You are on page 1of 18

Journal of Materials Processing Technology 212 (2012) 15731590

Contents lists available at SciVerse ScienceDirect

Journal of Materials Processing Technology


journal homepage: www.elsevier.com/locate/jmatprotec

Mechanics of fracture in single point incremental forming


Rajiv Malhotra, Liang Xue, Ted Belytschko, Jian Cao
Department of Mechanical Engineering, Northwestern University, Evanston, IL, USA

a r t i c l e

i n f o

Article history:
Received 8 September 2011
Received in revised form 1 February 2012
Accepted 29 February 2012
Available online 12 March 2012
Keywords:
SPIF
Fracture
Material instability

a b s t r a c t
Single point incremental forming (SPIF) is a sheet metal forming technique which has gained considerable
interest in the research community due to its enhanced formability, greater process exibility and reduced
forming forces. However, a signicant impediment in the industrial adoption of this process is the accurate
prediction of fracture during the forming process. This work uses a recently developed fracture model
combined with nite element analyses to predict the occurrence of fracture in SPIF of two shapes, a cone
and a funnel. Experiments are performed to validate predictions from FEA in terms of forming forces,
thinning and fracture depths. In addition to showing excellent predictions, the primary deformation
mechanism in SPIF is compared to that in conventional forming process with a larger geometry-specic
punch, using the deformation history obtained from FEA. It is found that both through-the-thickness shear
and local bending of the sheet around the tool play a role in fracture in the SPIF process. Additionally, it is
shown that in-spite of higher shear in SPIF, which should have a retarding effect on damage accumulation,
high local bending of the sheet around the SPIF tool causes greater damage accumulation in SPIF than
in conventional forming. Analysis of material instability shows that the higher rate of damage causes
earlier growth of material instability in SPIF. A new theory, named the noodle theory, is proposed to
show that the local nature of deformation is primarily responsible for increased formability observed in
SPIF, in-spite of greater damage accumulation as compared to conventional forming.
2012 Elsevier B.V. All rights reserved.

1. Introduction
In single point incremental forming (SPIF) a peripherally
clamped sheet is imparted a desired shape by moving a single hemispherical ended tool along a desired prole so as to locally deform
the sheet along this path (Fig. 1). The sum total of these local deformations gives the sheet its nal shape. Signicant advantages of this
process over conventional forming include greater formability, low
forming forces and generic tooling conguration. One of the major
research problems of considerable interest to the sheet metal forming community is the accurate prediction of fracture in SPIF. This
is important because an underestimation of the fracture depth will
result in a loss of the advantage of enhanced formability of the process and an overestimation will cause component failure during
the forming process itself. Furthermore, a better physical understanding of the mechanisms of deformation and fracture in SPIF
is of great importance since this can aid the choice of appropriate
process parameters for the process and can lead to modications
of the process to further enhance the achievable formability.
Early work in SPIF (Jeswiet et al., 2005) indicated that the maximum formable wall angle could be a good indicator for material

Corresponding author. Tel.: +1 847 467 1032.


E-mail address: jcao@northwestern.edu (J. Cao).
0924-0136/$ see front matter 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2012.02.021

formability in SPIF. More recently, Hussain et al. (2007) formed


axisymmetric funnel shapes in which the proles of the components were arcs of different radii of curvature and showed that
the maximum formable wall angle depended on the radius of the
curvature of the funnel components prole. This indicates that
formability in SPIF depends on a combination of the global shape of
the component and the process parameters, and therefore essentially on the deformation mechanics of the process. Filice et al.
(2006) explored the possibility of detecting fracture in real time
based on the trend of the forming force. Szekeres et al. (2007)
showed that while the force trend methodology for detecting failure can be used for a cone shape, it does not work for a pyramid
shape. This observation again highlights the fact that occurrence
of failure in SPIF depends signicantly on the process mechanics
itself.
Attempts have also been made to use the concept of conventional forming limit curves (FLCs) to characterize formability limits
in SPIF. Filice et al. (2002) demonstrated that the failure strain in
SPIF signicantly exceeds that in conventional forming. Hussain
et al. (2009) derived empirical forming limit diagrams (FLDs) which
used the reduction in cross sectional area at tensile failure as a
means of predicting failure in SPIF. However, they stated that their
empirically derived FLD depended on the process mechanics itself
and was only valid within the range of the tool diameters, incremental depths and feed rates used in their work. At the same time

1574

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

Fig. 1. Schematic of SPIF (dotted lines show motion of tool in the prole view).

Emmens and van den Boogaard (2009) showed that FLCs have certain drawbacks when it comes to predicting failure in SPIF. It is
known that FLCs are not valid when there is bending and throughthe-thickness shear, both of which are signicant in SPIF. As a result
modications of conventional FLCs by incorporating the effects of
changing strain paths (Yao and Cao, 2002) and the effects of large
normal contact pressures (Smith et al., 2003) would still not be
able to predict failure in SPIF accurately. Therefore, the prediction
of formability in SPIF with conventional FLCs might not be feasible.
Numerical investigations using FEA have also been conducted to
investigate the deformation force and mechanisms in SPIF. Henrard
et al. (2007) modeled the contact between the tool and the sheet
using a moving spherical tool method, in which a dynamic explicit
time integration scheme was used instead of the usual penalty
based contact algorithm. The main improvement was that a better force prediction was obtained using their methodology even
though computational time was reduced by using a larger element size. Cerro et al. (2006) simulated SPIF of a pyramid with a
75 wall angle with shell elements and obtained a 5% difference
between the maximum values of the measured and calculated tool
z forces. However, no attempt was made to predict fracture. van
Bael et al. (2007) extended a Marciniak-Kuczyisnki analysis (MK analysis) to predict localized necking and fracture in SPIF. They
showed that while the forming limit predictions were higher than
that for monotonic loading, their models still underestimated the
forming limits obtained experimentally in SPIF. This was attributed
to the fact that the input for the M-K model was obtained at a
pre-determined location through the thickness of the sheet which
meant that interaction between different layers of the sheet was
not considered. Huang et al. (2008) used Oyanes fracture criterion,
an empirical fracture criterion, to predict failure during forming of
a conical cup using SPIF. The model was found to capture forming
limits in SPIF reasonably well, however, the predictions of forming
forces were not satisfactory. Silva et al. (2009) extended a membrane analysis of SPIF to incorporate a damage model in which
damage accumulation depended on hydrostatic stress and showed
that such an approach could be used to predict fracture strains in
SPIF. Malhotra et al. (2010a) investigated the use of various material
models and element types to simulate SPIF using FEA and showed
that a fracture model considering triaxility and shear as presented
in Xue (2007a,b) can predict forming forces and fracture occurrence much better than other common material models. The results
were promising and have led to the further investigations on the
mechanics of fracture in SPIF, which will be presented in this paper.
While predicting forming force and failure limits in SPIF is
important, a more interesting challenge is to understand why SPIF
results in a much higher formability compared to the conventional
forming process. Emmens et al. (2009) proposed that while bending, shear, cyclic straining and hydrostatic stress are the dominant
deformation mechanisms in SPIF, pinpointing which factors are

