Sheet Incremental Forming: Advantages of Robotised Cells vs. CNC Machines

See
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/221785947
Sheet Incremental Forming: Advantages of

Robotised Cells vs. CNC Machines
Chapter December 2006
DOI: 10.5772/4913 Source: InTech
CITATIONS
READS
85
4 authors:
Massimo Callegari
D. Amodio
Universit Politecnica delle Marche
Universit Politecnica delle Marche
191 PUBLICATIONS 579 CITATIONS
SEE PROFILE
SEE PROFILE
Elisabetta Ceretti
Claudio Giardini
Universit degli Studi di Brescia
University of Bergamo
128 PUBLICATIONS 1,028 CITATIONS
SEE PROFILE
All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.
SEE PROFILE
Available from: Massimo Callegari

Retrieved on: 16 August 2016
25
Sheet Incremental Forming: Advantages of
Robotised Cells vs. CNC Machines
Massimo Callegari1, Dario Amodio1, Elisabetta Ceretti2, Claudio Giardini3
1Dipartimento
di Meccanica, Universit Politecnica delle Marche, Ancona, Italy

di Ingegneria Meccanica, Universit degli Studi di Brescia, Italy
3Dipartimento di Progettazione e Tecnologie, Universit degli Studi di Bergamo, Italy
2Dipartimento
Open Access Database www.i-techonline.com
1. Introduction
In recent years the traditional sheet metal forming processes, suitable for high volume
batches, do not correctly meet new market requirements characterised by high flexibility,
reduced time-to-market, low cost for small batch production, etc. Moreover, they are not
suitable for producing low cost prototypes and pre-series components. Thus new sheet
metal forming techniques are very often required and pursued by manufacturing industries
and have been intensively undertaken by scientific research groups (Siegert et al., 1997;
Amino et al., 2000; Shima, 2001; Kochan, 2001; Shim et al., 2001; Filice et al., 2002; Iseki &
Naganawa, 2002; Kim et al., 2003; Yoon et al., 2003; Ceretti et al., 2002, 2004; McLoughlin et
al., 2003; Allwood et al., 2005; Lamminen, 2005; Meier et al., 2005). Among the new
innovative technologies, the sheet Incremental Forming (IF) can be successfully used for
small pre-series batches or prototypes. IF is a process where common and simple tools
mounted on CNC machines, instead of complex die sets, are used to deform locally a
workpiece. In recent years many studies have been done on IF and many are still in
progress with the aim of finding both the most affecting process parameters and the suitable
machines and working centres to run experiments and production (Park & Kim, 2002, 2003;
Jeswiet et al., 2005a, 2005b; Duflou et al., 2005a, 2005b; Hirt et al., 2005; He et al., 2005a, 2005b;
Ambrogio et al., 2005; Bambach et al., 2005).
Unlike the standard metal forming process, fast production changes are possible thanks to
the very simple IF machine configuration. Even if the time required for making one product
is much longer than in the traditional press forming, the IF advantages are gained on tool
design and production in prototyping phase. IF could be also successfully applied in
completion flexible work cells, for example after hydroforming operations for slots or small
parts finishing. Furthermore, instead of using general purpose CNC machines, the modern
incremental sheet processes can be directly performed on robotised cells. This will enhance
the advantages in flexibility and production time reduction since a robotised cell equipped
with the proper tools can produce the part and, on the same fixture, realise the completion
operations such as flanging, trimming and so on.
To form the sheet into the desired shape an ad hoc tool, mounted on the machine spindle or
on a robot gripper, is moved according to the given tool path. Several IF strategies have
been developed which mainly differ for equipment and forming procedure. In particular,
Source: Industrial Robotics: Programming, Simulation and Applicationl, ISBN 3-86611-286-6, pp. 702, ARS/plV, Germany, December 2006, Edited by: Low Kin Huat
494
Industrial Robotics - Programming, Simulation and Applications
the process can be divided into:

Single Point Incremental Forming (SPIF), where the sheet is deformed by a single
tool (no support is present).
Two Points Incremental Forming (TPIF), sheet deformation is ensured by a tool
and a local support (a kind of partial die).
Full Die Incremental Forming, where the tool deforms the sheet against a die; this
die can be realized with cheap materials such as wood, resin or low cost steel; the
use of a die ensures a better and more precise shape of the final piece.
The above described strategies can be used on both CNC machines and industrial robots.
This paper presents an overview of the most recent researches carried out by the
Universities of Bergamo and Brescia and by the Polytechnic University of Marche about the
innovative IF process, describing the used working machines, the equipment devices and
the process experimental and simulative optimisation applied to Full Die IF.
More in details, the results of the use of an industrial robot for the IF with Full Die will be
analysed in terms of a potential increase of technological opportunities (e.g. the use of 5 axes
manufacturing) and of process versatility (e.g. the capability of performing extra flanging or
completing works) and compared with traditional 3 axes machining centre results.
2. Machines and equipment

