CIRP Annals - Manufacturing Technology 62 (2013) 267270

Contents lists available at SciVerse ScienceDirect

CIRP Annals - Manufacturing Technology

jou rnal homep age : ht t p: // ees .e lse vi er . com /ci r p/ def a ult . asp

A novel process for transforming sheet metal blanks: Ridged die forming
Mark A. Carruth, Julian M. Allwood (2)*
Department of Engineering, University of Cambridge, Trumpington Street, Cambridge CB2 1PZ, United Kingdom

A R T I C L E I N F O

A B S T R A C T

Keywords:
Sheet metal
Stamping
Blanking

Up to 20% of all sheet metal produced is scrapped as blanking skeletons. A novel process is therefore
designed and examined, aiming to transform tessellating pre-blanks in-plane into the real blanks
required for stamping. Prior to blanking, the sheet is formed with a set of ridged dies, from which preblanks are cut and then attened into true blanks. Several different approaches to designing ridged dies
are evaluated by simulation and experiment, and the best results demonstrate a potential reduction in
blanking yield losses for can-making from 9.3% to 6.9%.
2013 CIRP.

1. Introduction
Around 50% of annual global production of liquid metal is used
to make sheets, and around half of this never reaches a nal
product, but instead is scrapped (and recycled) at different stages
of production [1]. These high losses cause a signicant increase in
the specic embodied energy of the nal product (i.e. energy per
kg), particularly due to the compounding effect of yield losses
along the supply chain [2].
One of the biggest sources of scrap occurs during blanking,
when coils of strip metal are cut into shapes for use in downstream
forming operations. These shapes rarely tessellate, hence the waste
material from between the blanks (the skeleton) is scrapped.
However, the shape of the blank is dictated by the shape of the nal
part and the forming process used for its manufacture, and the
optimised design of this shape has already been subject to
intensive research. An approach to estimating workpiece deformation in deep drawing introduced in [3] and subsequently named
the geometrical mapping method [4] has been extended [5] for
use in optimisation routines [6]. This inverse approach describes
the target nal shape of the workpiece with a mesh of shell
elements, and then maps each node on the nal conguration to a
point on the initial at blank. By considering only the initial and
nal blank geometries, this approach gives a rapid solution for the
major strains in the workpiece, which can be used to optimise the
initial blank geometry to minimise nal part thickness variations.
The results show a reduction of 17% in material required
compared to the original blank shape. This method has been
applied for automotive components [7] and developed using ideal
forming theory [8], a faster inverse approach [9], the ESO technique
[10] and with approximate modelling [11]. These developments
provide faster solutions while offering similar reductions in blank
masses. However, despite these reductions, if the initial blanks are
cut with blanking-presses from constant width sheets, yield losses
may be unimproved.

* Corresponding author.
0007-8506/$ see front matter 2013 CIRP.
http://dx.doi.org/10.1016/j.cirp.2013.03.052

The use of tessellating blanks, for example using a hexagonal

blank to form a beverage can, could largely eliminate these losses.
However, without some change to the forming sequence, use of a
different blank shape would simply result in higher trim losses
after the rst stages of forming, moving the scrap burden further
down the supply chain but not reducing its magnitude. Is it
possible to design a forming process to transform a blank from one
shape to another, without reducing nal product quality?
This paper rst considers the theoretical basis for devising a
process to meet these requirements. A candidate, novel process,
ridged die forming, is then presented as one possible means to
create arbitrary in-plane deformation. The process is evaluated by
application to a case study on forming beverage cans, with the aim
of transforming a hexagonal pre-blank into a circular blank for
use in can bodies or ends.

2. Transforming blanks with uniform thickness change

An ideal in-plane blank transformation should maintain uniform blank thickness to allow downstream processing and assure
nal product quality. Assuming incompressibility, this imposes the
constraint that, during the transformation (over time 0  t  T, in
the plane xy, and assumed to have a constant rate), the in-plane
strain rates satisfy:

e xx x; y; t e yy x; y; t e zz constant

(1)

By mass conservation, the required thickness strain rate can be

calculated from the change in blank area, A, as:
1
T

e zz  ln

 
AT
A0

(2)

There is an innite family of deformation elds (vx, vy) satisfying

(1) and (2) while transforming a blank between given initial and
nal geometries. A useful reference case is the pattern of
deformation which requires minimum work and hence which

M.A. Carruth, J.M. Allwood / CIRP Annals - Manufacturing Technology 62 (2013) 267270

268

satises,
min W
vx ;vy

Z T Z
0

s e dV dt

(3)

where
2

2
3

e e 2xx e 2yy e 2tt 2e 2xy

(4)