primarily responsible for failure is difcult. Jackson and Allwood


(2009) showed experimentally that deformation in SPIF consists
primarily of stretching perpendicular to the toolpath and throughthe-thickness shear perpendicular to and along the direction of the
toolpath. They also showed that shear increased with the depth of
deformation, was greater on the inner side of the sheet and was
greater along the direction of toolpath motion than in the direction perpendicular to the toolpath motion. They observed that as
the component was formed the structure became stiffer and the
deformation transitioned from a more widely distributed area to
being concentrated into a local area around the tool contact zone.
In addition, they mentioned that unusual choice of material, sheet
thickness and wall angle used in the present experiments probably
caused strains that have some differences to more typical ISF experiments which use thinner sheets and steeper wall angles. Allwood
et al. (2007) also showed experimentally that signicant throughthe-thickness shear is present in SPIF, by tracing the history of a pin
inserted perpendicularly into the blank during deformation of the
blank. A valuable work with reference to the mechanisms responsible for increased formability in SPIF was performed by Allwood and
Shouler (2009) in which the M-K analysis was extended so that all
six components of the stress tensor were non-zero. This represents
the typical state of deformation in SPIF. This work provided signicant circumstantial evidence that through-the-thickness shear
might play a signicant role in fracture in SPIF.
The current work goes beyond the previous discussions focusing
on whether hydrostatic pressure or through-the-thickness shear
contributes to the signicant increase of forming limit in SPIF.
Instead, we uncover the unique role that material localization plays
in SPIF. The approach is to use a fracture model in FEA to analyze
the mechanics of deformation in the SPIF of a 70 wall angle cone
and a funnel shape. The experimental setup is shown in Section
2. The material model is briey described in Section 3, in which
the fracture envelope is expressed in the stress space and is a function of the hydrostatic pressure and the deviatoric stress state (Xue,
2007a). More recently, it has been shown (Xue, 2007b) that only
stress triaxiality effects cannot explain the phenomenon of fracture in shear and the material model used in this work combines the
effects of plastic strain, hydrostatic pressure and shear on fracture.
Therefore, this model is ideal for examining the combined effects
of stretching along the component wall and local bending around
the tool (indicated by the hydrostatic pressure and plastic strain)
and through-the-thickness shear (indicated by the Lode angle).
Note that stretching, bending and shear are among the deformation mechanisms said to very dominant in SPIF (Emmens et al.,
2009). Corresponding experiments are performed to compare the
forming force history and fracture location predictions from FEA.
The mechanics of SPIF are then analyzed in-depth in Section 4 by
examining the deformation history through the thickness of the
sheet at four locations along the formed component wall and the

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

1575

plastic strain (p ) and a reference fracture strain (f ). In particular, we chose the stress-based fracture envelope for its simplicity.
The damage coupled yield function is
= eq w(D) M 0

Fig. 2. Proles of (a) 70 cone (b) funnel formed as part of experiments.

primary deformation mechanisms which affect failure in SPIF are


found. Contributions from hydrostatic pressure and through-thethickness shear to fracture are quantied. In Section 5 additional
simulations are performed with a reduced tool-sheet friction coefcient to examine the effect that through-the-thickness shear has
on formability in SPIF. Furthermore, to better understand the failure
mechanism, simulations are also performed using the same material model for forming of the 70 cone with a conventional forming
process using a larger punch. The deformation mechanisms and
the occurrence of material localization in the larger punch case are
compared to those in the SPIF case in Section 6. Motivated by the
results from these two cases, a new noodle theory is proposed here
to explain the essence of enhanced formability seen in SPIF. Finally,
inferences on the nature of fracture in SPIF are discussed in Section
7 followed by conclusions in Section 8.
2. Experimental setup of forming a cone and a funnel using
SPIF
Two shapes, a 70 wall angle cone and a variable angle funnel
(Fig. 2) were formed using 1 mm thick AL5052 sheet, with a tool
diameter of 9.5 mm and feed rate of 150 mm/min. The blank size
available for forming inside the clamp was 80 mm 80 mm square
for the cone and an 80 mm diameter circular area for the funnel
case. No tool rotation was allowed and PTFE based grease was used
as lubricant at the toolsheet interface. The incremental depths
used for the cone and the funnel were 1.0 mm and 0.5 mm respectively. A spiral toolpath was used in which the tool followed the
three dimensional prole of the shape to be formed while moving
simultaneously in the x, y and z directions (Malhotra et al., 2010b).
The z forces on the tool throughout the forming process were
measured using a Kistler dynamometer model 9255A mounted
below the xture and the tool tip depth at which fracture occurred
was also recorded. Each experiment was performed three times and
the nal fracture depth for a component was taken as the average
of the observed fracture depths from the three experiments.

(1)

where  eq is the equivalent stress,  M is the undamaged matrix


stress which is a function of the plastic strain to include strain hardening, and w(D) is a weakening factor used to describe the material
deterioration. The weakening factor w is related nonlinearly to the
damage variable D by w(D) = 1 D , where is a material constant.
The weakening factor w is treated as a scalar for isotropic damage models like the present one. To model matrix resistance the
Swift-type hardening relationship is adopted as show in Equation
(2)

M = y0 1 +

p
0

n
(2)

where  y0 is the initial yield stress, 0 is the pre-strain, p is the


plastic strain and n is the hardening exponent. The associated ow
rule is enforced. Accumulation of damage is modeled as a nonlinear
function of the plastic strain (p ) and a reference fracture strain (f )
and is expressed in the rate form as

D=m

p
f

m1

p
f

(3)

where m is a material constant. In the nite element model an element completely loses its load carrying capacity and is removed
when the value of the damage variable D reaches 1.0. When all the
elements through the thickness of the sheet are removed then a
crack is said to have occurred at that location. The reference fracture strain (f ) is rst expressed as a stress envelope  Mf , which
takes the form of a modied Tresca type of pyramid (Xue, 2009) as
follows,



3
Mf = f 0 1 + kp p
(4)
2 cos L
where  f0 is a reference fracture stress, kp is a material constant
related to pressure sensitivity, p = ( kk /3) is the hydrostatic pressure and  L is the Lode angle, via which the deviatoric component
of the stress state is incorporated into the reference fracture strain.
An example of this reference fracture strain in the space of the principal strains (1 , 2 and 3 ) and the hydrostatic stress p is shown in
Fig. 3.

3. Fracture model and its implementation in FEA


This section describes the material model used and its implementation in FEA to predict fracture in incremental forming. The
predictions from FEA are compared to measurements obtained
from experiments.
3.1. Fracture model
The present work uses the damage plasticity model proposed
by Xue (2007a) in which two independent internal variables, i.e.,
the so-called damage variable and the plastic strain, are used to
describe the material status. The constitutive relationship consists
of a damage coupled yield function, the evolution laws for the

Fig. 3. Illustrative example of the reference fracture strain envelope used in this
work.

1576

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

The reference fracture strain f used in Equation (3) is obtained


from the inverse of the matrix stress as:
 (1/n)

(1/n)



f 0
3
f (p, L ) = 0
1 + kp p
1
(5)
y0
2 cos L
Therefore, damage accumulation in this material model
depends on the plastic strain (p ) as well as on the current state of
hydrostatic pressure and deviatoric stress state, via the reference
fracture strain (f ). Xue (2010) showed that this type of I1-J2-J3
constitutive model can predict both the onset of ductile fracture
and material instability at the same time. Note that in this material
model there are four material constants which cannot be obtained
from tensile tests and therefore, a calibration procedure with different mean stress and deviatoric stress states should be employed.
These parameters are  f0 , kp which quantify the reference fracture
strain (Equation (4)), and , m which are related to how the damage
and material softening evolves (Equations (1) and (3) respectively).
The material constant  f0 is the matrix stress at fracture under zero
mean stress condition, i.e., tensile condition. The parameter kp is
responsible for the contribution of hydrostatic pressure towards
the evolution of f . The material constant m signies the rate at
which damage accumulates in an element and therefore how fast
the element is removed. The inclusion of a nonlinear damage evolution law takes into consideration the common observation that
damage accumulation rate increases as the plastic strain increases.
The value of dictates how the material loses its capacity to carry
load as damage accumulates. Note that since m inuences the evolution of the damage (Equation (3)) and inuences how this
damage affects the weakening of the material (Equation (1)), both
these material constants are together responsible for the softening
behavior of the material as damage develops.
In addition to the prediction of damage accumulation and
fracture, the occurrence of diffused and localized necking during
deformation is predicted analytically using this material model. Diffused necking is predicted using a three dimensional generalization
of Consideres maximum force criterion to a maximum power criterion, as derived in Xue (2010). The closed form expression used
to predict the onset of diffused necking is shown in Equation (6),

h
wD D

+
cos L +
M
w
6

(6)

where wD = w/D, D = D/p , h =  eq /p and  L = Lode angle.


