Computer-Aided Design 41 (2009) 877–885

Contents lists available at ScienceDirect

Computer-Aided Design
journal homepage: www.elsevier.com/locate/cad

Tool path programming optimization for incremental sheet forming applications

Matthieu Rauch a , Jean-Yves Hascoet a,∗ , Jean-Christophe Hamann b , Yannick Plenel b
a
Institut de Recherche en Communications et Cybernetique de Nantes (IRCCyN) – UMR CNRS 6597, 1 rue de la Noe, BP92101, 44321 Nantes Cedex 03, France
b
Airbus, Saint Nazaire, France

article info abstract

Article history: Incremental sheet forming is an emerging process to manufacture sheet metal parts that is well adapted
Received 26 August 2008 for small batch production or prototypes. The adjustment time is short, as it is sufficient to modify the
Accepted 12 June 2009 tool motions to optimize the manufacturing process. Tool path generation therefore becomes a key topic
linked to incremental sheet forming, and process characteristics ask for dedicated tool paths. Hence,
Keywords: this paper first discusses the impact of tool path types and other programming parameters on process
Tool path optimization
implementation through an experimental campaign performed on a parallel kinematics machine tool.
Incremental sheet forming
Intelligent CAM programming
Then, a new approach to generate and control Intelligent CAM programmed tool paths is proposed. The
Forming force control major purpose of this innovative concept is to use process constraints for programming and controlling
CNC data the tool path, which are adapted during the running of the CNC program according to real-time process
data evaluation. Validation studies and an industrial implementation are finally presented to assess the
efficiency of the proposed approach.
© 2009 Elsevier Ltd. All rights reserved.