The minimisation of (3) subject to the constraint of (1) can be

solved by an appropriate constrained optimisation algorithm
applied to a velocity eld discretised in time and space. A code was
written to achieve this, for a rigid-plastic material, using the nite
element method, with reduced integration quadrilateral elements
and hourglass control, and a gradient based optimisation method.
The code was applied to a test case of transforming a hexagon (the
highest order regular tessellating polygon) to a circular blank. Fig. 1
shows one twelfth of the resulting deformation, when the
thickness strain rate is chosen so that the nal circle perfectly
encloses the original hexagon.

pre-blanks. Forming within the uncut strip creates a helpful set

of boundary conditions at the edge of each pre-blank: material
cannot be drawn into the pre-blank, and the sheet must instead
stretch. Then, after blanking these pre-blanks can be attened,
during which further in-plane deformation will occur, to create the
nal, transformed, blank.
The transformation created by this process depends on the
geometry of the dies used to form the pre-blanks. After preliminary
trials of a range of die geometries, the work leading to this paper
has considered dies shaped with a set of many relatively shallow
ridges: the attraction of this approach is that the spacing and depth
of the ridges allows more inuence over the distribution of surface
tractions than with a dome-like punch.
The process is illustrated in Fig. 2. The uncut sheet is stamped
with ridged dies, shown in Fig. 2(a), to create ridged pre-blanks.
The shaped pre-blanks, which (in plan view) tessellate are then cut
from the sheet, shown in Fig. 2(b). The cut pre-blanks are attened,
causing a change of the plan-view shape of the blank, to produce
the nal blank geometry, demonstrated in Fig. 2(c) and (d). The
success of the process depends on the design of the ridged dies of
Fig. 2(a).
3. Die design methods

Fig. 1. Minimum work transformation of hexagon to circle. (a) Mesh deformation;

(b) deformation vectors (plotted on initial mesh).

In order to create the deformation eld of Fig. 1, tools must be

applied to create appropriate surface tractions. The required
tractions can be calculated, subject to a free choice of hydrostatic
pressure, from the deviatoric stresses associated with the strain
rates in the deformation eld. However, in practice tools can apply
surface tractions only by compression and its associated friction.
Thus, although it might be possible to create the deformation
pattern of Fig. 1 through design of a segmented tool, such a tool is
likely to be complex. Potentially a simpler segmented tool could be
designed to create a different (non-ideal) deformation eld. But an
alternative approach that could be easier to implement in practice
would be to deform the strip out of plane prior to blanking to create

Three approaches were taken to the design of ridged dies.

Firstly, a design was developed using an analogy to the folding of
leaves, based on an approach to designing deployable membrane
structures [12]. Many tree leaves consist of a central rib, with sets
of smaller ribs emanating at a xed angle on either side of the rib.
Folding along these lines provides a degree of control over the plan
view geometry of the leaf. De Focatiis and Guest observed that by
treating each segment of a regular polygon (e.g. each 1/6th of a
hexagon) as a separate leaf, and arranging the ribs in an
appropriate manner, the plan-view geometry of the polygon could
be adjusted. In the case of a hexagon, the required rib runs centrally
along each 1/6th of the polygon, with the ribs running parallel to
the sides of each segment (i.e. at a 308 angle to the central rib). A
design with 9 ridges was chosen and the depth of the ridges xed
such that along the central rib, the total length of the die surface
was approximately equal to the target radius of the blank. This
leads to the design shown in Fig. 3(a).
Secondly, a die parameterisation, illustrated in Fig. 4 and
summarised in Table 1, was optimised by repeated trials. Each
ridge is dened by a compound Bezier curve, with the gradients at
each endpoint chosen to provide C1 continuity between the
different segments of the hexagon. Finite Element simulation was

Fig. 3. Summary of three die designs assessed in experimental trials. (a) Die set 1;
(b) die set 2; (c) die set 3.

Fig. 2. Overview of concept for ridged die forming process. (a) The uncut sheet is
stamped with ridged dies; (b) tessellating pre-blanks are cut from the sheet; (c) the
pre-blank is attened between at dies; (d) the pre-blank is transformed to the nal
blank geometry.

Fig. 4. Parameterisation for die designs by manual optimisation.