2.1 CNC milling machine
As reported in the previous paragraph, IF is characterised by different strategies. The one utilised
in the present research refers to Full Die Incremental Forming. In fact, since the final goal of this
research is to identify working rules for pre-series production, this methodology should allow
the achievement of the best results in terms of geometrical and surface tolerances. It must be
underlined that it is not necessary to use an expensive material for the die (such as particular
resistant steel) since the working loads and deformation areas are reduced, but polymeric
materials, thermosetting resins, hardwood and any other material satisfying strength, stiffness
and surface finishing requirements can be used so allowing an overall cost reduction.
IF with Full Die is based on the local deformation of a sheet on a die obtained by means of a punch
moved by a CNC machine over a die which can have a negative (Fig. 1a) or positive shape (Fig. 1b):
in this last case a movable blank holder is required. The main components of the equipment are:
1. a punch mounted on the spindle of the CNC machine;
2. the die, which reproduces the part geometry;
3. the blank holder to keep the sheet in the right position during the process (fixed or
movable).
Punch
Fig. 1. The set up components.
Blank Holder
Die
(a)
(b)
Sheet Incremental Forming: Advantages of Robotised Cells vs. CNC Machines
495
The die is mounted on the table of the CNC machine and the moving punch deforms the
sheet up to the die. The punch path must be optimized by means of an accurate study of the
final geometry, in order to obtain a good surface finishing. When the die has a positive
geometry the blank holder is moved by four hydraulic actuators to keep the sheet steady in
the working position (Fig. 1b) otherwise is fixed (Fig. 1a).
The punch has a cylindrical body with a spherical head. The sphere dimension is an
important process variable. In fact, large headed tools assure better material flow and
production time reduction, while small sphere dimensions are needed to accomplish the
dimensional features of the part (the tool radius must be equal or lower than the minimum
radius of curvature of parts surface).
To generate the punch trajectory CAD, CAM and CAE techniques were used. Since the
tool movement strategy is very important for obtaining safe components, several tool
paths were investigated in previous studies (Ceretti et al., 2003; Giardini et al., 2004a,
2004b, 2004c, 2005a, 2005b) and the best one must be identified by taking into account the
positive or negative shape of the die, the shape of the part, the tool dimensions and the
sheet material.
2.2 Robotized cell
Conventional industrial robots do not have the stiffness nor the accuracy that are generally
required for the incremental forming operations (Lamminen, 2005) but the rather new
parallel kinematics machines can be designed so as to show good features from this point of
view: for the present research the COMAU Tricept HP1 robot has been used, a six axes
machine with hybrid structure (a serial wrist is mounted on top of a parallel shoulder),
shown in Fig. 2. It is able to apply a maximum thrust of 15 kN over a work envelope of 2000
mm x 600 mm, with a repeatability better than 0.03 mm. The drawbacks of the machine
mainly lie in the great anisotropy of its workspace, that is also characterised by a complex
shape and a limited dexterity; therefore a simulation tool has been implemented in order to
be able to analyse off-line machine kinematics and assess beforehand the feasibility of a
certain task.
Fig. 2. The hybrid robot Tricept HP1.
496
A pneumatic gripper holding a punch with a spherical tip is attached to the robot flange.
Various punches with different radii are used but in any case the small dimensions of the
ball tips allow to obtain the required high local pressures by the application of forces
considerably smaller than in conventional stamping. The die and the blank are clamped
together by means of several clips, whose design and setting are very important for the
quality of the operation: in fact a rigid constraint at frame boundaries can be desired in
some cases as well as the possibility to control the material flow towards the die in other
situations (Ceretti et al., 2003). The die with the supporting frame and the robot are mounted
on the same (thick) metal sheet: an automatic tool changing system was available beside the
robot but it has not been used in the present experimentation.
3. Analysis of the whole production process

The design of the production process begins from the definition parts geometry, that is
usually made by importing the CAD model of either the part or the die. Then the tool path
is generated by a CAM program. The part program used to machine the die is useless and a
specific path must be generated instead, to take into account that a plastic deformation is
being induced and to satisfy the needs of realizing good quality parts. The CL file is then
imported in FEM packages to evaluate the maximum forces that are required to complete
the forming process (ANSYS) and to simulate the operation (PAMSTAMP) in order to
forecast the final shape and the thickness distribution of the part. This is a fundamental task
when studying the part feasibility.
A kineto-static model of the robot has been implemented in Matlab by the Authors therefore
it is now possible to verify that tool path is entirely inside the workspace of the manipulator
and that the forces required at tool tip can be actually developed by the motors at the
actuated joints. If the check is not passed, it is necessary to change the punch (e.g. a smaller
size one) or the relative position between the die and the robot (e.g. by means of a rotating
table) or the kind of tool path (e.g. the tool can be differently oriented in such a way as to
load it mainly with axial forces).
At this phase the availability of a commercial multibody simulator can be useful, e.g. to
verify that the joints are not running out of their admissible stroke or that no impact with
other machinery in the cell occurs. The IGRIP package, that has been used to this aim, also
allowed to generate robots task in controllers own programming language, i.e. PDL in this
case; then the preliminary version of the program has been directly uploaded to Tricept
controller for the final tuning. The off-line programming is also enabled by an IGRIP
calibration utility, that has been used for the initial setting of cells CAD model according to
the real data measured in the laboratory.
Some modules of the recalled procedure have been previously outlined in (Callegari et al.,
2006) and will be explained in more details in the following sections.
3.1 CAD/CAM module
The solid model of the die is based on the geometry of the part and corresponds either to
convex or to concave die shapes. The tool path can be different in the two cases, since for
positive die the tool usually deforms the blank starting from the inner part and moving
towards the boundary, while the opposite approach is used for negative forming.
A CAD package has been used for modelling the part and the die and for the related CAM
497
processing: a fixed slicing of the part was used, with a step depth (or pitch) that has been
varied from 0.5 mm to 1.0 mm. Both 3 axes (for both CNC machine and robot) and 5 axes (for
robot) paths have been generated, see Fig. 3 and 4.
(a)
(b)
Fig. 3. Negative forming with concave die: 3 axes path (a) and fixed step slicing (b).
(a)
Fig. 4. Negative forming with concave die: 5 axes path (a) and robot wrist (b).
(b)
When the die is concave, different punch movement strategies can be used. In particular,
one strategy starts the process from the part boundary and progressively deforms the sheet
through trajectories drawn at constant Z levels (direct negative forming); the other strategy
is characterised by a straight tool movement inside the blank till the maximum depth of the
die is reached and then a spiral movement towards the part boundaries starts (inverse
negative forming), see Fig. 5. This last approach can be used only with limited die depths
and assures a better material flow from the boundary so guaranteeing a lower thinning of
the part. Unfortunately, this strategy could not be applied with the robotised cell because
the robot was not able to develop the high lateral thrusts that were required. To overcome
this limit a progressive solution was tested: in this case the punch is moved as in inverse
negative forming but it moves downwards up to the bottom of the die following
progressive spiral paths with different geometrical dimensions whose effects were
investigated.
The velocity of the tool together with the path variables (spiral width, pitch and so on)
have a great relevance for the quality of the part and must be chosen according to the
other parameters of the process, like sheet material and thickness, depth of deformation,
etc.
498
If the surface finishing of the part is an important aspect of the process, the path conducted
as a pocketing with constant step depth needs to be modified. In fact, this step down path
from the top to the bottom of the pocket, in which the tool follows a series of consecutive
contours with fixed step depth, is the simplest one but presents two disadvantages. In
particular the sheet is marked at the transition point between consecutive layers and the
quality of flat or near to flat surfaces is poor when high step depth or pitch (more than 0.5
mm) are used.
In this case to conduct pocketing with constant scallop height (Sc) can give better results. This
is a step down path from the top to the bottom of the pocket, in which the tool follows a series of
consecutive contours with variable step depth (the maximum of which must be furnished) in
order to keep constant the value of the scallop height (Fig. 5d). This kind of path reduces the
disadvantages of the first type and, in particular, the flat surfaces show a better quality.
Step depth, scallop height and type of tool path influences were studied during the
experimental tests, changing their values.
Spiral
width
(a)
(b)
Pitch
(c)
(d)
Fig. 5. Direct (a) and inverse (b) negative forming with progressive downward movement
(c) at fixed step depth or fixed scallop height (d).
3.2 Analysis of the forming process
Analytical model
A rough estimation of the state of stress of the blank under deformation can be obtained by
analytical models, while more detailed FEM simulations must be performed to go inside the
matter and assess the influence of the main process parameters.
For sake of simplicity, it has been closely followed the approach of Iseki (2001), which refers
to a planar state of deformation. This approximation can be rather well satisfied in case of
spherical rollers, with good lubrication and relatively small curvature paths. Referring to the
geometrical setting and notation of Fig. 6, representing a negative forming, the z axis is
taken along the direction of the incremental depth steps, and the y axis along the current
motion direction.
499
(a)
(b)
Fig. 6. Sketch of the negative incremental forming setting (a) and planar model of the arising
state of deformation (b).
The contact angle between tool and blank is:
+
h
b
d
arctan z
hx2 + hz2
hx
= arcsin
(1)
It is noted that half of the initial thickness of the metal sheet, t0, must be added to both
curvature radiuses of tool and die, b and d respectively. The hypothesis of plain strain
state leads to:
A2
A2
=
L + Rd A1 A 3 hx A 3
x = z = ln
(2)
where : A 2 = b and A 3 = hx cos hz sin + d .
Fig. 7. Equilibrium of metal sheet during negative incremental forming.

