Dry cutting study of an aluminium alloy (A2024-T351): a numerical and

experimental approach
M. Asad, F. Girardin*, T. Mabrouki, J.-F. Rigal
LaMCoS, INSA-Lyon, CNRS UMR5259, F69621, France
*
Corresponding Author: Francois.Girardin@insa-lyon.fr

Abstract

In the present contribution, experimental and numerical methodologies concerning orthogonal cutting are
proposed in order to study the dry cutting of an aeronautic aluminium alloy (A2024-T351). The global aim
concerns the comprehension of physical phenomena accompanying chip formation with respect to cutting
speed, such as chip segmentation and fragmentation. For experimental validation, series of tests are carried
out concerning geometrical analysis of the chip; video sequences of chip formation with a high-speed camera,
and high-frequency sampling measurements of the cutting force signal are realised. For the numerical
approach, the material and its ductile shear failure behaviour are based on the Johnson-Cook laws. The
material failure model exploited considers both damage evolution and energy coupling. Numerical results
concerning cutting force and segmentation frequency are compared to experimental ones. Moreover ,an
analysis of damage distributions is presented.

Keywords: Serrated chip formation, Orthogonal cutting simulation, Damage, Experimentation, A2024-T351

1 INTRODUCTION For experimental exploration, series of tests are

Today, in automotive and aeronautic industries it is
noted an increased request for the use of materials,
which have high strength to weight ratio, for
example aluminium alloys. These materials are often
subjected to machining operations where the
criterion of minimization of lubricant use is more
and more of topicality. However, in dry (said also
green) machining like the cutting of aluminium
alloys, parameters of working are not yet optimised.
This is mainly due to a lack of the physical
phenomena comprehension accompanying the
cutting operations.
In this framework, the present work point out multi-
physical phenomena accompanying the cutting of
the aeronautic aluminium alloy A2024-T351. The
present study, deals with both a numerical approach
to simulate an orthogonal cutting operation by
exploiting the capabilities of ABAQUS/Explicit
software and an experimental methodology for
validation.
carried out concerning geometrical analysis and
measurements at high frequency sampling of cutting
force signal. From the numerical point of view, a
presentation of an Abaqus/Explicit methodology
explains an optimised approach based on the
coupling between damage and fracture energy for
building an orthogonal cutting model with realistic
chip formation.
2 EXPERIMENTAL APPROACH
During this work, a classical turning operation was
exploited. Two working parameters were
considered: the cutting speed Vc = 200-400-800
m/min and the feed rate f = 0.3-0.4-0.5 mm/rev, with
a constant cutting depth ap = 4 mm. In the following,
after devices description, it is proposed to study the
formation of the chip thanks to different methods:
photo and video captures, and force measurements.
2.1 Experimental device Finally, cutting force signal was sampled at 45 kHz.
Force average value was calculated to obtain the
The cutting tool used is composed of an uncoated reference level for numerical simulation comparison.
carbide insert (CCGX 12 04 08-AL H10 fixed on The frequency spectrum was also computed.
tool-holder SCLCR 2020K 12). In order to reproduce Because no correcting method was employed, such
orthogonal cutting case, the insert cutting edge is as accelerometric compensation [2] or frequency
orthogonal with the feed rate and cutting speed (Kr = response function [3], the pass-band was limited to
90° and λs = 0°). Also, workpiece was prepared with 1000 Hz (a third of the first dynamometer natural
concentric cylindrical grooves (figure 1). frequency). By investigating frequency spectrum,
fragmentation frequency was localised.
m
50 4m
N Φ1 ap = Cutting
Radius face Chip 2.3 Results
20µm breaker
0.7mm
B B
17.5° Table 1 gives the experimental results concerning
Vf
7° the evolution of cutting force (Fc), fragmentation
Tool-holder Insert
Section B-B frequency and segmentation one according to cutting
speed and feed rate variations.
Fig. 1. Workpiece preparation
Table 1. Experimental results
The measuring equipment is composed of a standard
Fc(N) - Fragmentation (Hz) / Segmentation (kHz)
dynamometer (Kistler 9257B) and a high frequency
f Vc (m/min)
data acquisition device (National Instrument NI (mm) 200 400 800
4472). Data treatment was developed with Matlab 778 N 769 N 769 N
software. Videos were performed with a high-speed 0.3
128Hz / - 290Hz / 37.8kHz 500Hz / 90.7kHz
camera (MotionScope 8000 Redlake). 988 N 978 N 976 N
0.4
120Hz / 10.3kHz 351Hz / 32.4kHz 889Hz / 64.8kHz
2.2 Analysis methodology
1216 N 1196 N 1192 N
0.5
256Hz / 16.2kHz 476Hz / 22.7kHz 1026Hz / 45.3kHz
At first, chip morphologies were photographed using
a microscope and saw-tooth shapes (segmentation)
Those results are reference points for a numerical
can be recognized on chips (figure 2).
approach that aims at building a model which first
a) b)
reproduces experimental tendencies, such as chip
geometry or frequencies evolution, and second,
numerical values, particularly for segmentation
frequency.

