously changing contact state
plug surplus energy is the energy that is in excess of what is to be simulated first and then conduct experiments to ensure cor-
required for the deformation of the plug. Energy plots obtained rectness in overall result. Although it is important to emphasise
from FE analysis provide further understanding of these energy the merit of the FE simulation as a valuable investigative tool
conversion procedures leading to the improvement in the form- for complex problems, its acceptance is enhanced only when
ing process. As shown in Fig. 8, initially the energy transferred to supportive experimental evidence is cited.
the plug manifests mainly as kinetic energy. As time progresses,
a major portion of this kinetic energy is transferred to the plate, Acknowledgement
giving rise to the energy balance shown in Fig. 9. Since the strain
energy of the plate, according to Fig. 7 is much higher than the The authors would like to express their appreciation to Prof.
kinetic energy, it is evident that energy received by the plate is K. Kormi of the Leeds Metropolitan University for his construc-
mainly utilised to deform the blank. tive suggestions and contribution.
The blank shape employed for this analysis is as shown in
Fig. 1. This is obtained from square plates with corners machined
References
circular. The reason for this is, when the corners exist, the blank
tend to rupture due to excessive resistance experienced by the [1] Y.Q. Guo, J.L. Batoz, H. Naceur, S. Bouabdallah, F. Mercier, O. Barlet,
material at the corners. Experimental work also provides evi- Recent developments on the analysis and optimum design of sheet metal
dence in this regard [10]. forming parts using a simplified inverse approach, Comput. Struct. 78
The logarithmic hoop strain variation ln(r/r0 ) plotted in (2000) 133–148.
[2] Z. Zimniak, Implementation of the forming limit stress diagram in FEM
Fig. 10 consists of a positive strain closer to the centre of the cup
simulations, J. Mater. Process. Technol. 106 (2000) 261–266.
and a negative strain closer to the edge of the cup. This shows [3] S.A. Meguid, M.H. Refaat, Finite element analysis of the deep drawing
the direction of movement of the plate material. When the cup is process using variational inequalities, Finite Elem. Anal. Des. 28 (1997)
drawn, material at the centre thins out, as evident from ln(t/t0 ), 51–67.
 V.N. Wijayathunga, D.C. Webb / Journal of Materials Processing Technology 172 (2006) 139–145 145

[4] L.F. Menezes, C. Teodosiu, Three dimensional numerical simulation of [8] F.M. He, T. Zheng, W. Ning, Z.Y. Hu, Explosive forming of
the deep drawing process using solid finite elements, J. Mater. Process. thin-walled semi-spherical parts, Mater. Lett. 45 (2) (2000) 133–
Technol. 97 (2000) 100–106. 137.
[5] T. Naka, F. Yoshida, Deep drawability of type 5083 aluminium– [9] A.A. Ezra, Principles and Practice of Explosive Metal Working, Indus-
magnesium alloy sheet under various conditions of temperature and trial Newspapers, London, 1973.
forming speed, J. Mater. Process. Technol. 89–90 (1999) 19–23. [10] K. Kormi, Studies in the deep drawing of cups using an underwater
[6] D.J. Mynors, B. Zang, Applications and capabilities of explosive form- explosive technique, Ph.D. Thesis, UMIST, 1964.
ing, J. Mater. Process. Technol. 125 (2002) 1–25. [11] Hibbit, Karlsson & Sorenson, Inc., ABAQUS Theory Manual Ver.6.2,
[7] W. Johnson, Literature on Explosive forming, CME-Chartered Mech. 2001.
Eng. 34 (12) (1987) 11.