With reference to Fig. 7, the total effort causing material deformation, F, can be well
approximated being equal and opposite to the resultant of the 2 traction forces T at the
borderline of contact surfaces. The tensile force T on the metal sheet is:
(3)
T = 2 Rbt
If actual sheet thickness t is expressed as a function of the initial thickness t0 and of the
deformation , it can be obtained:
n +1
2
(4)
( x ) n
T = 2 Rb
Kt0 e x
3
where the material hardening law = k n has been considered. The two Cartesian
components of interaction force at the end effector can now be simply expressed as:
500
Fx = T (1 cos )
(5)
Fz = T sin
The simple inspection of (1-5) shows that, as expected, the lateral force to be developed at
the tool increases with the contact angles (and therefore with the depth h) and decreases
with the distance between the local centres of curvature of tool and die.
The results of the current model have been compared with the corresponding FEM
simulations and experimental trials, whenever available: a good agreement has been proved
in many situations, with results generally over-estimated and therefore conservative for the
sake of process safety. On the other hand, the model is not able to emulate closely the actual
thinning of the sheet (and therefore forecast the possible occurrence of tears or wrinkles) or
the paths with small curvature radii, where clear three-axial states of deformation establish.
If these cases must be investigated, a FEM simulation is unavoidable, while the analytical
model still preserves its validity for a first assessment of the forces occurring at tool tip, that
is an important input information for the kineto-static feasibility check that will be
explained in detail in next section.
Numerical process simulation
In order to verify the process ability in realizing the desired shape, that is the process
feasibility, it is important to conduct FEM analyses of the sheet deformation. Once the FEM
model has been validated by comparing the obtained results with the experimental ones, it
is possible to use the model itself to forecast the process parameters influences on the final
part shape and on sheet thinning which are relevant aspects in part quality evaluation.
Moreover, different punch movement strategies can also be analysed and the best one can
be identified prior to the experimental phase.
The simulations were conducted using both implicit (DEFORM) and explicit
(PAMSTAMP) codes. This latter has the advantage of reducing the computational time
with respect to the implicit code, in addition the explicit code results are closer than the
implicit to the experimental thickness values. As a consequence, in the following of this
paper, only the results of the explicit FE code will be presented. In both cases, the FE
programs move the punch using as input the CL files obtained from the CAM module. The
model components are the sheet (modelled as elastic-plastic element), the die, the punch
and the blank-holder (considered as rigid bodies), see Fig. 8. The friction was modelled
according to the Coulomb law.
A first set of simulations was run to identify the most suitable values for friction coefficient
and blank-holder force by comparing the experimental thickness distribution along the
piece section. Once the model was validated it was possible to use the model itself to
forecast the feasibility of a given part shape and to evaluate the part quality, in
correspondence of variations of tool path or other process parameters (e.g. tool diameter,
feed rate).
To allow the simulation in 3D DEFORM environment, a suitable program able to move the
punch (treated as the upper die) along a defined trajectory as obtained from the CL file has
been developed and implemented in a user subroutine. This module allows also to restart
the simulation from an intermediate point of the trajectory as required when a remeshing
occurs. Moreover, another module has been developed in order to allow the thickness
calculation and representation using proper variables specifically defined in the preprocessing phase.
501
(a)
(b)
Fig. 8. FEM model objects: sheet, punch, die and blank-holder (a); the thickness
representation (b).
3.3 Feasibility kineto-static checks
The only kind of robots presently able to develop the required thrusts at the tool are parallel
kinematics machines, that are also characterised by high stiffness and good accuracy:
unfortunately, the closed-loop architecture generally yields a workspace of complex shape and
characterised by high anisotropy, which complicates the planning of effective operations (Merlet,
2005). On the other hand, the information provided by robot manufacturers are often too poor to
enable an off-line assessment of the feasibility of the current incremental forming operation.
Therefore more detailed simulations must be performed beforehand to check that all the
generated 5-dimensional paths are included inside the 6-dimensional workspace of the
manipulator and that, at every position, the motors torques required by the task can be actually
developed. Since the velocities of the robot are usually quite low, a kineto-static model of the
manipulator is sufficient to assess the kinematic feasibility of the required tool positions and to
correlate workspace efforts to joint space torques. The Tricept HP1 used in the tests has a
maximum thrust of 15 kN in the axial direction when the tool is aligned along the same z axis but
its static performances dramatically downgrade as the structure leaves such favourable
configuration (see Fig. 9 for a qualitative representation). Therefore a complete kineto-static
model of the robot has been developed and implemented in the Matlab programming language,
as explained with details in (Callegari et al., 2005) and summarised in the following sections.
Fig. 9. Static performances of the robot Tricept HP1 (Comau, 1995 and 2001).
502
Robot kinematic model