The equal sign in Equation (6) denotes the onset of diffused necking. The effect of damage accumulation and weakening on material
instability is included in this expression via the term (wD D /w).
Note that higher rates of weakening or damage accelerate the onset
of diffused necking. The occurrence of localized necking, i.e., of local
shear bands post diffused necking, is also predicted in a HadamardHill sense by examining the positive deniteness of the acoustic
tensor at the current stress state, as shown in Xue and Belytschko
(2010). The form of the acoustic tensor A is as follows
A = h : Cedp : h

(7)

where Cedp is the elastic-damage-plasticity tangent matrix and h is


the direction in which instability might develop. The tangent matrix
Cedp is expressed as
Cedp = wC0 +

 2G   w D
D
0
w

h + 3G0

2wG0 r r

(8)

where r is the deviatoric stress direction, = 1/3 kk I + 2/3 eq r, G0


is the shear modulus of the material, C0 is the isotropic Hookean
matrix for the undamaged material. C0 is expressed as shown in
Equation (9).

C0 = 2G0 II + K0

2
G0 I I
3

(9)

Table 1
FEM model details.
Simulation case

Tool speed
(mm/s)

Target time
Simulation
increment for
time (CPU h)
mass scaling (s)
1.5e 06
1.2e 06
1.5e 06

SPIF: cone
3000
SPIF: funnel
1500
Conventional forming of cone 3000

78
115
13

Table 2
Material properties calibrated from tensile tests.
 (kg/m3 )

E (GPa)

 y0 (MPa)

2680

68.6

0.3

117

0.22

0.0045

where K0 is the bulk modulus of the material, I and II are the secondorder and fourth order identity tensors respectively. Equation (10)
shows the condition in which the material is stable in a HadamardHill sense, where g is a vector denoting the direction of particle
velocity. The material has lost stability at a given state when, for
some direction h and velocity g, the left hand side of Equation (10)
is lesser than zero.
g:A:g0

(10)

3.2. Implementation of the material-model in FEA


The above described material model was implemented using a
user subroutine in LS-DYNA. A schematic of the FEA model used
for simulating SPIF is shown in Fig. 4a. The blank material was
discretized using eight reduced integration linear brick elements
through the thickness of the sheet. The blank was meshed so that
the region to be formed had a radial mesh with a maximum inplane element size of 0.50 mm in the radial direction and 0.35 mm
in the circumferential direction (Fig. 4b). The tools and the top and
bottom clamps were discretized using planar shell elements with
an element size of 0.20 mm for the tool and 2.5 mm for the clamps.
Furthermore, the tool speed was articially increased and mass
scaling was used to speed up the simulation. The friction coefcient at the toolsheet interface was specied as 0.15 (Eyckens et al.,
2010). The Belytschko-Tsay hourglass formulation was used to control hourglassing. All simulations were performed on a workstation
with four processors at 3.66 GHz speed. Details on tool speed, mass
scaling and simulation time are shown in Table 1.
The material parameters E,  y0 , n and 0 were obtained from
uniaxial tensile tests and the values are shown in Table 2. The
four material constants  f0 , kp , m and were calibrated manually by matching the tool z forces from simulation with the same
obtained from experiments for the case of the 70 cone (Table 3).
The methodology used to nd the values for these materials constant is as follows.
The values of kp and  f0 control the initial slope of the force
curve. A higher value of kp and  f0 results in a greater slope and vice
versa. The values of the parameters m and are usually between 2.0
and 3.0 for most metals (Xue, 2007a,b). If the value of m is too high
then the occurrence of fracture is delayed and the softening part of
the force curve is higher than that from experiments. In this case,
the value of m is decreased to increase the rate of damage accumulation so that fracture occurs at the correct depth. After this, if the
Table 3
Reference fracture strain parameters calibrated using trial simulations with 70
cone.
 f0 (MPa)

kp (MPa1 )

490

0.0001

1.80

2.5

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

1577

Fig. 4. (a) Schematic of the FEA model used to simulate SPIF. (b) Top view of the mesh used to discretize the blank.

initial slope of the force curve and the fracture depth are found to be
matching well but excessive weakening is observed then the value
of needs to be increased. In this work three trial simulation runs
were performed to calibrate these material parameters (Table 3).
To show the quality of the damage model calibration the forming
forces in the z direction obtained from experiments and simulation
for the 70 cone are compared in Fig. 5a.
3.3. Verication of the material calibration results
The calibrated FEA predicted that the cone fractured at a tool
tip depth of 16.1 mm in experiments as compared to a tool tip
depth of 14.8 mm predicted by FEA. Furthermore, the maximum
thinning just before fracture was 64% from experiment and 63%
from simulation. The same model was then used to predict forming force and fracture depth for the funnel case (Fig. 2b). In this case,
the experimentally measured fracture depth from experiments was
15.2 mm and that predicted by FEA was 14.5 mm. The tool z forces
from experiments and simulation for the funnel are compared in
Fig. 5b. The thinning before fracture was measured to be 65.8% from
experiments and 63.64% from simulation.
The forces on the tool relate to the state of stress in the material, maximum thinning relates to the strain experienced by the
material and the fracture height relates to how well the damage
evolution function (Equation (3)) incorporates the physical effects
that govern fracture in SPIF. It can be seen that the tool z force prediction, the maximum thinning and the fracture depth prediction
from FEA all agree quite well with those from experiments.

3.4. FEM model of the conventional deep drawing process


To analyze the difference between SPIF and the conventional
deep drawing process, the forming of the 70 cone with a larger
punch instead of with a SPIF tool was also simulated. A schematic
of the FEA model used for this simulation is shown in Fig. 6a. The
corner radius of the punch was the same as the SPIF tool radius, i.e.,
4.75 mm, and the material properties and mesh size of the blank
as well as contact properties were the same as those in SPIF. In the
SPIF simulation, the outer periphery of the blank was constrained
so that the blank was completely xed in the xy plane, i.e., there
was no material draw in. The top and bottom clamps were used
to constrain motion of the unformed region of the blank in the z
direction. For the conventional forming simulation, the boundary
of the blank was not constrained, i.e., draw in was allowed. The
blank holder and the die were only used to prevent motion of the
unformed region in the z direction. The punch was displacement
controlled such that if the SPIF tool tip was at a depth z at time
t then the at face of the punch was also at the same depth z at
time t. The coefcient of friction between the xture and the blank
was specied as 0.15, the same as in the case of SPIF. The punch
movement in the negative z direction was set such that the at face
of the punch was at the same z depth as the tip of the SPIF tool at
any point during the simulation. The predicted fracture depth for
this larger punch forming case was 13.5 mm, i.e., the formed depth
in SPIF was greater than that in the punch forming case. The plastic
strain just before fracture was 1.43 for the conventional forming
case, as compared to a strain of 1.83 for SPIF. Also in the punch

Fig. 5. Comparison of tool z forces between FEA and experiments for (a) 70 cone (b) funnel.

1578

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

Fig. 6. (a) Schematic of the FEA model (b) occurrence of crack predicted by FEA, for the punch forming case.

forming case fracture occurred as a continuous crack all along the


circumference of the component (Fig. 6b).
A comparison of the hourglass control energy, kinetic energy
and internal energy of deformation for SPIF and conventional forming simulations showed that the kinetic energy and hourglass
control energy were less than 4% of the internal energy of deformation. Therefore, the effects of hourglass control, mass scaling
and articially speeding up the tool on the simulation results was
negligible.