1. Introduction
Incremental Sheet Forming (ISF) is an emerging process to
manufacture sheet metal parts that is well suited for small batch
production or prototyping [1,2]. Indeed, it is highly flexible and
needs only low set-up costs: the required equipments are an
NC milling machine tool (or equivalent) equipped with dedicated
tooling; process implementation consists in making the tool,
performing computer-generated tool paths at preset levels which
decrease according to given increments. As a consequence, the
adjustment time is very short, as it is sufficient to modify the
tool motions to improve the manufacturing process. Tool path
generation therefore becomes a key topic linked to ISF.
The majority of the first publications did not focus on forming
tool paths [3–7] and were using forming strategies built by CAM
software (dedicated to material removal processes). However,
some interesting works studied tool path related parameters.
Kopac et al. showed in [8] that, in Single Point Incremental Sheet
Forming (SPIF), it is better to run the tool path from the edge of
the part to the centre in order to improve the accuracy. In [7],
Ceretti et al. worked on Two Points Incremental Forming (TPIF)
∗ limits away, especially the wall angle. Both the accuracy and part
Corresponding author. Tel.: +33 240376949.
E-mail addresses: matthieu.rauch@irccyn.ec-nantes.fr (M. Rauch),
jean-yves.hascoet@irccyn.ec-nantes.fr (J.-Y. Hascoet),
jean-christophe.hamann@airbus.com (J.-C. Hamann), yannick.plenel@airbus.com
(Y. Plenel).
0010-4485/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cad.2009.06.006
and tested two tool path types. According to their results, spiral
strategies are better than contour parallel strategies. In the same
way, other works highlighted the fact that the choice of forming
strategy has a great effect on the process performances; see for
example [2]. Attanasio et al. worked on the influence of the axial
increment in TPIF [9] and advise using a small scallop height and
variable axial increments to design the tool paths. A variant of TPIF
was implemented by Mao et al. in [10]. Instead of a complete shape,
a bolster rod holds the sheet metal and stands for a die. A way to
limit the circumferential ridges that appear between the tool paths
is to adopt a forward and backward forming strategy. As a result,
the surface roughness is better.
Recent research works have been carried out to overcome two
major drawbacks which are observed in ISF: formed part accuracy
and sheet metal integrity. To compensate the elastic springback ef-
fect, Ambrogio et al. program a higher slope than the CAD model
until the middle of the part [11]. In [12], this concept is extended
and named ‘‘vitiated trajectories’’, because the tool paths, which
are deliberately false, lead to accurate parts. An experimentally
based method is proposed by Hirt in [13]: the tool paths are mod-
ified after having measured the first produced part and its defects.
More recently, Duflou explored a promising approach in [14] by
implementing multi-step tool paths that allow moving the forming
integrity are increased. A similar method is proposed by Skjoedt
in [15]. Micari, in [16], discusses several methods to improve SPIF
capability and concludes that tool path optimization approaches
are the most promising.
878 M. Rauch et al. / Computer-Aided Design 41 (2009) 877–885
20
100
24
160
Fig. 1. Test part.
However, CNC tool path implementation and control during the
ISF process have not yet been widely discussed. On that point,
CAD-CAM competences are very useful. Moreover, tool path gener-
ation has been studied widely for the milling process. Even though
it is very different from ISF processes, the equipment and imple-
mentation (tool motion control) are very similar. It is consequently
profitable to rely on works made about tool path programming for
milling applications to propose dedicated ISF tool paths.
Two research directions were particularly studied in milling
tool path generation and control: a geometrical approach and a
technological approach. The geometrical approach looks for the
best tool path shape according to the characteristics of the man-
ufactured feature and of the cutting tool [17,18]. The technologi-
cal approach integrates process characteristics (machine dynamic
models, CNC functionalities, . . . etc) to generate the most adapted
tool motions [19,20]. Finally, technological process simulation [21]
can sharpen the results. Improvements are still being observed,
even though this manufacturing process has been widespread
worldwide for decades.
The aim of tool path improvement for ISF is of course not to
employ the same tool path geometries and parameterizations but
to study the methods developed to adapt the tool path generation
to a specific manufacturing process. Indeed, no CAM tool paths
specifically designed for ISF processes have been proposed yet.
One major purpose of this paper is to input ISF related constraints
into the tool path generation process. To carry this out, a new
approach to program and control incremental forming tool paths
is proposed. This approach is called Intelligent Computed Aided
Manufacturing (ICAM), and it focuses on CAD-CAM issues. It is
based on real-time process data evaluation to compensate the tool
path control on-line. To be effective, the ICAM approach has to
adapt to specific constraints associated with the process.
Thus, the first section of this paper evaluates the effects of form-
ing tool paths on the process implementation. The study focuses
not only on tool path parameters such as axial increment or fee-
drate but also on the tool path shape itself. As they are based on ex-
perimental tests, quantitative results are not valid for other sheet
metal materials. The objective of the first section is to discuss the
central effects of tool paths and their programming on the pro-
cess and not to propose prediction laws. Furthermore, deformation
mechanisms in incremental forming are still little understood. Few
works have been published about theoretical studies and models,
for example [22,23]; their authors admit that experimental analy-
ses are even now very effective to optimize the process. Then, the
second section of this paper presents the implementation of the
ICAM approach for ISF applications. Validation studies and an in-
dustrial implementation are also provided to assess its efficiency
in flexible manufacturing environments.
2. Tool path effects on process implementation: First results
This section presents the results of an experimental study which
aimed to evaluate the effects of the process parameters when
forming an entire shape. This action was carried out within an
industrial environment. Its goal was to identify the capability of
an ISF application to compete with already implemented processes
such as deep drawing or hammering. It focused on a specific group
of aeronautical parts; that is why some parameters (material,
thickness) are already selected.
2.1. Experimental setup
The experimental study consisted of the manufacturing of
an entire shape (Fig. 1). The aim was to measure the influence
of the forming parameters on the accuracy of the formed part
(especially the depth), the forming time, and the forming force.
Three parameters were studied: feedrate, axial increment and tool
path shape. For each of these factors, two values were tested, and
a design of experiments made of 12 runs was built.
Strategies 1 and 2 were made of constant level contouring
(Fig. 2). In Strategy 1, the axial increment was taken plainly along
the tool axis. In Strategy 2, the axial increment was obtained
gradually along one side of the square shape. Strategy 3 was a spiral
tool path: a quarter of the axial increment is taken along each side
of the square. This work did not intend to propose new forming
strategies but underscores the effects of tool path shape on the
process capability. Indeed, similar tool paths have already been
employed, for example in [2,7].
The other parameters remained the same for all experiments.
Sheets were made of aluminium alloy (5086) and had a 0.6 mm
thickness in order to stick to the industrial case of study. After a
short investigation based on experimental forming tests that com-
pared several tool radii (5 mm, 10 mm, 15 mm), a hemispherical
head tool with a 10 mm diameter was used. This choice meets the
results of [5].
2.2. Results
The design of experiments was performed through three man-
ufacturing criteria, which were forming loads, forming times and
formed surfaces properties. Table 1 provides the parameterization
and the results associated to each test of the design of experiment.
For each measured criterion, the results of the analysis of variance
are given in Table 2.
Forming forces were acquired during the process thanks to a
Kistler 3D component dynamometer. Measured force peaks are
higher for Strategy 1 than for the other strategies because the
axial increment is taken by a single motion along the tool axis,
 M. Rauch et al. / Computer-Aided Design 41 (2009) 877–885 879

Strategy 1 Strategy 2 Strategy 3

1 2
1 4 1 2,6
2 6
10 10
3,7 8 5 5 3 5
4
10 3
6 4 9 7
9,13 14 12 11 9 7,11 8
8 11
15 12
12

Fig. 2. Strategy 1–Strategy 2–Strategy 3.