In order to develop the kineto-static model of the Tricept HP1, the conventional steps of
direct and inverse kinematic analysis had to be performed. It is well known that the
resolution of the position problem is always the most challenging phase of the work, usually
ending up with a complex non-linear system of algebraic equations. In this case the
approach suggested by Siciliano (1999) for the resolution of the (3 dof) parallel shoulder has
been followed, by adapting it to the case of a 6 axes hybrid machine: due to the difficulty of
the problem, direct kinematics has been solved numerically, while inverse kinematics
admits closed-form solutions. Velocity kinematics, on the other hand, can be expressed in
the usual linear form:
1
vx

v
2
y
(6)
3
vz
= J (1 , " , 6 )
4
x

y
5

z
6
where in the left hand side all the terms of tool linear and angular velocities are collected in
a single vector, while i , i=1,,6 are motors speeds and J is a proper Jacobian matrix. It is
noted that the Jacobian incorporates all the terms of the velocity mapping, up to ball screws
pitches or gear reducers ratios.
Manipulability ellipsoids
Let us consider the set of joint velocities of constant unit norm:
T = 1
(7)
It is desired to describe the Cartesian space velocities of the end-effector that can be
generated by such velocities at a given configuration in space. The image of the sixdimensional sphere (7) in the operational space of tool linear and angular velocities is called
(velocity) manipulability ellipsoid and is given by the quadratic form:
[v
v
v = 1
(8)
where the 6x6 square matrix v is defined by:
v () = J T () J 1 ()
(9)
Cartesian space velocities are high along the direction of the major axis of (9), while only
low velocities are yielded along its minor axis: therefore an isotropic (kinematic) behaviour
of the manipulator is manifested when the ellipsoid becomes a sphere. Just to visualise the
matter, in case of a planar robot with two equal motors, the form (8-9) would represent an
ellipse whose major and minor axes (the eigenvectors of v) indicate the directions along
which the tool can be moved with the maximum and minimum velocities respectively (that
are given by the square roots of the eigenvalues of v).
Due to the well-known duality between kinematics and statics, for every pose of the
manipulator inside its workspace (position and attitude of the tool) the 6 joint torques
balancing the external wrench of forces F and moments M applied at tool tip are given by:
F
= JT
M
503
(10)
Following the same approach previously outlined for the characterisation of robot
velocities, the ellipsoid representing all the Cartesian forces and moments that correspond
to unit motors torques, T =1, can be derived. In the case of the present application the
external moments at tool tip are null, therefore only the three-dimensional ellipsoid of endeffector forces is meaningful:
FT f F = 1
(11)
where f is the 3x3 upper left sub-matrix of J JT.

The knowledge of matrix f clearly characterises the static performances of the manipulator and
also represents an effective means to visualise them; it is noted that f depends on all the 6
coordinates of robots pose. As an example, Fig. 10 shows in dark patterns the force
manipulability ellipsoids at various points of Tricept workspace, when the tool is asked to lay
parallel to the Cartesian axes of global frame. Of course to gain more insight into the
performance of the actual robot under consideration, proper weights must take into account the
different sizes of joints motors; moreover, due to the linearity of static relations, the application
of higher forces/torques can be represented by simply re-scaling the resulting graphs.
(a)
(b)
Fig. 10. Force manipulability ellipsoids of the robot Tricept HP1 in two Cartesian planes:
upper (a) and side (b) view (end-effector parallel to global frame).
Static performances
For every position of the manipulator inside its workspace (position and attitude of the tool)
the 6 joint torques balancing the external wrench of forces F and moments M applied at
tool tip are given by (10), where the 6x6 manipulator Jacobian matrix is highly varying with
tool position in our case. Once a task has been assigned and the arising forces have been
assessed, the feasibility checks can be easily performed: the joint torques that are computed
by simulation must be compared with the actual rated torques of the motors driving
shoulder and wrist axes, about 13 Nm and 3 Nm respectively, but of course the gear ratio of
the Harmonic Drives and the efficiency of all the transmission has to be accounted for.
Figure 11 shows a sample analysis performed by means of the developed simulation
package: tool tip spans a vertical plane at constant height (z=1600 mm) with the tool
perpendicular to the plane itself and charged by an axial load of 15 kN. The figure plots the
value of the maximum torque requested to any one of shoulder motors: it is noted that the
limit value of 13 Nm is easily overcame when the tool approaches workspace boundaries;
wrist motors, on the other hand, are almost idle since the external force passes through the
centre of the spherical wrist.
504
Fig. 11. Maximum torque delivered by shoulder motors (z=1600 mm, Fz= 15 kN).
In the simulation of Fig. 12, instead, the configuration of the robot is still the same but an
external force of 500 N is vertically directed along the x axis: in this case shoulder motors are
scarcely charged but wrist motors cannot sometimes deliver all the torque that should be
required.
(a)
(b)
Fig. 12. Maximum torque delivered by shoulder (a) and wrist (b) motors (z=1600 mm, Fz= 0.5
kN).
3.4. Cell simulation
The analysis of all the cell is performed by means of the IGRIP package by Delmia, Fig. 13:
it is a multibody simulation code for applications of industrial automation and its libraries
already have all the information needed to simulate and program off-line the main
industrial robots presently on the market. Its use is particularly important in case several
intelligent pieces of equipment cooperate simultaneously in the cell, therefore needing a
strict coordination of the tasks.
In our case, tool trajectory has been previously defined by means of CAD/CAM programs
and then statically validated to be sure robot motors can actually push the punch against
the blank with the required forces. Then, the path has been imported into the IGRIP
505
environment to perform a multibody simulation, providing the kinematic feasibility of