4. Deformation analysis of incremental forming


The predictions in terms of forming force, fracture depth and
thickness reduction obtained from the FEM model embedded with
the fracture model provide a good foundation for further analyzing the fundamental deformation mechanics in SPIF. The question
to be answered is, which factors contribute the most to the occurrence of fracture in SPIF? This section analyses the key indicators
that can be obtained from the FEA to answer this question. These
indicators are: the damage variable (D) in Section 4.1, plastic strain
(p ) in Section 4.2, hydrostatic pressure (p) in Section 4.3, throughthe-thickness shear (13 , 23 ) in Section 4.4 and fracture strain (f )
in Section 4.5. The combined effect of these indicators is analyzed
in Section 4.6, to pinpoint the primary deformation mechanisms
affecting fracture in SPIF. All the analysis starts with an examination of the deformation history at four sections A, B, C and D along
the prole of the deformed shapes in the simulation, section D being
closest to the axis of symmetry of the formed component (Fig. 7a).
At each of these sections the four elements through the thickness
of the sheet are labeled 1 to 8, where 1 is the element on the
inner side of the sheet which is in contact with the tool and 8 is
the element on the outer side of the sheet (Fig. 7b). This work concentrates on examining the deformation of elements 1 and 8 at each
of the four sections A, B, C and D. The key deformation indicators
mentioned above are plotted versus the z displacement of the tool
tip.

It can be seen that in both the cone and the funnel cases, the damage variable evolves faster for element 8, i.e., on the outer side of the
sheet, than for element 1 which is on the inner side of the sheet. As a
result, in FEA the element on the outer side of the sheet is removed
rst. This phenomenon was also conrmed by a visual examination of the FEA results. Physically this implies that in SPIF the crack
initiates on the outer side of the sheet and propagates inwards.
Note that in Equation (3) the damage variable D is directly proportional to plastic strain p and inversely proportional to the reference
fracture strain f . Therefore, to investigate the reasons behind the
trends for damage evolution shown in Fig. 8, the evolution of p and
f at sections A, B, C, D shall be examined next.
4.2. Evolution of plastic strain (p )
Fig. 9 shows the evolution of plastic strain (p ) for elements 1
and 4 at sections A, B, C and D for the cone and the funnel cases.
For both the cone and the funnel cases, the plastic strain at any
section is greater for the outer side of the sheet (element 8) than
it is for the inner side of the sheet (element 1). This is because of
local stretching and bending of the sheet around the tool which
causes the element on the outer side of the sheet (element 8) to
stretch more as compared to the element on the inner side (element 1). This results in a higher plastic strain on the outer side of
the sheet. Since, damage accumulation is directly proportional to
plastic strain (Equation (3)), a straightforward conclusion is that by
itself, local bending of the sheet around the tool will attempt to increase
damage accumulation on the outer side of the sheet as compared to
the inner side of the sheet.
However, attributing the damage behavior observed in Fig. 8
entirely to this phenomenon would be premature. As shown in
Equation (3) it is not only the plastic strain p but also the reference fracture strain f that plays a role in determining damage
accumulation. The following sections will explore this facet further
by examining in greater detail the factors that affect f , i.e., hydrostatic pressure in Section 4.3 and through-the-thickness shear in
Section 4.4.

4.1. Damage evolution indicated by damage index (D)

4.3. Hydrostatic pressure (p)

The damage plasticity model uses a damage variable D to signify the accumulation of damage and loss of the materials ability
to take stresses during deformation. An element is removed when
the damage variable D reaches a value of 1.0 and a crack is said to
occur at a location where all elements through the thickness of the
sheet have been removed. Since we are interested in the occurrence
of fracture therefore, in this sub-section, the evolution of damage
variable, D, in SPIF will be examined rst. Fig. 8a and b show the
evolutions of the damage variable D from section A to section D for
the cone and the funnel, respectively.

The evolution of hydrostatic pressure (p) in forming the cone


and the funnel is shown in Fig. 10. The hydrostatic pressure, p, is
positive on the outer side of the sheet (i.e., element 8) and negative
on the inner side of the sheet (i.e., element 1). This is because once
the tool has passed over a certain region of the sheet, that region of
the sheet undergoes local springback. This results in the hydrostatic
pressure on the outer side becoming positive and on the inner side
becoming negative. This effect is seen more clearly by examining
the hydrostatic pressure contours in a certain region of the blank at
consecutive time steps during the simulation, as shown in Fig. 11. In

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

1579

Fig. 7. (a) Schematic of sections A, B, C and D along the component wall at which the deformation history from FEA is examined. (b) Nomenclature of elements through the
thickness of the sheet (contours of damage variable D shown).

Fig. 8. Evolution of damage variable D at sections A, B, C, and D for the (a) 70 cone (b) funnel shape.

Fig. 9. Evolution of plastic strain p at sections A, B, C, and D for the (a) 70 cone (b) funnel shape.

1580

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

Fig. 10. Evolution of hydrostatic pressure at sections A, B, C, and D for the (a) 70 cone, (b) funnel shape.

the contours shown in Fig. 11a the highlighted region of the blank is
in contact with the tool. Fig. 11b shows the same region of the blank
after the tool has passed over this region and local springback has
occurred. It can be seen that the local springback results in negative
hydrostatic pressure on the inner side of the sheet and positive
hydrostatic pressure on the outer side of the sheet.
In terms of the effect of hydrostatic pressure (p) on f (Equation (5)), the trends shown in Fig. 11 mean that the pressure term

(1 + kp p)1/n in Equation (5) will be greater than 1.0 for element 8


and lesser than 1.0 for element 1.
4.4. Through-the-thickness shear (zx , zy )
Since the Lode angle  L , in Equation (5) denotes the shear component of the stress it is also worthwhile to look at the shear stresses
in SPIF. Along the direction of the tool motion (i.e., along the hoop

Fig. 11. Contours of hydrostatic pressure on the deformed blank at simulation time of (a) 2.3885 s (b) 2.3942 s.

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

1581

Fig. 12. Evolution of through-the-thickness shear along the tool motion direction (zx ) at sections A, B, C, D for the (a) 70 cone (b) funnel shape.

direction of the component) the tool drags the material along with
it, causing a through-the-thickness shear zx . Fig. 12 shows that zx
increases from sections A to D and is higher on the inner side of the
sheet, i.e., for element 1 than for element 8.
Since a spiral toolpath is being used, the tool continuously moves
down in the z direction while moving in the x and y directions and
therefore drags material in a direction perpendicular to the toolpath as well. Fig. 13 shows the evolution of through-the-thickness
shear in a direction perpendicular to the tool motion, i.e., zy . Again,
shear is greater on the inner side of the sheet as compared to
the outer side of the sheet. Note that there are three important
trends associated with shear, i.e., (1) greater shear on the inner
side of the sheet, (2) increase in shear along the toolpath motion
with deformation depth, and (3) shear along the toolpath direction
being greater than shear in a direction perpendicular to the toolpath motion. All these trends have been shown in the past work of
Jackson and Allwood (2009). The effects of the observed trends of
hydrostatic pressure and through-the-thickness shear on f will be
discussed below.
4.5. Reference fracture strain (f )
In the present material model f depends on the hydrostatic
pressure p and the on the shear, via the Lode angle  L as shown

in Equation (5). The product term [(1 + kp p) ( 3/2 cos  L )](1/n) , in


Equation (5), signies the combined effect of the hydrostatic pressure and the Lode angle of the current stress state on f . To examine
the individual effects of p and  L , this product term is split up into a

pressure term (1 + kp p)(1/n) and a Lode angle term ( 3/2 cos  L )(1/n) .
The evolution of these factors with plastic strain is individually plotted (Figs. 14 and 15 respectively), for elements 1 and 4, at section D,
i.e., where the through-the-thickness crack rst begins. Note here
that a higher value of the product term implies a higher f which in
turn means that damage accumulation is retarded, and vice versa.
It can be observed that for both the cone and the funnel the
pressure term is lesser than 1.0 for element 1 and greater than 1.0