Table 1
Parameterization and results of the experimental study. 1
Strategy Axial Feedrate Maximal Forming Depth Flatness a/2
inc. (m/min) forming time error error of
(mm) force (min) (mm) the bottom
(N) plan (mm) a/2
0.5 637 23 0.18 0.32 2
0.5
1.5 647 8 0.24 0.34
1
0.5 890 12 0.05 0.50
1.0
1.5 855 4 0.05 0.50 a
0.5 564 29 0.21 0.33 Localization of the tool paths
0.5
1.5 545 10 0.24 0.34
2
0.5 690 15 0.07 0.50 Complete Part Zoom of the center
1.0 of the part
1.5 671 5 0.18 0.53
0.5 598 23 0.55 0.25
0.5 Fig. 3. Measured points to check the part accuracy.
1.5 578 7 0.52 0.25
3
0.5 746 12 0.79 0.39
1.0 Fastened
1.5 726 4 0.70 0.36 Fastening Area
System
1
Table 2
Analysis of variance for the design of experiments. α
2
Maximal forming Forming Depth Flatness
Free Area
forces (%) time (%) error (%) error (%)

Strategy 30.8 3.7 87.3 28.5

Axial inc. 64.0 27.1 0.1 69.2
Feedrate 0.7 68.0 0.1 0.1 Working Area
Strategy — Axial inc. 4.6 0.5 10.8 1.6
Strategy — Feedrate 0.0 0.7 1.2 0.6
Fig. 4. Distribution of the areas on the fastening system.