the task by taking into account actual joints strokes and possible impacts with obstacles or
other equipment. For larger parts the use of a tilting table, synchronised with the robot
and governed by the same robot controller, can be necessary, as well as for some
operations, e.g. flanging, where the required line of action of tool force would not be close
to the preferred thrust direction of the Tricept: in these cases a proper off-line
programming reduces the set-up time of the operation and enhances the feasibility
chances of the task.
Since IGRIP data base codes all the kinematic data of the robot and is also aware of control
characteristics, robot task can be generated in the proper programming language, i.e. PDL in
our case, and directly uploaded to the controller: all production information like cycletimes, occupation of resources, etc. can be estimated off-line at this phase of the work. Of
course the PDL program needs to be refined at robot side in any case, but a closer
agreement between off-line program and the final code is obtained by exploiting the
calibration module, that is purposely developed.
Fig. 13. Multibody simulation of the cell.
4. Experimental and simulative results

The first laboratory tests have been performed on a square socket of small size,
37 mmx37 mm side and a depth of 21 mm in Ancona by using the robotised cell and
40 mmx40 mm side and a depth of 20 mm in Brescia, where the CNC machine was used.
Figure 14 shows the metallic and wooden dies, the blank holder and all the fixtures
mounted on a vertical frame.
506
This set of tests was difficult for the robot due to the particular shape of the die, whose
lateral walls were almost vertical; in fact this geometry determines high reaction forces (due
to the large contact surface between tool and blank) whose lines of action are almost
perpendicular to the (preferential) axial direction, therefore often causing the block of the
motors.
During these preliminary experiments it has been shown that, notwithstanding the
relatively low velocities of the tool during the process, the velocity profile of the robot
influences the quality of the product and even more the performances that can be obtained.
In fact during accelerations the inertia forces are added to the elasto-plastic reactions
coming from the blank under deformation, while the forming task is easier for the robot
during deceleration ramps. Moreover, it resulted that the smaller the radii of the corners are,
the higher torques are required to the motors for the impending state of three-axial
deformation: better results and lower torques have been registered by using tools with
smaller sphere diameters.
It must be emphasized that, when the CNC machine is used to perform the experimental
tests, no limits can be identified for this simple geometry. In fact, the machine stiffness and
the motor powers allow the deformation of the sheet since the working loads are low. In
addition, increasing the feed rate gives the possibility of reducing the working time without
affecting the part quality.
Fig. 14. Metallic and wooden dies with blank holder and clamping fixtures.
The first attempts of forming a steel blank have been performed by using 3 axes paths on a
0.6 mm thick AISI 304 stainless steel sheet; in this way it was possible to draw a comparison
with the results obtained using both the robotised cell and the CNC machine. In the case of
the robotized cell a maximum depth of about 14 mm was reached in direct forming, while
the motors blocked at the depth of only 10 mm in inverse forming (see Fig. 15). In the first
case a set of square paths at higher depths has been progressively realised, with a pitch of
0.5 mm without working the bottom of the pocket; in the second case, after reaching the
desired depth, a spiral path was traced with a spiral step of 1 mm. The quality of surface
finishing is quite different in the two cases, since the pocket bottom does not come into
contact with the tool in direct forming, while in second case it is deformed by the tool. In
this second case the quality is better, even if in the part bottom it is still visible the point
corresponding to the straight path of the tool before starting the spiral path which deforms
the bottom (Fig. 15). A very tight fixing of the blank holder is required in order to avoid the
formation of wrinkles and the rotation of the sheet.
To reduce the working loads, it was necessary to lower the tool pitch to 0.25 mm instead of 1
mm. The consequence of this change was an increase of the total cycle time. A reduction of
507
total cycle time was obtained by using a different language primitive for the path
interpolation. Instead of requiring the robot to pass exactly through the programmed points,
and therefore to stop when cornering, a spline interpolation allowed the tool to pass close
to the programmed points, without stopping every time. The quality of the part is still good,
while cycle times have been strongly reduced.
In a second set of tests the capabilities of the 6 axes robotized cell were used to orient
properly the tool with respect to the sheet local normal and to the feed direction. First
of all, it must be made clear that in robot trajectory planning 3 degrees of freedom are
constrained by the path to be followed by tool tip, while 2 more degrees of freedom can
be used to orient the punch along a suitable direction in space; therefore, by having
the availability of a 6 axes machine, infinite possible trajectories can still be specified
after all.
Fig. 15. Deformed steel parts for direct (left) and inverse (right) forming obtained with the
robotized cell.
It must be pointed out that the kinematics of the robot is rather complex, therefore the
optimal alignment of the tool does not necessarily mean that the tool itself is aligned
along the direction of robot motion. In our trials, the tilting of the tool with respect to the
feeding direction has been set so as to try to minimise the moments acting at wrist centre
(by the way, the tool is charged by axial loads in this case). Moreover, due to wrist
mechanical structure and to the limited winding of its joints, it was necessary to invert the
sense of rotation of the motion around the die at every pass. This was also visible on the
surface of the final part.
508
In future work, the redundancy of the robot with respect to the task could be used to:
take automatically into consideration the constraints coming from the possible
mutual collisions between blank/frame and tool/wrist;
try to avoid the time consuming re-winding of wrist axes, therefore speeding up
the process;
align the whole robot structure along the main direction of the manipulability
ellipsoids.
It must be said that the comparison with the 3 axes forming previously performed did not
show major differences, apart from a slight higher quality and precision of the finished part
mainly due to the smaller bending of the tool.
Considering the experiments conducted with the CNC machine, only the limit of the
maximum reachable slot depth was found since it affects the sheet thinning (when thinning
is too high the part breaks). Several tool paths of the direct forming type were tested by
changing the spiral width and the spiral step according to Giardini et al. (2005a). The
experiments conducted showed that roughness of the produced pieces was influenced only
by spiral step and was not influenced by the tool path type. The thickness distribution,
analysed along a part transversal section, is affected by the tool path and by the spiral step.
In particular, summarising the results obtained in terms of part roughness, sheet thickness
and maximum reachable punch depth it is possible to define the optimal Full Die IF process
parameters, that is inverse negative forming with spiral step 0.5 mm, tool pitch 1 mm and
feed rate 400 mm/min.
The sheet thickness shows a minimum in correspondence of the bottom radius of the
pocket, see Fig. 16. In one of the studied cases, the spiral step was so large that the
final geometry of the part had a very low quality due to large wrinkles on the pocket
bottom.
Fig. 16. Deformed steel obtained with the CNC machine.