for element 8. This is to be expected since the hydrostatic pressure


is negative for element 1 and positive for element 8 in both cases
(Fig. 10). Therefore, if the Lode angle terms for elements 1 and 8
are comparable, the product term for element 8 should be higher
than that for element 1. This should in turn cause an increase in the
corresponding f for element 8. However, Figs. 14 and 15 show that
for both the cone and the funnel the modication term is actually
lower on the outer side of the sheet (element 8) than it is on the side
in contact with the tool (element 1). This is because in-spite of the
increasing pressure term the Lode angle term is so dominant that it
reduces the modication term more signicantly on the outer side
of the sheet.
Since the Lode angle is representative of the deviatoric component of the stress, this difference in Lode angle terms can
be attributed to the difference in through-the-thickness shears
between the outer and inner sides of the sheet (Figs. 12 and 13).
Therefore, it is the through-the-thickness shear and not the hydrostatic pressure which dominates evolution of the reference fracture
strain f in SPIF. Furthermore, since f is inversely proportional to the
damage variable D, taken by itself, higher f on the inner side of the
sheet will cause retardation in damage accumulation on the inner side
of the sheet as compared to the outer side.
Therefore it can be said that higher shear on the inner side of the
sheet is a deformation mechanism which will try to reduce damage accumulation on the inner side of the sheet as compared to
the outer side. This also correlates well to past work by (Allwood
et al., 2007) which proposed that greater shear can enhance
formability.

4.6. Combined effect of local bending and through-the-thickness


shear
The analysis performed till now has shown two deformation
mechanisms, which drive damage accumulation in SPIF. These
mechanisms are as follows:

1582

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

Fig. 13. Evolution of through-the-thickness shear perpendicular to the tool motion direction (zy ) at sections A, B, C, and D for the (a) 70 cone (b) funnel shape.

Fig. 14. Comparison of modication factor, pressure factor and Lode angle terms for 70 cone.

Fig. 15. Comparison of modication factor, pressure factor and Lode angle terms for the funnel shape.

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

1. Local bending of the sheet around the tool, which causes greater
p on the outer side of the sheet. The effect is to increase damage
accumulation on the outer side of the sheet.
2. Through-the-thickness shear, which is lower on the outer side
of the sheet and causes f to be lower on the outer side of the
sheet. The effect is to further increase damage accumulation on
the outer side of the sheet than on the inner side of the sheet.
The cumulative effect of these phenomena is to increase damage
accumulation on the outer side of the sheet causing the crack to
begin on the outer side of the sheet. Therefore, it is a combination of
local bending and through-the-thickness shear that governs evolution
of damage, failure occurrence and the crack initiation location in SPIF.
5. Noodle theory for fracture in SPIF
As mentioned in Section 3, the forming of the 70 cone with a
larger punch was also simulated and the predicted fracture depth
was about 2 mm lesser than that of the same cone simulated using
SPIF. This section attempts to uncover the reasons behind this
increased formability in SPIF by comparing the deformation history
in SPIF to that in the punch forming case in Section 5.1 followed by
a description of the proposed noodle theory in Section 5.3.
5.1. Deformation mechanisms in punch forming case
Fig. 16a shows that for the larger punch forming case damage
accumulation for both elements 1 and 4 is the same. Fig. 16b shows
that the plastic strain on either side of the sheet is similar as well.
Examining the shear strains at sections A, B, C and D (Fig. 17)
shows that the shear is several orders of magnitude smaller than in
SPIF, indicating that the dominant deformation mechanism in the
larger punch forming case is tension along the component wall.
Observe that in Fig. 16a, i.e., in the punch forming case, the plastic
strain and damage are concentrated at section D. Also there is a
sudden increase in the plastic strain and damage rate at section
D at a z displacement of around 11 mm. In comparison, in SPIF
(Figs. 8a and 9a) the plastic strain and the damage are much higher
very early on during the deformation, accumulate more uniformly
along sections A, B, C and D and at no point there is a sudden increase
in the plastic strain rate or damage rate. Physically this means that
in SPIF plastic strain and damage are spread out more evenly along
the formed component wall as compared to the punch forming case.
5.2. Impact of shear on increased formability in SPIF
The objective of this sub-section is to compare fracture in SPIF
and in conventional forming, and to answer the question, Does
greater shear in SPIF as compared to conventional forming completely explain the higher formability observed in SPIF?
Fig. 17 shows that shear in conventional forming is negligible as
compared to shear in SPIF (Figs. 12 and 13). Past work has noted
this fact (Jackson and Allwood, 2009) and it has also been proposed (Allwood et al., 2007; Allwood and Shouler, 2009) that this
increased shear in SPIF might be the reason for increased formability in SPIF as compared to conventional forming. Plotting the
components of the fracture envelope for the punch case and the
SPIF case (Fig. 18) shows that f is indeed much lower for the punch
case as compared to SPIF, especially on the inner side of the sheet.
Since the Lode factor dominates the value of f in SPIF, and the Lode
angle is indicative of shear, higher shear is the primary reason for
higher f in SPIF. If shear were the only factor increasing formability
in SPIF and since f is inversely proportional to damage (Equation
(3)), the result should have been lower damage accumulation in
SPIF as compared to conventional forming.

1583

However, Figs. 8a and 16a show that in comparison to conventional forming, damage accumulation is much higher in SPIF.
Comparing the ratio of plastic strain to fracture strain at sections A,
B, C, D for SPIF and the punch forming case (Fig. 19) shows that the
ratio (p /f ) is much higher in SPIF than in conventional forming.
Therefore, in-spite of a higher value of f , the value of p is so
much higher in SPIF that it overwhelms the effect of higher f .
Essentially, while greater shear in SPIF does attempt to retard damage accumulation as compared to conventional forming, greater
local bending in SPIF overwhelms this effect and causes greater
damage accumulation. This should result in the formability in
SPIF being reduced as compared to conventional forming, which
does not happen. Therefore, increased shear only partially explains
increased formability in SPIF as compared to conventional forming.
The obvious question that arises is, if damage accumulation is
greater in SPIF then what is the reason for increased formability
in SPIF? The answer is obtained by comparing material instability
between SPIF and conventional forming, which will be discussed
next.
5.3. Increased formability in SPIF: the noodle theory
As described in Section 3, the material model used in this
work also incorporates material localization in the form of diffused and localized necking. Examining these predictions provides
an explanation for the increased formability in SPIF as compared to
conventional forming. Note that here material localization refers to
the onset of material instability. Jackson and Allwood (2009) mentioned localization of deformation from a more widely distributed
area to a local area around the tool contact zone as the formed
structure became more rigid. However, by localization they meant a
geometric concentration of deformation into the local contact area
around the tool. The geometric concentration of deformation into
a local area around the tool contact may or may not cause material
instability.
Figs. 20 and 21 show the localization contours, z depths and
plastic strains for the punch forming case and SPIF, respectively.
In the localization ag plots, the localization ag has a value of 1.0
(blue color) when the material is in a diffused necking state and a
value of 3.0 (red color) when the material has reached a localized
necking state (i.e., local shear bands have formed). The contours
shown in (a) are at the onset of diffused necking, in (b) are at the
onset of localized necking and in (c) are just before fracture occurs.
Fig. 20a and b show that in the punch forming case diffused
necking begins at a z depth of 8.0 mm and localized necking begins
at about 11.0 mm. Once localized necking starts, the plastic strain
becomes highly concentrated into this unstable region of the material, till fracture occurs in this region (Fig. 20c) leading to fracture
within about 2.6 mm (i.e., at a z depth of 13.6 mm). This phenomenon of localized necking being followed rapidly by fracture
is a very well documented phenomenon in conventional forming
processes. In fact the prevention of localized necking has been a
well known way of preventing failure in conventional forming processes.
In contrast, in SPIF diffused necking begins earlier at a z depth
of around 5.4 mm and localized necking begins at around 8.4 mm
(Fig. 21a and b) respectively. This earlier onset of diffused necking
can be attributed to the inherently local deformation in SPIF. Local
deformation in SPIF causes greater plastic strains to be induced
in the material very early on as compared to the punch forming
case, evident in sections AC in Figs. 9a and 16b. This increases the
damage rate (Figs. 8a and 16a) and weakening rate very quickly,
early on during the deformation in SPIF. Since the onset of diffused
necking is accelerated by greater weakening rate and damage
rate (Equation (6)), and early on during the deformation both of
these are higher in SPIF than in the punch forming case, therefore,