which creates an impact on the sheet metal. Nevertheless, the
loads stay inside an acceptable working domain. The feedrate has
no significant effect on the forming forces (Table 2). This result is
confirmed by others experiments conducted with feedrate values
of 3 m/min and 5 m/min.
Concerning the manufacturing times, feedrates are low in ISF
in comparison to the kinematic capacities of current CNC machine
tools. Thus, the kinematics effects are very limited and the fastest
tool paths are the shortest. As a result, Strategy 1 and Strategy 3
return similar results because they have nearly the same length.
Strategy 2 is longer and less efficient.
The geometric accuracy of the formed parts was evaluated.
Two levels were investigated: a macro-geometric level (the depth
accuracy) and a micro-geometric level (the surface roughness). The
choice of the depth error as an accuracy criterion was motivated by
the functional requirements of the part and by its easy evaluation
process.
To quantify the depth accuracy, the bottom plan of each test
part was measured using a 3D touch probe. The measured points
are shown in Fig. 3. Their initial position was previously acquired
on the clamped sheet metal before starting the process. The mea-
sured length between the initial and final positions of these points
was compared to the height of the CAD model. The difference be-
tween these two values was interpreted as the depth error. The
results are given in Table 1.
Due to the symmetry of the formed parts, the flatness error of
the bottom plan was calculated from the difference between the
final positions of points 1 and 2 (Fig. 3). The measures of the two
other points confirmed the symmetry hypothesis.
The significance results in Table 2 confirm that feedrate has
no influence on the accuracy of the formed parts for the chosen
values (0.5 m/min and 1.5 m/min). Nevertheless, additional tests
and other works, such as [24], show that at over 2 m/min the sheet
metal formability decreases while the feedrate increases.
In contrast, the axial increment and forming strategy have
major effects on the geometry of the final part. Strategies 1 and
2 present almost the same results, because they are very similar.
Indeed, for Strategy 1 and Strategy 2, the flatness error of the plane
is larger than the depth error of the point 1, which means that
the bottom surface is domed. With Strategy 3, the flatness error
of the bottom plan is lower but the depth error is larger. From a
quantitative point of view, depth errors with Strategy 3 are three
times higher than with the others strategies.
For all the test parts, the pyramid height is always lower than
the programmed value (24 mm). This comes from a combination
of the elasticity of the aluminium alloy and the shape of the sheet
fastening system (Fig. 4).
Furthermore, the bottom plan has always the same shape. The
points that are located on the tool path always have the smallest
depth error. Further tests highlighted an excellent repeatability of
this process: when using the same parameters, the positions of the
measured points and the flatness error were similar with a margin
of 0.02 mm.
When considering these results, Strategy 1 appears as the best
solution in regards to the overall accuracy. However, the way to
880 M. Rauch et al. / Computer-Aided Design 41 (2009) 877–885
µm Length = 4 mm Pt = 29 µm Scale = 50 µm
20
10
0
-10
-20
0 0.5 1 1..5
Tool paths
Fig. 5. Surface profile of a formed part.
perform the incremental step along the tool axis and always in the
same corner of the shape can penalize the surface roughness and
even affect the sheet metal integrity. A closer look at the surface
properties, by measuring the surface roughness, can sharpen these
conclusions.
Concerning micro-geometrical issues, the surface roughness of
the formed part was inspected using a rugosimeter. The surface
profile can be directly linked to the axial increment. It meets
the results obtained by Hagan in [25]. Fig. 5 shows the profile
inspection using an increment of 0.5 mm and a slope of 30◦ :
the oscillation period matches the tool paths. Nevertheless, the
corresponding amplitude is limited to 0.02 mm. Moreover, the part
formed with Strategy 1 does not present a defect of the surface
integrity at the corners of the contour. This unexpected property
can be explained by the springback effect of the sheet metal.
It is consequently necessary to find a compromise between
surface roughness and part accuracy to select the most suitable
axial increment. In addition, it is quite hard to predict the sheet
metal behaviour during the process.
2.3. Synthesis
According to the experimental results (Table 2), the feedrate has
no significant influencing effect on process implementation, ex-
cept the forming times. In contrast, the tool path shape and the
axial increment have a great influence on the process implemen-
tation. However, coupled effects are also observed, so that process
implementation depends on a proper combination of all tool path
parameters.
It is also important to note that repeatability tests were
conducted in addition to the design of the experiment presented
here. They showed a dispersion lower than 5%, which attests the
validity of the conclusions proposed in this section.
As a consequence, CAM-generated tool paths cannot be directly
used; they need to be adjusted to ISF applications. It is a fact
that most of the studies reported use such strategies and obtain
interesting results. But any productivity or accuracy improvement
needs to adapt these CAM milling orientated tool paths according
to the stakes conveyed by the forming process.
3. Intelligent CAM programming tool path generation
In ISF, it is difficult to predict the sheet metal position after
tool release because of the elastic springback. Forming tool path
programming is consequently quite complicated because the
2 2.5 3 3.5 mm
Observed
profile
position of the sheet metal after the tool path cannot be known
with a high precision. Moreover, in TPIF, if the tool paths are not