Figure 17 shows an example of the comparison between simulations and experiments
conducted on the CNC machine in terms of sheet thickness measured along a
transversal cross-section. In particular the simulative results correspond to the
PAMSTAMP optimized model, that is the model with the most suitable values for
friction coefficient (0.17 between sheet and blank-holder and sheet and die, 0.15
between sheet and punch) and blank-holder force (50 kN), referred to
120 mmx120 mmx0.6 mm AISI 304 sheet with spiral step of 0.5 mm, feed rate of 400
mm/min and punch diameter of 40 mm.
509
Fig. 17. Simulative and experimental thickness distributions along the transversal cross
section of the realised part.
Finally some automotive components have been formed. Starting from CAD models of
the parts, the resin dies were produced and then the part programs for the incremental
forming have been developed. Figure 18 shows a component used in chair back
assembly; two different kinds of steel blanks have been formed with this component:
AISI 304 and DC04. The results are satisfactory in both cases but the higher quality of the
AISI 304 part shows that this material can be worked better through the incremental
forming technique.
During the working of the central slot, a certain compliance of the wrist became apparent,
even if the overall quality of the part was good. Moreover, a little misalignment of the
planes of the blank and the die caused the wrinkles that are visible outside the working
area: a better realisation and tuning of the blank holder would certainly mitigate this
problem.
Fig. 18. Formed chair backs: AISI 304 (left) and DC04 (right).
510
The research project analyzed also the feasibility study of some finishing operations just to
show the potential versatility of an automated incremental forming cell: to this aim, the
hatch of a multipurpose vehicle has been flanged by pressing the outer edge initially bent at
90. The part was initially clamped between a punch and the die, then a specially designed
tool, see Fig. 19a, rolled along all the outside border. The use of this rolling tool to deform
locally the outer edge, allowed to flange the part without any problem and to avoid the
expensive hydraulic presses generally used in this case or the manual operations that are
time consuming and cost expensive but are necessary when the attention is focused on preseries production. Similar application was studied using a roller mounted on a 3 axes CNC
machine. The results are reported in Fig. 19b.
(a)
(b)
Fig. 19. Hemming of sheets in a robotised cell (a) and on a CNC machine (b).
5. Conclusions
The incremental forming of metal sheet parts can be an interesting alternative to manual
forging of prototypes and pre-series blanks or to the manufacturing of shells for resin dies
used for the production of small batches. Such characteristics of small-volume production
would call for an increase in the level of flexibility and automation, possibly leading to the
use of CNC machines or robotised cells able to produce or complete the parts. In particular,
the use of robotised cells, with automatic tool change, can dramatically reduce the process
time since on the same fixture it is possible to deform the part, cut the part, bend or flange
the borders, load/unload the part, etc.
The present contribution has described the crossed experiments performed at the
Polytechnic University of Marche in Ancona and at the Universities of Brescia and Bergamo
to assess the feasibility of the automated processing by using both a traditional CNC
machine and an industrial robot. It is noted that the research and industrial processes of
incremental forming realised so far have ever used 3-axes CNC milling machines, apart
from the hammering process patented by the Fraunhofer Institute for Manufacturing
Engineering and Automation of Stuttgart (Shaefer & Schraft, 2005).
Unfortunately the conventional serial robots do not have the required stiffness and are not
able to apply the necessary forces to deform incrementally the blank, but the rather new
family of parallel robots has characteristics similar to CNC machining centres, while still
keeping the versatility of a robot. The complex kinematics of the machine needed the
development of a special purpose simulation environment to design beforehand the
experiments and assess their feasibility. The necessary force at tool tip has been evaluated
both analytically, with a simplified approach based on a plane strain state, and numerically,
511
by means of a commercial FEM code: different kinds of tool trajectories have been generated
by CAM programs, based on actual part geometries. A final assessment of whole cell layout
and working has been performed by means of the IGRIP package, exploiting also the
possibilities of the coordination with other external axes (e.g. a revolving table during
flanging, a tool changing system, etc.) and its capability to generate the part program for the
different commercial controllers.