M.A. Carruth, J.M. Allwood / CIRP Annals - Manufacturing Technology 62 (2013) 267270

269

Table 1
Summary of ridge design parameters used in Section 3.
Parameter

Description

N
kR/L
li
hi
Dli
Dhi

Number of ridges
Radial length of ridge prole
Radial location of ridge i on u = 0
Difference in height between ridges i and i + 1
Radial shift in location of ridge i from u = 0 to u = p/6
Change in height of ridge i from u = 0 to u = p/6

applied to predict the deformation created by each die, and in the

subsequent attening stage. A manual optimisation process was
then performed to maximise the predicted yield savings. This led to
the design shown in Fig. 3(b).
Although the parameterised approach provides exibility in
describing a range of die designs, it would be cumbersome to apply
to more complex blank transformations. A third design approach
was therefore applied, using the minimum work deformation eld
of Fig. 1. The eld was used to calculate a displacement vector from
each location on the initial blank to its location in the nal
conguration. The die ridge lines were then oriented perpendicularly to these displacement vectors, with ridge depth proportional to total displacement. The resulting design is shown in
Fig. 3(c).
4. Experimental design and numerical analysis
The experimental setup is shown in Fig. 5. To simulate the
boundary conditions of the real process (in which many pre-blanks
would be formed simultaneously), a small hexagonal region was
formed within a larger hexagonal blank which was clamped rmly
in a blankholder to prevent any drawing in of material. The
blankholder plates also act as a guide for the ridged dies, to ensure
they are correctly aligned throughout the process. The ridged die
surface is complex, so two different manufacturing methods were
evaluated. Some dies were manufactured by direct metallic laser
sintering (DMLS) on a thick plate. The prole was created by
incrementally building up layers of a discrete size (either 40 mm
and 20 mm), and the surfaces then treated with micro shotblasting
to reduce faceting from the DMLS process. This approach created
dies with good accuracy and strength, but poor ridge surface
quality creating some surface damage in the blanks. Other dies
were therefore manufactured by CNC milling, with a tolerance of
0.1 mm. This gave superior surface nish but reduced accuracy.
The blank material tested was an aluminium AA1100-H14 sheet
with 0.315 mm starting thickness. The entire assembly was
compressed in a universal testing machine, as shown in Fig. 6,
allowing controlled vertical displacement, to create different ridge
heights. After release from the blankholder, the deformed
hexagonal pre-blank was cut from the larger blank and then

Fig. 5. Schematic of experimental test assembly.

Fig. 6. Instron testing rig used for experimental trials.

attened between two at plates. High pressure forming grease

was used as lubrication in all stages of the experiment.
In parallel with these physical trials, Finite Element simulations
of all the experimental trials were performed using Abaqus/
Standard v6.10. The symmetry (rotational and reective) of the
problem was exploited to simplify the model, meaning only 1/12th
of the problem was simulated. The workpiece was meshed using
approximately 5000 C3D8I fully integrated 8-node brick elements
with incompatible modes enabled to improve the behaviour of the
elements in bending, with two elements through the thickness.
The tools were modelled as discrete or analytical rigid surfaces. The
rst experimental results were used to calibrate the FE model,
showing that predictions were highly sensitive to the friction
coefcient. A friction coefcient of 0.2 gave useful agreement
between simulation and experiment, and is in-line with prior work
on aluminium sliding against steel [13].
5. Results
Two metrics were used to assess the performance of the ridged
dies. The rst measures the radial displacement of the sheet along
each midline the bisector of the 6 segments of the hexagonal
pre-blank. This is an indicator of whether sufcient displacement is
being achieved to create a circular blank. The second predicts the
reduction in yield losses from the new process. This metric is
calculated from the largest circle that could be cut from the
attened blank in the experimental and FE results.

Fig. 7. Radial displacement achieved along midlines. (a) Die set 1; (b) die set 2; (c)
die set 3.

M.A. Carruth, J.M. Allwood / CIRP Annals - Manufacturing Technology 62 (2013) 267270

270

Fig. 8. Transformed blank shapes after forming (shading in the FE results is

proportional to radial displacement).

potentially reduce this loss to 6.9% If this could be achieved in

practice, it would have signicant value due to the high volumes of
global annual can manufacture and potentially, further die
renement could reduce yield losses further. However, the trials
have also demonstrated two limitations to the proposed process: a
lack of uniformity in through thickness strain and surface cracking
at maximum ridge height.
The areas of the blank which underwent most bending during
the folding stage showed extra thinning. Fig. 8 also shows that the
surface of the blanks have been damaged by an embossing effect
during folding and attening. This embossing suggests that the
actual level of thinning may be higher than that predicted by the FE
simulations, and hence the non-uniformity in thickness may be
greater than that suggested in Fig. 9.
Surface cracking at the location of maximum bending is a
common forming defect, suggesting that the bend radius used in
the ridged die was too severe. The dies reported in this paper
typically had nine ridges across their diameter, so future trials
should explore use of fewer ridges of greater depth, to increase the
minimum bend radius. A further feature was that in the attening
stage, the formed pre-blanks changed shape less than expected. A
different approach to attening, possibly aided by use of fewer
ridges, might allow a more effective attening stage.
The specic ridged die forming process evaluated in this paper
has demonstrated a potential improvement in material yield, but
with unacceptable surface damage and uneven thinning. Future
design effort may help to reduce these effects. However, the
general approach of the paper forming the strip prior to preblanking and attening is novel and shows considerable
potential for future developments, including the exploration of
many other combinations of forming before and after blanking.