well adapted, sheet metal can be gripped between the die and the
tool; this can not only affect part quality and durability but also
damage the forming equipment. As a consequence, CAM milling
strategies are not suitable anymore; it is necessary to control and
adjust the forming tool paths according to real-time machine tool
performance feedback.
In a previous work, a new method to program and control
NC machining processes by integrating process constraints was
proposed [26]. Using this approach, tool path generation is shared
by both the CAM software and the CNC. The basic tool paths
are generated by the CAM software and sent to the CNC. Then,
during the running of the program, these tool paths are optimized
by the CNC controllers according to the real-time process data
evaluation. Another aspect of this approach is to limit the use of
additional equipment neither to collect the process data nor to
perform adaptative tool path control. On the contrary, the purpose
is to access these data using the CNC controller of the machine tool.
In this study, an ICAM approach is implemented for ISF
applications. As shown by Fig. 6, the proposed method is divided
into two stages. The first stage consists in evaluating process
data (such as tool forces or probing) directly from CNC data. In
the second stage, the tool motions are controlled not only by
CAM-generated tool paths but also by the process data estimation
obtained in real time in the first stage.
Two types of ICAM approach implementations are proposed
here. The first implementation is based on tool load evaluation to
adapt the programming parameters. The second implementation,
which has been validated within an industrial environment, uses
probing as process data to optimize the tool path programming.
3.1. Process data evaluation
The objective of ICAM approach is not to perform process mon-
itoring operations, but to dispose of an assistance tool dedicated
to tool path adaptation during the process implementation. It is
nevertheless necessary to evaluate process data which provide a
good image of the process implementation. For ISF, forming loads
are suitable, as shown by Duflou in [27]. Filice et al. have proposed
an interesting method for SPIF [28]. However, this approach is
based on 3D component dynamometer acquisitions and prediction
models; these two aspects limit its portability and its flexibility:
dynamometer statements are only valid under precise working
 M. Rauch et al. / Computer-Aided Design 41 (2009) 877–885
Theoretical
programming
parameters
Tpp
CAD CAM
CAD Model
Programming
parameters optimization
Accuracy
Fig. 6. From CAD to real part with the ICAM approach.
Fig. 7. Tool loads evaluation process.
conditions (part and clamping sizes, etc.) and prediction models
need several experimental validations.
In contrast, the proposed method consists in evaluating the
tool loads directly from the CNC data, without any additional
equipment. Indeed, recent CNC controllers let the user access sev-
eral process parameters such as tool centre positions, servomotor
torques or power amperage during the running of the NC program.
It is possible to compute these data using the available calculation
capability of the CNC controller or using an additional computer
(connected to the CNC as a memory and calculation extension).
As shown in Fig. 7, forming load evaluation is divided into three
steps. Step 1 consists of catching the servomotor torques Tkinematics
generated by the displacement of the machine tool axis (without
any process load): the NC program is executed without the sheet
metal and the torque values are acquired using CNC data. Step 2
is performed during the forming process of the part: servomotor
torques and associated forming tool positions are captured using
CNC data during the process. The measured values Ttotal contain
both forming and dynamics contributions. Step 3 is carried out
simultaneously with Step 2. Using the torque values Tkinematics from
Step 1 and Ttotal from Step 2, forming torques Tforming are calculated.
As a result, for each position of the forming tool, the forming
torques on each carriage are known. From these torques, it is
possible to estimate the load applied to each carriage thanks to the
servomotor’s technical properties. Finally, the forming load applied
on the tool is estimated using these data, the tool position data
and a mechanical equilibrium of the machine tool structure. At
881
ERRORS
CNC
Optimized
Manufactured
programming
parameters Part
Opp
Forming
process
Process Data Evaluation
(tool loads, probing, …)
this stage, the geometrical transformation model of the selected
machine tool is needed.
As a result, it is possible to estimate the instantaneous forming
forces from NC data and the geometrical transformation model
of the machine tool. Step 2 and Step 3 are performed during
the forming process, so it is possible to optimize the tool paths
‘‘on-line’’. Furthermore, the purpose is only to have an evaluation
of these data, not highly accurate statements. CNC controller
capabilities (clock frequency, etc.) are sufficient to do these
evaluations.
Other process parameters can be evaluated as well. For exam-
ple, strain paths and fracture limits are also useful parameters to
optimize the forming tool paths. In the same way, probe measures
will be employed to carry out the industrial implementation of TPIF
proposed in the last section of this paper.
3.2. Forming tool path control
The second stage of the ICAM method consists in adapting the
tool paths according to the results of forming force estimation.
In ISF, the major objective is to prevent any sheet metal damage
caused by a too high forming force. As a result, the developed tool
path compensation aims at clearing the tool when the forming
force overtakes a preset value. Two implementation procedures
were identified: to modify the forming tool jog or to call a routine
that clears the tool.
882 M. Rauch et al. / Computer-Aided Design 41 (2009) 877–885