Several experimental tests have been performed in order to validate the complex
methodology for system design and prototyping and to study the several parameters
playing a significant role in the process, as for instance: the different types of materials (e.g.
AISI 304 and DC04 steel, copper), the number of axes of the task (3 or 5), the kind of
interpolation between the points, the path imposed to the tool (e.g. depth first or breadth
first), the size of punch end, etc. Many tests have been performed on a die specifically
developed in-house for the execution system trials, then a few tests have been performed on
commercial components, on dies provided by a car manufacturer supplier, with the
execution of simple flanging operations too.
At the end of the research a development environment has been set up, able to interface the
different software tools in order to support the process designer in making the correct
choices. It must be said that, even if the simulation environment proved to be powerful and
reliable, the whole design process of an experiment seemed very complicated by the
complexity of the used equipment: therefore the use of more powerful robots, with larger
workspaces would be desirable. It's Authors opinion that parallel robots can be a viable
alternative to CNC machines for the execution of incremental forming processes, especially
if it is possible to exploit the high versatility of the machine for further completing
operations. More tests should be needed to completely assess the benefits of robotics, with
the possible availability of more powerful machines, that are already available on the
market. As for the system design and prototyping tool that has been developed, it proved to
be effective and reliable, even if resulted to be quite complex and more integration would be
needed between the single software modules.
6. Acknowledgments
This work has been developed within the research programme Ultraflex (see Lamier,
2005), partially funded by the Italian Ministry of Industry and by Lamier SpA (La Loggia,
Italy).
7. References
Allwood, J.M.; Houghton, N.E. & Jackson, K.P. (2005). The Design of an Incremental
Sheet Forming Machine. Proc. Shemet 2005, pp. 471-478, Erlangen, Germany,
April 5-8
Ambrogio, G.; Filice, L.; Gagliardi, F. & Micari, F. (2005) Sheet Thinning Prediction in Single
Point Incremental Forming, Proc. Shemet 2005, pp. 479-486, Erlangen, Germany,
April 5-8.
Amino, H.; Makita, K. & Maki, T. (2000). Sheet Fluid Forming and Sheet Dieless NC
Forming, Proc. New Developments in Sheet Metal Forming 2000, pp. 39-66, Stuttgart,
Germany, May 23-24.
512
Bambach, M.; Azaouzi, M.; Campagne, L.; Hirt, G. & Batoz, J.L. (2005). Initial Experimental
and Numerical Investigations into a Class of New Strategies for Single Point
Incremental Sheet Forming (SPIF), Proc. Esaform 2005, pp. 671-674, Cluj-Napoca,
Romania, April 27-29.
Callegari, M.; Gabrielli, A.; Palpacelli, M.-C. & Principi, M. (2006). Robotised Cell for the
Incremental Forming of Metal Sheets, Proc. ESDA 2006, 8th Biennial ASME
Conference Engineering Systems Design and Analysis, Turin, July 4-7.
Callegari, M.; Palpacelli, M. & Principi, M. (2005). Analisi della manipolabilit del robot
industriale Tricept, Atti del XVII Congresso AIMeTA di Meccanica Teorica e Applicata,
Firenze, 11-15 Settembre 2005.
Ceretti, E.; Giardini, C.; Attanasio, A. & Maccarini, G. (2002). Some Experimental
Edvidences in Sheet Incremental Forming on CNC Machines. Proc. Numisheet 2002,
pp. 429-434, Jeju Island, Korea, October 21-25.
Ceretti, E.; Giardini, C. & Attanasio, A. (2003). Sheet Incremental Forming on CNC Machines,
Proc. SheMet 2003, pp. 49-56, University of Ulster, Newtownabbey, UK, April 14-16.
Ceretti, E.; Giardini, C.; Attanasio A. (2004). Experimental and Simulative Results in Sheet
Incremental Forming on CNC Machines, Journal of Materials Processing Technology,
vol. 152, pp. 176-184.
Comau (1995). Tricept HP1. Maintenance Manual. rel 3.X/4.X
Comau (2001). Tricept HP1. Users Manual. rel 3.X/4.X
Duflou, J.R.; Lauwers, B.; Verbert, J.; Tunckol, Y. & De Baerdemaeker, H. (2005b).
Achievable Accuracy in Single Point Incremental Forming: Case Studies, Proc.
Esaform 2005, pp. 675-678, Cluj-Napoca, Romania, April 27-29.
Duflou, J.R.; Szekeres, A. & Vanherck, P. (2005a). Force Measurements for Single Point
Incremental Forming: An Experimental Study, Proc. Shemet 2005, pp. 441-448,
Erlangen, Germany, April 5-8.
Filice, L.; Fratini, L. & Micari, F. (2002) Analysis of Material Formability in Incremental
Forming, Annals of CIRP, Vol. 51, No. 1, pp. 199-202.
Giardini, C.; Ceretti, E.; Attanasio, A. & Pasquali, M. (2004a). Analysis of the Influence of
Working Parameters on Final Part Quality in Sheet Incremental Forming, Proc. 3rd
International Conference and Exibition on Design and Production of Dies and Molds, pp.
191-198, Bursa, Turkey, June 17-19.
Giardini, C.; Ceretti, E.; Attanasio, A. & Pasquali, M. (2004b). Feasibility Limits in Sheet
Incremental Forming: Experimental and Simulative Analysis, Proc. Esaform 2004,