1584

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

Fig. 16. Evolution of (a) damage variable: D (b) plastic strain: p , at sections A, B, C and D for the punch forming case.

the onset of diffused necking occurs earlier in SPIF than that in


the conventional forming case. Consequently, the occurrence of
localized necking is also earlier than in the punch forming case.
If material localization is used as an indicator of fracture then one
would expect that since localized necking begins earlier in SPIF
the fracture depth in SPIF would also be lower. However, this is
not the case. The reasons for this will now be discussed.
A very signicant difference between SPIF and the punch forming cases arises after localized necking begins. The z depth at
fracture in SPIF is 14.81 mm, i.e., about 6.4 mm of the already formed
material is in a state of localized necking before fracture nally
occurs (Fig. 21c). This is more than twice that in the punch forming case. This is because, in SPIF, after the tool deforms a certain
region and causes localized necking it moves on and deforms new
material. As a result the amount of deformation experienced by the

previously unstable region is lesser than it would be with a more


global deformation as in the punch case. So the shear bands do
not grow as fast as expected. Therefore, local deformation in SPIF
is responsible for the existence of a large localized necking region
before fracture nally occurs. The effect of the previously formed
material in SPIF undergoing localized necking without going all the
way to fracture is that this previously formed region is able to take
up some of the deformation caused in subsequent passes of the tool.
As a result the component can be formed to a greater z depth and
a greater plastic strain without fracture using SPIF as compared
to the punch forming case. This effect is also seen by examining
the plastic strain contours for both cases which show that in the
punch case (Fig. 20c) the plastic strain becomes concentrated very
quickly into the shear bands in the localized necking region before
fracture. However, in SPIF (Fig. 21c) the plastic strain is distributed

Fig. 17. Evolution of through-the-thickness shear along (a) hoop direction (zx ) (b) component wall (zy ), at sections A, B, C and D for the punch forming case.

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

1585

Fig. 18. Comparison of modication factor, pressure factor and Lode angle terms for 70 cone formed with (a) SPIF (b) large punch.

over a much larger localized necking region and has greater magnitude before fracture, as compared to the punch case. Therefore,
the component wall has a more uniform strain distribution in SPIF.
This phenomenon can be better understood by a simple analogy,
i.e., the so-called noodle theory which is as follows. Consider a
single string of wheat noodles that is held at one end and then
needs to be stretched as much as possible.
One obvious strategy is to start pulling at the free end of the
string (Fig. 22a). As a result, at some location on the string the
material will begin to localize, as shown in red vertical stripes in
Fig. 22b, a strain concentration will develop and eventually fracture
will occur (Fig. 22c). This is similar to what happens in conventional
forming. An alternate strategy would be to stretch by smaller increments all along the string. One would start at a point a little bit away
from the xed end of the string (section AA in Fig. 22d) and stretch
by a small increment (say s), while moving the location at which
the deformation is applied by small regular increments (say c)
towards the free end of the string (Fig. 22e), i.e., from section AA to

DD . In this case, material instability would begin much earlier as


compared to the previous strategy (Fig. 22d). However, if s is low
enough the localized material would not go all the way to fracture.
So, after some time, when the string is being stretched at section
DD (Fig. 22e) some of the deformation would be taken up by the
previously localized region, i.e., at sections AA , BB and CC . As a
result, the strain would get distributed more uniformly along the
entire length of the string. With the right combination of c and
s at each section the string could be stretched to a greater length
without breaking (Fig. 22f). This is very similar to what happens in
SPIF.
It might be thought that if the previous neck is taking up some
of the deformation in subsequent tool passes then it should grow.
The question might arise that if this is the case then why fracture
in SPIF does not occur at the originally formed local shear band,
instead of at the contact zone around the tool tip. The ability of
a local shear band to share some of the subsequent deformation
without going to fracture depends not only on the extent of the

Fig. 19. Ratio (p /f ) at sections A, B, C, and D for 70 cone formed using (a) SPIF (b) large punch.

1586

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

Fig. 20. Contours of localization ag, z depth and plastic strain at (a) at onset of diffused necking (b) at onset of localized necking (c) just before fracture, for punch forming
of the 70 cone.

deformation but also on the location of application of the deformation. Note that in the schematic representation of the noodle
theory shown in Fig. 22(ef), for the SPIF strategy, the localized
neck growth is more in the region near the actual section of load
application. This is because as the distance of the neck from the
actual point of load application increases the ability of this neck to
share some of the deformation reduces. As a result, after the onset
of localized necking, it is always the neck closest to the contact zone

that grows to fracture. Essentially, the portion of the necking region


that is responsible for sharing most of the subsequent deformation
also moves along with the location of load application.
Further supporting evidence for the noodle theory is provided
by the plots of plastic strain shown in Figs. 9a and 16b. As mentioned earlier, these plots show that the plastic strain increases
gradually and very regularly along the prole of the SPIF cone,
i.e., from sections A to D, indicating a more uniform distribution

Fig. 21. Contours of localization ag, z depth and plastic strain at (a) at onset of diffused necking (b) at onset of localized necking (c) just before fracture, for SPIF of the 70
cone.

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

1587

Fig. 22. (a) Stretching the string at the free end (b) material localization at a single location on the string. (c) Fracture at location of material localization. (d) Stretching the
string by s at location c from the free end (e) continuous material localization along length of the string (f) elongation to a greater length without fracture.

of plastic strain along the component wall. This also means that
even after the tool has moved on the previously formed region
is actually taking up some deformation in the subsequent tool
passes. On the other hand for the punch forming case, the plastic
strain is concentrated at section D. In fact, at section D after a
certain point (at which point this region is in a state of localized
necking) the plastic strain rate increases dramatically indicating a
rapid concentration of strain in this region after localized necking.
The existence of a larger localized region in SPIF is further supported by the observation of material localization all along the outer
surface of the formed SPIF components (Fig. 23). The components
shown are funnels formed with an incremental depth of 0.5 mm.
The z depth between the localized bands was measured using a
depth gauge to be approximately 0.5 mm. Furthermore, it was visually observed during the forming process that these shear bands
initiated and grew in the regions where the tool was currently in
contact with the sheet. Fracture always occurred at a previously
generated shear band closest to the current position of the tool.
6. Discussion
The goal of this section is interpret the results obtained in this
work in an attempt to correlate it to work done in the past and
to examine the implications of the observations made, on the SPIF
process.

plastic strain on the outer side of the sheet cause greater damage
accumulation on the outer side of the sheet (Fig. 8). This causes the
crack in SPIF to begin on the outer side of the sheet and propagate
inwards. Therefore, when considering failure in SPIF the combined
effect of both local bending and shear must be accounted for.
6.2. Inuence of shear on formability in SPIF as compared to
conventional forming
It has been proposed in the past (Allwood et al., 2007; Allwood
and Shouler, 2009) that increased shear in SPIF could be the reason for increased formability in SPIF as compared to conventional
forming. This work partially supports this theory by showing that
increased shear does increase f in SPIF (Fig. 18) which should
cause lesser damage accumulation in SPIF. At the same time, as
compared to conventional forming, the increase in damage accumulation caused by local bending of the sheet around the SPIF tool
overwhelms the reduction in damage accumulation due to higher
shear (Fig. 19). Consequently, damage accumulation is faster in SPIF
than in conventional forming (Figs. 8a and 16a). Therefore, attributing the increased formability in SPIF as compared to conventional
forming, solely to shear, might not be a complete explanation. This
raises the question of why formability is higher in SPIF than conventional forming.
6.3. The Noodle theory of failure in SPIF