a linear modification of the tool height between two successive

control points of the tool paths (Fig. 8).
A difficulty conveyed by this tool path compensation lies in its
Adapted Toolpath duration. The final accuracy can be affected by tool jog variations if
several tool path adaptations are needed during the process. This
tool path adapting technique is consequently rather dedicated to
TPIF where a too high forming force is often caused by a wrong
position of the forming support.
Another procedure to prevent the sheet metal from damaging
is to call a tool clearance routine. This routine generates a retract
movement along the tool axis, so that it is possible to control the
tool load even in five-axis strategies. From a practical point of
Initial Toolpath
view, as soon as the tool load estimation overtakes a preset value,
the forming NC program calls the clearance subroutine (Fig. 9).
Fig. 8. Tool path adaptation by modifying the tool jog. The retraction value is defined by the user before running the NC
program. As the tool path compensation is only local, this approach
is efficient for local tool overloads. This solution is also suited to
SPIF.
Other techniques can be developed to adapt the forming tool
paths to the process data evaluation. For example, another method
Adapted Toolpath has been developed for the industrial application presented in
the next section of this paper. It consists in stretching the initial
tool path shape according to the probe measures between two
transitional shapes. Thus, the forming increment can be controlled
and the tool motions can follow the actual part profile.
From a general point of view, although various process data can
be evaluated and although various tool path compensations can
be proposed, the major point is to make both stages of the ICAM
method coherent.
Initial Toolpath

3.3. Validation study and industrial implementation

Fig. 9. Tool path modification by using a clearance routine.
For the first procedure, the tool jog parameters are modified in
‘‘real time’’ into the CNC database. If the forming load estimation
asks for tool path adaptation, the tool length parameter is modified
into the tool jog database. This modification leads in practice to
Guide j ( j [1;3] )
L3
This section introduces two different implementations of the
ICAM concept for ISF applications. First, an experimental study
is presented. Its objectives are to validate the first stage of the
ICAM approach. As a result, no tool path compensation is proposed
for this first study. The process data evaluated are the tool loads
during the forming process. Second, an industrial implementation