pp. 515-518, Trondheim, Norway, April 28-30.
Giardini, C.; Ceretti, E. & Contri, C., (2004c). Analysis of Material Behavior in Sheet
Incremental Forming Operations, Proc. 8th NUMIFORM 2004, pp. 1016-1021, vol.
712, Columbus-Ohio, USA, June 13-17.
Giardini, C.; Ceretti, E. & Attanasio, A. (2005a). Further Experimental Investigations and
FEM Model Development in Sheet Incremental Forming, Proc. Shemet 2005, pp.
501-508, Erlangen, Germany, April 5-8.
Giardini, C.; Ceretti, E. & Attanasio, A. (2005b). Optimization of Sheet Incremental Forming
Process by Means of FE Simulations, Proc. Esaform 2005, pp. 691-694, Cluj-Napoca,
Romania, April 27-29.
513
He, S.; Van Bael, A.; Van Houtte, P.; Szekeres, A.; Duflou, J.R.; Henrard, C. & Habraken,
A.M. (2005a). Finite Element Modeling of Incremental Forming of Aluminum
Sheets, Proc. Shemet 2005, pp. 525-532, Erlangen, Germany, April 5-8.
He, S.; Van Bael, A.; Van Houtte, P.; Tunckol, Y. & Duflou, J.R. (2005b). Effect of FEM
Choices in the Modelling of Incremental Forming of Aluminium Sheets, Proc.
Esaform 2005, pp. 675-678, Cluj-Napoca, Romania, April 27-29.
Hirt, G.; Bambach, M. & Junk, S. (2003). Modelling of the Incremental CNC Sheet Metal
Forming Process, Proc. SheMet 2003, pp. 495-502, University of Ulster,
Newtownabbey, UK, April 14-16.
Iseki, H. (2001). An Approximate Deformation Analysis and FEM Analysis for the
Incremental Bulging of Sheet Metal Using a Spherical Roller, Journal of Material
Processing Technology, Vol. 111, pp. 150-154.
Iseki, H. & Naganawa, T. (2002). Vertical Wall Surface Forming of Rectangular Shell Using
Multistage Incremental Forming with Spherical and Cylindrical Rollers, Journal of
Materials Processing Technology, Vol. 130-131, pp. 675-679.
Jesweit, J.; Duflou, J.R. & Szekeres, A. (2005a). Forces in Single Point and Two Point
Incremental Forming, Proc. Shemet 2005, pp. 449-456, Erlangen, Germany, April 5-8.
Jesweit, J.; Young, D. & Ham, M. (2005b). Non-Traditional Forming Limit Diagrams for
Incremental Forming, Proc. Shemet 2005, pp. 409-416, Erlangen, Germany, April
5-8.
Kim, Y. H. & Park, J. J. (2003). Effects of Process Parameters on Formability in Incremental
Sheet Metal Forming Technique, Journal of Material Processing Technology, Vol. 140,
p. 447-453.
Kochan, A. (2001). Dieless forming. Assembly Automation. Vol. 21, No. 4, pp. 321-322.
Lamier. (2005). New Advances in Process Flexibility: the Perspective of an OEM Supplier,
Proc. 9th International Conference "FLORENCE ATA 2005 - Vehicle architectures:
evolution towards improved safety, low weight, ergonomics and flexibility", Florence,
May 11-13.
Lamminen, L. (2005). Incremental Sheet Forming with an Industrial Robot Forming Limits
and Their Effect on Component Design, Proc. Shemet 2005, pp. 457-464, Erlangen,
Germany, April 5-8.
McLoughlin, K.; Cognot, A. & Quigley, E. (2003). Dieless Manufacturing of Sheet metal
Components with non Rigid Support, Proc. SheMet 2003, pp. 123-130, University of
Ulster, Newtownabbey, UK, April 14-16.
Meier, H.; Dewald, O. & Zhang, J. (2005). A New Robot-Based Sheet Metal Forming Process,
Proc. Shemet 2005, pp. 465-470, Erlangen, Germany, April 5-8.
Merlet, JP. (2005). Parallel Robots, 2nd Ed., Springer, Dordrecht.
Park, J.-J. & Kim, Y.-H. (2002). Effect of Process Parameters on Formability in Incremental
Forming of Sheet Metal. Journal of Materials Processing Technology. Vol. 130-131, pp.
42-46.
Park, J.-J. & Kim, Y.-H. (2003). Fundamental Studies on the Incremental Sheet Metal
Forming Technique. Journal of Materials Processing Technology. Vol. 140, pp. 447453.
Schafer, T. & Schraft, R.D. (2005). Incremental Sheet Metal Forming by Industrial Robots.
Rapid Prototyping Journal. Vol. 11, No. 5, pp.278286.
514
Shim, M., & Park, J. (2001). The Formability of Aluminium Sheet in Incremental Forming,
Journal of Materials Processing Technology, Vol. 113, pp. 654-658.
Shima, S. (2001). Incremental Forming: State of the Art, Proc. IPMM 2001: Intelligent
Processing and Manufacturing of Materials, Vancouver, British Columbia, Canada,
July 29 - August 3.
Siciliano, B. (1999). The Tricept Robot: Inverse Kinematics, Manipulability Analysis and
Closed-Loop Direct Kinematics Algorithm. Robotica. Vol. 27, pp. 437-445.
Siegert, K.; Rennet, A. & Fann, K.J. (1997). Prediction of the Final Part Properties in Sheet
Metal Forming by CNC-Controlled Stretch Forming, Journal of Materials Processing
Technology, vol. 71, pp. 141-146.
Yoon, S. J. & Yang, D.Y. (2003). Development of a Highly Flexible Incremental Roll Forming
Process for the Manufacture of a Doubly Curved Sheet Metal, Annals of CIRP, Vol.
52, No. 1, pp. 201-204.

Sheet Incremental Forming: Advantages of Robotised Cells vs. CNC Machines

Uploaded by

Copyright:

Available Formats

Sheet Incremental Forming: Advantages of Robotised Cells vs. CNC Machines

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sheet Incremental Forming: Advantages of Robotised Cells vs. CNC Machines

Uploaded by

Copyright:

Available Formats

See