Fig. 9. Contours of thickness strain for blanks formed using die set 2.

Acknowledgements

Fig. 7 shows a strong linear relationship between the radial

displacement along the midlines and the height of the ridges after
the folding of the pre-blank. The maximum attainable displacement for die sets 1 and 2 were approximately equal, but die set 3
performed worse, with around 40% less displacement.
In all cases, the maximum achievable ridge height was limited
by the onset of cracking during the folding stage. Cracking was
found to occur along the line of the ridges. Although the location of
the crack was well predicted by the FE simulations, corresponding
to the region of the largest equivalent plastic strain, it was not
possible to predict the ridge height at which cracking would occur
from the FE results.
Fig. 8 displays the nal perimeter shape of 1/6th of the nal
blanks from each die set. It demonstrates that although die sets 1
and 2 achieve approximately equal displacement, die set 2
provides a far greater yield improvement as it transforms the
blank to a more favourable shape. Although die set 3 achieves less
displacement, the improvement in yield is only slightly less than
that of die set 2, as it induces a more favourable shape
transformation to the perimeter of the blank.
Fig. 9 demonstrates contours of thickness strain for the results
of die set 2, taken from FE simulations. Although a signicant area
of the nal blank demonstrates nearly uniform thickness strain,
there are several small regions where the strain is much greater
than this, at around 10%.
6. Discussion and conclusions
If it were possible to cut close-packed circular blanks from
rolled strip, the minimum yield losses would be 9.3%. (In practice,
existing blanking processes must leave a narrow strip between
each circle, so current yield losses in can blanking are greater than
this.) The best results of Section 5 demonstrate that die 2 could

Both authors are supported by a Leadership Fellowship

provided by the UK EPSRC reference EP/G007217/1.

References
[1] Cullen, J.M., Allwood, J.M., Bambach, M.D., 2012, Mapping the Global Flow of
Steel: From Steelmaking to End-Use Goods, Environmental Science & Technology, doi:10.1021/es302433p, in press.
[2] Milford RL, Allwood JM, Cullen JM (2011) Assessing the Potential of Yield
Improvements, Through Process Scrap Reduction, for Energy and CO2 Abatement in the Steel and Aluminium Sectors. Resources Conservation and Recycling
55(12):11851195.
[3] Sowerby R, Che E, Duncan J (1982) Determination of Large Strains in Metal
Forming. The Journal of Strain Analysis for Engineering Design 17(2):95101.
[4] Gerdeen JC, Chen P (1989) Geometric Mapping Method of Computer Modelling
of Sheet Metal Forming. Numiform 89:437444.
[5] Guo YQ, Batoz JL, Detraux JM, Duroux P (1990) Finite Element Procedures for
Strain Estimations of Sheet Metal Forming Parts. International Journal for
Numerical Methods in Engineering 30(8):13851401.
[6] Guo YQ, Batoz JL, Naceur H, et al (2000) Recent Developments on the Analysis
and Optimum Design of Sheet Metal Forming Parts Using a Simplied Inverse
Approach. Computers & Structures 78(13):133148.
[7] Kim J, Kim N, Huh M (2000) Optimum Blank Design of an Automobile Subframe. Journal of Materials Processing Technology 101(13):3143.
[8] Park SH, Yoon YW, Yang DY, Kim YH (1999) Optimum Blank Design in Sheet
Metal Forming by the Deformation Path Iteration Method. International Journal
of Mechanical Sciences 41(10):12171232.
[9] Tang B, Zhao Z, Lu X, et al (2007) Fast Thickness Prediction and Blank Design in
Sheet Metal Forming Based on an Enhanced Inverse Analysis Method. International Journal of Mechanical Sciences 49(9):10181028.
[10] Naceur H, Guo YQ, Batoz JL (2004) Blank Optimization in Sheet Metal Forming
Using an Evolutionary Algorithm. Journal of Materials Processing Technology
151(13):183191.
[11] Cwiekala T, Brosius A, Tekkaya AE, Grydin O, Schaper M (2008) Efcient
Modelling and Simulation of Process Chains in Sheet Metal Forming and
Processing. Steel Research International 79(10):731737.
[12] De Focatiis DSA, Guest SD (2002) Deployable Membranes Designed from
Folding Tree Leaves. Philosophical Transactions Series A Mathematical Physical
and Engineering Sciences 360(1791):227238.
[13] Fratini L, Lo Casto S, Lo Valvo E (2006) A Technical Note on an Experimental
Device to Measure Friction Coefcient in Sheet Metal Forming. Journal of
Materials Processing Technology 172(1):1621.