Ai
L2
L1

Vj

ZT
Bi
R T OT
za i

XT YT
O Y
X
RO
L

CO
D

C
B ZM Tilting axis (B)
OC
d

RM

OM
YM
XM Upper face of
the table

Fig. 10. Geometry of the parallel kinematics module of the VERNE machine tool.
 M. Rauch et al. / Computer-Aided Design 41 (2009) 877–885 883

700
Forming force (N) 600
500
Dynamometer
400
300
200 Sheet Metal
100 CNC
0
0 1000 2000 3000 4000 5000 6000
Time (ms) 3D dynamometer

Fig. 11. Comparison of CNC evaluated tool loads and dynamometer statements for the first experiment.

Fig. 12. Comparison of CNC evaluated tool loads and dynamometer statements for the second experiment.

of the ICAM concept is detailed. In this case, the process data

evaluated are on machine inspection results. Then, according to
these evaluations, the forming tool paths are adapted.
The experimental study was performed on the five-axis parallel
kinematics machine of the laboratory. Its name is VERNE, and it
is equipped with a Siemens Sinumerik 840D NC controller. The
servomotor torques in the axis carriages were acquired by using
the CNC data. The purpose of using a parallel kinematics machine
tool is to prove the efficiency of the ICAM approach with atypical
machine architectures.
The main characteristic of VERNE is that it combines a serial
module, a tilting table, and a parallel module, made of six struts
with constant length (Fig. 10). These struts join the spindle to three
carriages, with two struts linked to each carriage. The NC controller
commands the position of the carriages and hence controls the
position of the tool centre point. The use of a geometrical trans- Fig. 13. Work holder for the industrial application.
formation algorithm makes this stage transparent for the user, so
that tool paths can be programmed thanks to the CAM software The second experiment focused on SPIF as well. It consisted of
as for any other machine tool. However, this transformation algo- carrying out a 100 mm diameter tool path using a 2 mm increment
rithm had to be taken into account during the implementation of on 0.6 mm sheet metal. The forming forces are calculated from
the ICAM concept in order to match the acquired real-time data servomotor acquisitions according to the ICAM approach. The
with the Cartesian tool positions. results are shown in Fig. 12. They show a good concordance
The capability of this parallel kinematics machine tool has al- with Kistler dynamometer statements obtained in the same
ready been studied [29,30]. The stiffness of its structure is adequate experimental conditions.
to carry out milling and forming applications. The force peaks are
As a result, both validation tests highlighted the efficiency of
not compensated by any deformation of the machine structure. So,
the ICAM approach. It is possible to evaluate the forming loads
this machine tool is as efficient as a serial machine tool to imple-
without a dynamometer. The objective is not to perform process
ment ISF applications.
monitoring but to get forming force evaluations with the view to
Two experiments were conducted to carry out this study.
During the first experiment, the forming punch was animated with adapting the tool paths. Thus, during these experiments no tool
a linear axial motion to progressively load the sheet metal. The path compensation is performed.
resulting tool loads were acquired from CNC data and compared After these validations tests, a complete implementation of the
to the statements of a Kistler 3D dynamometer. However, the ICAM approach was carried out within an industrial environment.
dynamometer was only employed to provide a reference curve. The case consisted in manufacturing aeronautic parts made of
The results of the tool load evaluations are given in Fig. 11. They aluminium alloy. Their final shape is almost hemispherical and
are very satisfying: as soon as the tool load surpasses 50 N, there the work pieces are sheet metals. These parts were manufactured
is a good correlation between both curves. As a result, tool load until now by hammering. The use of the ICAM approach led the
evaluation using CNC data is possible by using the ICAM approach. industrial concern to shift to the ISF process.
884 M. Rauch et al. / Computer-Aided Design 41 (2009) 877–885
Fig. 14. Developed approach for the industrial application.
As shown in Fig. 13, the sheet metal is clamped at the top
of a dedicated counter form. The forming tool paths make the
sheet metal adjust to this hemispherical shape. A major functional
requirement is to prevent any thickness variation of the sheet
metal during the manufacturing process. Usual forming tool paths
such as spiral or parallel contours have to be avoided because they
form sheet metal with a variable thickness. In contrast, high level
tool paths are needed to guide the material flows and reproduce
the deformation phenomenon associated to handwork.
The forming tool path generation therefore becomes a very
complex issue because the effects of elastic springback have to be
taken into account too.
The ICAM approach was implemented as explained by the
following.
– The evaluated process data were probe measures. A 3D touch
probe was mounted into the tool holder of the machine so that
on-machine inspections could be carried out. The geometry of
the formed part was acquired between two successive forming
tool paths. The results of the on-machine part inspection were
used to rebuild the 3D model of the part and to generate the
next tool paths.
– The basic tool paths were generated by CAM software. Their
initial shape enabled the part to be formed without any
thickness variation. Then, tool path ‘‘morphing’’ was done: the
tool paths were stretched to adapt to the geometry of the part
generated using the process data.
Moreover, a step-by-step approach based on transitional shapes
was developed, as proposed in Fig. 14. It was impossible to obtain
the final part with one single tool path because the deformation
rate of the sheet metal was high.
The generated tool paths deformed the sheet metal from the
current transitional shape to the next one. The tool path shape was
designed to make the material flow rise towards the top of the part
during the forming process (as for handwork). Indeed, the purpose
was to stay in the compressive domain. The chosen interpolation
format was B-splines in order to adapt to any transitional shape
and to ease the morphing process, which was carried out by
moving the control points. Actually, tool path compensation made
these tool paths adapt to the real geometry of these transitional
shapes. As a result, the first tool path patterns were strongly
modified at some stages to stick to the real position of the sheet
metal. The springback effect was corrected by this method as well.
As a result, this industrial application demonstrated the efficiency
of the ICAM approach, although its implementation was based
on empirical results. The produced part presented an acceptable
accuracy and the forming times met the industrial requirements.
Furthermore, the capability to produce accurate sheet metal parts
using the compression phenomenon in TPIF was made clear.
This implementation of the ICAM approach was based on pro-
cess data evaluations which were geometrical probing. Tool path