6.1. Deformation mechanisms in SPIF


Experimental work in the past (Jackson and Allwood, 2009) has
shown that in SPIF shear along the toolpath increases with depth of
deformation, is greater along the direction of the toolpath motion
than perpendicular to it and is greater on the inner side of the
sheet than on the outer side. This is supported by the current work
(Figs. 12 and 13). Furthermore, it is shown that the outer side of
the sheet is subjected to greater plastic strain due to local bending
of the sheet around the tool (Fig. 9). The lower shear and greater

Martins et al. (2008) proposed that failure in SPIF occurs by uniform thinning without evidence of localized necking. This work
shows what appears to be localized necking on the outer side of
the components formed with SPIF (Fig. 23). At the same time it is
important to note that the occurrence of localized necking is a fairly
subjective phenomenon as far as experimental observations are
concerned. It is very difcult to decide whether the occurrence of
what might appear to be a neck is purely a geometric phenomenon,
purely a material deformation affect or a combination of both of

1588

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

Fig. 23. Regions along the outer surface of SPIF components indicating material localization.

these. However, their claim that in SPIF a neck does not grow completely to failure upon initiation is supported by the current work
as well as shown in Section 5.3.
Emmens et al. (2009) examined formability in SPIF in terms of
the suppression or retardation of necking. They noted that while
localization of material in SPIF is inevitable, the increase in formability as compared to conventional forming can be explained in
terms of mechanisms which reduce stress at the location of the
originated neck to a level below that required for further growth of
the neck. This work shows that material localization is very much an
essential characteristic of SPIF due to the local nature of deformation in the process (Figs. 20 and 21). This happens because in-spite
of higher shear the accumulation of plastic strain, and therefore
of damage, is much higher very early on during the deformation in
SPIF. This should cause earlier onset of material instability and fracture in SPIF as compared to conventional forming. While material
instability does begin early on, actual fracture occurs much later in
SPIF than in conventional forming.
A new noodle theory is proposed that explains the increased
formability in SPIF as compared to conventional forming in-spite
of the fact that damage accumulation and onset of material instability is faster in SPIF. The theory goes farther than taking material
localization as an indicator of the occurrence of fracture by analyzing what happens after material localization. It is shown that
the inherently local nature of deformation in SPIF allows the generation of a larger region of unstable, but not fractured, material
before actual failure occurs (Fig. 21). It is proposed that it is the
ability of this region, in essence, to share some of the deformation in the subsequent passes of the tool which is the root cause for
increased formability in SPIF (Fig. 22). The fact that the crack occurs
around the tool contact zone instead of at the rst originated neck is
explained by the fact that the extent to which a neck grows depends
on not only the extent of the subsequent deformation but also on
the location of the application of this deformation.
This theory might also explain the reason for the inability to
accurately predict failure in SPIF using conventional FLCs. Conventional FLCs predict the occurrence of material localization. Using
FLCs to predict failure in conventional sheet forming (Filice et al.,
2002; Hussain et al., 2009) makes the assumption that the transition from material localization to fracture is so fast that keeping
a margin of safety from the occurrence of material localization is
enough to prevent fracture. This assumption is true for conventional processes, as is shown in the simulation of the punch forming
case, where material localization is quickly followed by fracture
(Fig. 20). However, for SPIF the transition from material localization
to actual fracture is much slower in SPIF than in conventional forming due to the local nature of deformation in the process. This is why

FLCs, which predict material localization, are unable to accurately


predict fracture in SPIF.
A practical concern that arises from the observation of a larger
localized necking region in SPIF is that parts formed by SPIF will
be severely damaged even before going into actual operation. It is
important to note that any material has internal voids which are
subjected to damage under even a small tensile stress. The damage
evolution in general plays a more important role in nal fracture
compared to the damage initiation. Therefore, simply taking the
initiation of localized necking at one material point as an indicator for damage is premature. Additionally, another relevant work
(Agrawal et al., submitted for publication), a study on a channel part
has shown that SPIF formed parts have a signicantly longer fatigue
life compared to parts obtained from conventional machining or
bending processes.

6.4. Effect of process parameters on fracture in SPIF


Process parameters such as incremental depth, tool size, tool
rotation, feed rate and friction at the toolsheet interface affect
both local bending and shear in SPIF. Therefore, they also control
damage accumulation and the occurrence of localized and diffused
necking which subsequently controls the nal fracture depth. In
addition to predicting fracture and explaining higher formability
in SPIF as compared to conventional forming, this link between
operational parameters and the occurrence of fracture provides a
powerful means to qualitatively predict the effect of operational
parameters on fracture in SPIF.
For example, plastic strain evolution is one factor that affects
damage accumulation. The phenomenon of higher plastic strain
on the outer side is induced by local bending of the sheet and is
therefore qualitatively inherent to SPIF. A change in the incremental depth and the tool size, might reduce the rate at which this
plastic strain and therefore damage accumulates. This will cause a
delay in onset of diffused necking and subsequent localized necking
which would in turn lead to a greater fracture depth. This might be
a possible explanation for the well documented observation that a
reduction in incremental depth or tool size results in an increased
fracture depth.
Whether shear or local bending dominates fracture in an SPIF
operation is a question that can only be answered subjectively
with reference to another SPIF operation with different operational
parameters. The answer depends on which operational parameters
dominantly affect plastic strain and shear strain and by how much,
and on which parameters are different between the two operations
being compared.

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

7. Conclusions
This work uses a fracture model in FEA to predict forming forces,
thinning and fracture in SPIF. The material model is calibrated using
the forming force history for a cone shape and validated using a
funnel shape by comparing experimental forming forces, thinning
and fracture depths with predicted values from FEA. Additionally,
to compare SPIF to a conventional forming case, the forming of the
same cone with a larger punch is also simulated using the same
material model. It is shown that the fracture depth in SPIF is greater
than that in the punch forming case. Eight elements through the
thickness of the sheet are used and it is ensured that hourglass
control, mass scaling and speeding up of the tool have a negligible
effect on the simulation results. The deformation mechanisms in
SPIF are found to correlate well to past experimental results, in a
qualitative manner.
The deformation mechanisms in SPIF are analyzed in-depth
using results from FEA. The following are the main conclusions:
1. Fracture in SPIF is controlled by both local bending and shear.
Local stretching and bending of the sheet around the hemispherical end of the tool cause higher plastic strain on the outer
side of the sheet, which increases damage on the outer side as
compared to the inner side of the sheet. Greater through-thethickness shear on the inner side of the sheet also has a similar
effect on damage accumulation. The combined effect of these
two phenomena is the initiation of the crack on the outer side of
the sheet which then propagates towards the inner side of the
sheet.
2. Greater shear in SPIF only partially explains the increased formability as compared to conventional forming. Shear is greater in
SPIF than in conventional forming which does have an effect of
retarding damage accumulation in SPIF by increasing f , especially on the inner side of the sheet. However, local bending in
SPIF is so much higher that it overwhelms the effect of increased
shear and increases damage evolution in SPIF as compared to
conventional forming. In-spite of the greater damage accumulation in SPIF, fracture depth and plastic strain before fracture
are greater than in conventional forming. Therefore, increased
shear in SPIF cannot be held solely responsible for increased
formability as compared to conventional forming.
3. The local nature of deformation in SPIF is the root cause for
increased formability as compared to conventional forming. A
so-called noodle theory, is proposed as an explanation for the
increased formability in SPIF as compared to conventional forming. This theory proposes that local deformation in SPIF increases
plastic strain and damage very early on during the deformation and causes earlier onset of material instability, as compared
to conventional forming. However, due to the local nature of
deformation in SPIF this region with material instability is not
deformed all the way to fracture. In subsequent tool passes this
unstable region takes up some of the deformation and the maximum plastic strain and deformed depth before fracture are
increased. Therefore, the local nature of deformation is the primary reason for increased formability in SPIF as compared to
conventional forming.
4. The link between process parameters and their effect on fracture
can be explained via their effect on shear and local bending. This
work establishes a link between shear and local bending, the
consequent occurrence of material instability and subsequent
fracture in SPIF. Developing a further link between operational
parameters and their qualitative and quantitative inuence on
shear and local bending can be used to qualitatively explain the
effect of process parameters on formability.
5. Future work. While this work examines the effects of local bending around the tool, stretching along the component wall and