programming compensation focused on the NC parameterization
and on the tool path shapes by using the tool path ‘‘morphing’’
concept.
However, the ICAM method can be successfully carried out for
other ISF applications under the conditions that both process data
evaluation and tool path programming compensation are relevant
in regard to the specificities of the equipment and of the part.
4. Conclusion
Tool path generation is a key topic in incremental sheet forming.
It is essential to develop dedicated tool paths to improve the effi-
ciency of this process. In this work, an experimental study focused
on the effects of the forming strategy and other programming re-
lated parameters on the produced part accuracy and on the process
implementation. The results showed that basic CAM tool paths are
not suitable to carry out ISF applications.
Hence, the implementation of Intelligent CAM programmed
tool paths for incremental sheet forming is proposed. This new ap-
proach adapts the tool paths during the manufacturing of a part ac-
cording to process data evaluations. Tool path adaptation is carried
out by using CNC data. This ensures a high flexibility. A validation
study was achieved in a conclusive way. Finally, an implementa-
tion of the ICAM approach was presented within an industrial en-
vironment. The capability for TPIF to manufacture accurate parts
that have a constant thickness was shown.
Acknowledgements
This work was carried out within the context of the working
group Manufacturing 21 which involves 18 French research labora-
tories. The topics covered are: the modelling of the manufacturing
process, virtual machining, and the emergence of new manufactur-
ing methods.
References
[1] Hagan E, Jeswiet J. A review of conventional and modern single-point
sheet metal forming methods. Proceedings of the Institution of Mechanical
Engineers, Part B: Journal of Engineering Manufacture 2003;217(2):213–25.
[2] Jeswiet J, Micari F, Hirt G, Bramley A, Duflou J, Allwood J. Asymmetric single
point incremental forming of sheet metal. CIRP Annals — Manufacturing
Technology 2005;54(2):623–49.
[3] Shim M-S, Park J-J. The formability of aluminium sheet in incremental
forming. Journal of Materials Processing Technology 2001;113(1–3):654–8.
[4] Park J-J, Kim Y-H. Fundamental studies on the incremental sheet metal form-
ing technique. Journal of Materials Processing Technology 2003;140(1–3):
447–53.
[5] Kim YH, Park JJ. Effect of process parameters on formability in incremental
forming of sheet metal. Journal of Materials Processing Technology 2002;
130–131:42–6.
[6] Murata A. A new trial on incremental forming of sheet metal parts. R&D
Review of Toyota CRDL 1999;34(3).
[7] Ceretti E, Giardini C, Attanasio A. Experimental and simulative results in
sheet incremental forming on CNC machines. Journal of Materials Processing
Technology 2004;152(2):176–84.
[8] Kopac J, Kampus Z. Incremental sheet metal forming on CNC milling machine-
tool. Journal of Materials Processing Technology 2005;162–163:622–8.
[9] Attanasio A, Ceretti E, Giardini C. Optimization of tool path in two points
incremental forming. Journal of Materials Processing Technology 2006;
177(1–3):409–12.
[10] Mao F, Mo JH, Huang SH. Study on instability of the point bolster sheet metal
dieless forming. Journal of Materials Processing Technology 2006;176(1–3):
13–8.
 M. Rauch et al. / Computer-Aided Design 41 (2009) 877–885
[11] Ambrogio G, Costantino I, De Napoli L, Filice L, Fratini L, Muzzupappa M.
Influence of some relevant process parameters on the dimensional accuracy
in incremental forming: A numerical and experimental investigation. Journal
of Materials Processing Technology 2004;153–154:501–7.
[12] Ambrogio G, De Napoli L, Filice L, Gagliardi F, Muzzupappa M. Applica-
tion of incremental forming process for high customised medical product
manufacturing. Journal of Materials Processing Technology 2005;162–163:
156–62.
[13] Hirt G, Ames J, Bambach M, Kopp R. Forming strategies and process modelling
for CNC incremental sheet forming. CIRP Annals — Manufacturing Technology
2004;53(1):203–6.
[14] Duflou JR, Verbert J, Belkassem B, Gu J, Sol H, Henrard C, et al. Process win-
dow enhancement for single point incremental forming through multi-step
toolpaths. CIRP Annals — Manufacturing Technology 2008;57(1):253–6.
[15] Skjoedt M, Bay N, Endelt B, Ingarao G. Multi stage strategies for single point
incremental forming of a cup. In: ESAFORM. 2008.
[16] Micari F, Ambrogio G, Filice L. Shape and dimensional accuracy in single point
incremental forming: State of the art and future trends. Journal of Materials
Processing Technology 2007;191(1–3):390–5.
[17] Held M. Voronoi diagrams and offset curves of curvilinear polygons.
Computer-Aided Design 1998;30(4):287–300.
[18] Tang Y. Optimization strategy in end milling process for high speed machining
of hardened die/mold steel. Journal of University of Science and Technology
Beijing, Mineral, Metallurgy, Material 2006;13(3):240–3.
[19] Pateloup V, Duc E, Ray P. Corner optimization for pocket machining. In-
ternational Journal of Machine Tools and Manufacture 2004;44(12–13):
1343–53.
[20] Hascoet JY, Rauch M. Trochoïdal milling and plunging strategies in rough
pocket milling of aluminium alloys. In: International conference on high
speed milling, 2. 2006. p. 649–61.
885
[21] Dugas A, Lee JJ, Hascoet JY. Feed rate and tracking errors simulation in high
speed milling. In: 4th international ESAFORM conference on material forming.
2001. p. 667–70.
[22] Martins PAF, Bay N, Skjoedt M, Silva MB. Theory of single point incremental
forming. CIRP Annals — Manufacturing Technology 2008;57(1):247–52.
[23] Silva MB, Skjoedt M, Martins PAF, Bay N. Revisiting the fundamentals of single
point incremental forming by means of membrane analysis. International
Journal of Machine Tools and Manufacture 2008;48(1):73–83.
[24] Ham M, Jeswiet J. Single point incremental forming and the forming criteria
for AA3003. Annals of CIRP 2006;55(1):241–4.
[25] Hagan E, Jeswiet J. Analysis of surface roughness for parts formed by computer
numerical controlled incremental forming. Proceedings of the Institution of
Mechanical Engineers, Part B: Journal of Engineering Manufacture 2004;
218(10):1307–12.
[26] Hascoet JY, Rauch M. A new approach of the tool path generation in manufac-
turing operations using CNC data. In: International conference on high speed
milling. 2006. p. 101–13.
[27] Duflou J, Tunckol Y, Szekeres A, Vanherck P. Experimental study on force
measurements for single point incremental forming. Journal of Materials
Processing Technology 2007;189(1–3):65–72.
[28] Filice L, Ambrogio G, Micari F. On-line control of single point incremental
forming operations through punch force monitoring. Annals of CIRP 2006;
55(1):245–8.
[29] Terrier M, Dugas A, Hascoet JY. Qualification of parallel kinematics machines
in high-speed milling on free form surfaces. International Journal of Machine
Tools and Manufacture 2004;44(7–8):865–77.
[30] Terrier M, Gimenez M, Hascoet JY. VERNE — A five-axis parallel kinematics
milling machine. Proceedings of the Institution of Mechanical Engineers, Part
B: Journal of Engineering Manufacture 2005;219(3):327–36.