1589

through-the-thickness shear there are other effects that should


also be incorporated to obtain an even more accurate picture of
damage evolution and fracture in SPIF. In the near future kinematic hardening and sheet anisotropy will be incorporated into
the fracture model. Furthermore, the undamaged ow stresses
may depend on strain rate when strain rate sensitive materials,
like stainless steel, are used with high feed rates and high spindle
speeds, which will also be considered. Future work will also focus
on a more detailed mesh convergence analysis as well as independent calibration of the fracture model parameters without
using SPIF experiments. Furthermore, experiments and simulations will be conducted to examine and exploring the effects of
various operational parameters and shape types on the occurrence of fracture with the objective of improving formability in
the process even further.
Acknowledgements
The authors of this work gratefully acknowledge the support
provided by the National Science Foundation (CMMI-0758607 and
CMMI-0727843) for this work.
References
Agrawal, A., Ziegert, J., Smith, S., Woody, B., Cao, J. Study of dimensional repeatability
and fatigue life for deformation machining bending mode. Submitted to ASME
Journal of Manufacturing Science and Technology for publication.
Allwood, J.M., Shouler, D.R., 2009. Generalized forming limit diagrams showing
increased forming limits with non-planar stress states. International Journal of
Plasticity 25, 12071230.
Allwood, J.M., Shouler, D.R., Tekkaya, A.E., 2007. The increased forming limits of
incremental sheet forming processes. Key Engineering Materials 344, 621628.
Cerro, I., Maidagan, E., Arana, J., Rivero, A., Rodrguez, P.P., 2006. Theoretical and
experimental analysis of the dieless incremental sheet forming process. Journal
of Materials Processing Technology 177 (13), 404408.
Emmens, W.C., van den Boogaard, A.H., 2009. An overview of stabilizing deformation mechanisms in incremental sheet forming. Journal of Material Processing
Technology 209, 36883695.
Emmens, W.C., vanderWeijde, D.H., vandenBoogaard, A.H., 2009. The FLC, enahnced
formability, and incremental sheet forming. In: Proceedings of the International
Deep Drawing Research Group IDDRG 2009 International Conference, 1-3 June
2009, Golden, CO, USA.
Eyckens, P., Duou, J., Van Bael, A., Van Houtte, P., 2010. The signicance of friction
in the single point incremental forming process. International Journal of Mater
Form 3 (Suppl. 1), 947950.
Filice, L., Fratini, L., Micari, F., 2002. Analysis of material formability in incremental
forming original research article. CIRP Annals Manufacturing Technology 51
(1), 199202.
Filice, L., Ambrogio, G., Micari, F., 2006. On-line control of single point incremental
forming operations through punch force monitoring. CIRP Annals Manufacturing Technology 55 (1), 245248.
Huang, Y., Cao, J., Smith, S., Woody, B., Ziegert, J., Li, M., 2008. Experimental and
Numerical Investigation of Forming Limits in Incremental Forming of a Conical
Cup. Transaction of the North American Manufacturing Research Institution of
SME.
Hussain, G., Gao, L., Hayat, N., Zirana, X., 2009. A new formability indicator in single point incremental forming. Journal of Materials Processing Technology 209,
42374242.
Hussain, G., Gao, L., Hayat, N., Qijia, L., 2007. The effect of variation in the curvature of
part on the formability in incremental forming: an experimental investigation.
International Journal of Machine Tools & Manufacture 47, 21772181.
Henrard, C., Boufoux, C., Duchene, L., Duou, J.R., 2007. Validation of a new nite
element for incremental forming simulation using a dynamic explicit approach.
Key Engineering Materials 344, 495502.
Jackson, K., Allwood, J., 2009. The mechanics of incremental sheet forming. Journal
of Materials Processing Technology 209 (3), 11581174.
Jeswiet, J., Micari, F., Hirt, G., Bramley, A., Duou, J., Allwood, J., 2005. Asymmetric
single point incremental forming of sheet metal. CIRP Annals Manufacturing
Technology 54 (2), 88114.
Malhotra, R., Huang, Y., Xue, L., Cao, J., Belytschko, T., 2010a. An investigation on
the accuracy of numerical simulations for single point incremental forming
with continuum elements. Proceedings of the 10th International Conference
on Numerical Methods in Industrial Forming Processes NUMIFORM 2010, AIP
Conference Proceedings 1252, 221227.
Malhotra, R., Reddy, N.V., Cao, J., 2010b. Automatic 3D spiral toolpath generation
for single point incremental forming. Journal of Manufacturing Science and
Engineering 132 (6), 10.
Martins, P.A.F., Bay, N., Skjoedt, M., Silva, M.B., 2008. Theory of single point incremental forming. CIRP Annals Manufacturing Technology 57, 247252.

1590

R. Malhotra et al. / Journal of Materials Processing Technology 212 (2012) 15731590

Silva, M.B., Skjoedt, M., Bay, N., Martins, P.A.F., 2009. Revisiting single-point incremental forming and formability/failure diagrams by means of nite elements
and experimentation. Journal of Strain Analysis 44, 221234.
Smith, L.M., Averill, R.C., Lucas, J.P., Stoughton, T.B., Matin, P.H., 2003. Inuence of
transverse normal stress on sheet metal formability. Journal of Plasticity 19,
15671583.
Szekeres, A., Ham, M., Jeswiet, J., 2007. Force measurement in pyramid shaped parts
with a spindle mounted force sensor. Key Engineering Materials 344, 551558.
van Bael, A., Eyckens, P., He, S., Boufoux, C., Henrard, C., Habraken, A.M., Dulfou,
J., van Houtte, P., 2007. Forming limit predictions for single point incremental
sheet metal forming. Proceedings of the 10th ESAFORM Conference on Material
Forming, AIP Conference Proceedings 907, 309314.
Xue, L., 2007a. Damage accumulation and fracture initiation in uncracked ductile
solids subjected to triaxial loading. International Journal Solids and Structures
44 (16), 51635181.

Xue, L., 2007b. Ductile fracture modeling-theory, experimental investigation and numerical verication. PhD thesis. Massachusetts Institute of
Technology.
Xue, L., 2009. Stress based fracture envelope for damage plastic solids. Engineering
Fracture Mechanics 76 (3), 419438.
Xue, L., 2010. Localization conditions and diffused necking for damage plastic solids.
Engineering Fracture Mechanics 77 (8), 12751297.
Xue, L., Belytschko, T., 2010. Fast methods for determining instabilities of elasticplastic damage models through closed-form expressions. International Journal
for Numerical Methods in Engineering 84, 14901518.
Yao, H., Cao, J., 2002. Prediction of forming limit curves using an anisotropic yield
function with prestrain induced backstress. International Journal of Plasticity 18
(8), 10131038.

You might also like