Fig. 2. Chip morphology for f=0.4 mm/rev.
a) VC=200 m/min, b) VC = 800 m/min
Supposing incompressible material [1],
segmentation and fragmentation frequencies were
calculated starting from number of saw-teeth, chip
thickness, chip length, feed rate and cutting speed.
Secondly, video sequences were performed. Video
acquisition frequency (4 kHz) was too slow to
observe chip segmentation. Nevertheless, chip
fragmentation frequency can be analysed. It was
determined by numbering chip fracture on the video
and calculating the mean value with corresponding
time. The results confirm the first evaluation from
chip morphology analysis.
3 NUMERICAL APPROACH
3.1 Geometrical model and hypothesis
To improve physical comprehension, the capabilities
of Abaqus/Explicit have been exploited. A 2-D
orthogonal cutting model with plane strain
assumption was considered. Quadrilateral continuum
elements were used for a coupled temperature-
displacement calculation. Interactions between
contacting bodies are defined with a Coulomb’s
friction law [1,4]. To optimize the management of
the contact between chip and cutting tool, a four-part
model was developed (figure 3). Tool section shape
is similar to that used in experimentation (figure 1).
 Chip formation occurs in two steps following the
Damaged zones:

Part 2 Part 1
Part 2 + Part 3

ductile failure phenomenon. At the beginning, the

Part 3 damage in element is initiated when the scalar
20 µm damage parameter ω exceeds 1. This parameter is
Part 4 based on a cumulative law defined as:
∑ ∆ε
f
1.2 mm

ω= (3)
ε 0i
No displacement
Y along Y direction
12 mm
where ∆ε is increment of equivalent plastic strain.
X Fixed boundaries
In a second step, the damage evolution is specified
Fig. 3. Grid model and boundary conditions in terms of fracture energy dissipation with a stress-
displacement response thanks to Hillerborg's fracture
3.2 Material behaviour and chip formation energy proposal [8]. The fracture energy Gf used as
criterion material parameter is given by equation (4). It takes
into account the characteristic length L of element to
Johnson-Cook material model (equation (1)) is used alleviate mesh dependency of the results.
for cutting simulation [5], and is associated with
εf uf
Johnson-Cook shear failure model (equation (2)) G f = ∫ Lσ y d ε = ∫ σ y du (4)
which corresponds to the damage initiation criterion. ε0i 0

 ε&    T − Troom  
( )
m
 Damage evolution law can be expressed in linear
σ =1 A4+2 B4  n 
ε3 1 + C ln   1 − 
   (1)
ε
&  T − T   (equation (5)) or exponential form (equation (6))

 4424
Elasto− plastic 1  430    melt room  
1 44 4
4 2 4 4 4 4
3
Lε
term
Viscosity Softening u
term term D= = (5)
uf uf
 P    ε&     T − Troom   (2)
ε 0i =  D1 + D2 exp  D3   × 1 + D4 ln    1 + D5    u σ 
  σ    ε
&0
    Tmelt − Troom   D = 1 − exp  − ∫ du  (6)
 0 G 
where σ is the equivalent plastic flow stress, ε is the  f 
equivalent plastic strain, ε& is the plastic strain rate, where the equivalent plastic displacement at
-1
ε&0 is the reference strain rate (1.0 s ) and ε&0 i is the failure, u f , is computed as:
plastic strain at damage initiation. Coefficients of
laws are given table 2. 2G f
uf = (7)
σy
Table 2. Johnson-Cook parameter values for A2024T351 [6]
A (MPa) B (MPa) n C m where σy is the material yield stress.
352 440 0.42 0.0083 1
D1 D2 D3 D4 D5
0.13 0.13 -1.5 0.011 0 4 RESULTS AND DISCUSSION

The properties of workpiece and cutting tool are The present section deals with the mechanisms
mentioned in table 3. governing the chip tooth shape genesis process. The
parametric studies considered are VC = 200, 400, 800
Table 3: Workpiece and tool physical parameters [7] m/min and f = 0.4 mm/rev. Globally experimental
Physical parameter Workpiece (A2024-T351) Tool and numerical results such as the values of
Density, ρ (Kg/m3) 2700 11900
segmentation frequency and cutting force are in
Elastic modulus, E (GPa) 73 534
Poisson’s ratio, ν 0.33 0.22
good comparison as shown in table 4.
Specific heat, Cp (Jkg-1°C-1) CP = 0.557 T + 877.6 400
Table 4. Numerical results and deviation with experiment
Thermal conductivity, 25<T<300: λ = 0.247T + 114.4 50 Vc (m/min) 200 400 800
λ (W m-1C-1) 300<T< Tmelt : λ = −0.125T + 226.0
Fc (N) 898 N – 4% 994 N – 6% 901 N – 8%
Expansion,αd (µm.m-1°C-1) α d = 8.910−3 T + 22.2 ×
Segmentation
Non available 31.4 kHz – 3% 66.3 kHz – 2%
Tmelt, (°C) 520 × (kHz)
Troom, (°C) 25 25
Figure 4 shows the distribution of damage D for a
complete chip formation. On figure 4-c), it is
remarked that this damage is localised mainly in the
shearing zones and in the locations corresponding to
tool/workpiece interaction. In order to bring more
comprehension of chip genesis, the attention is
focused on the steps characterising the formation of
one chip segment.
At the beginning of chip formation, no segmentation
was observed (figure 4-a)). As soon as the chip is in
contact with the free surface, the segmentation
phenomenon is initiated (Fig. 4-b)). This contact will
increase chip curvature radius and also shear stresses
in the shearing primary zone, that causes the
beginning of segmentation. After a certain cutting
time, the curled-up chip is subjected to high loads:
those due to self-contact on chip free surface and
those caused by the tool advancement. These will
cause a bending state on the chip and increase
locally chip damage (figure 4-d)). In comparison
with video capture (figure 4-e), that may induce the
chip fragmentation.
As shown by previous published papers [1], the
segmentation is the result of a softening state during
tool/workpiece interaction, but it can be favoured by
other phenomena like pre-existing micro-cracks,
machine tool vibrations or chip self-contact during
its curl-up as explained previously. We can noticed
that chip breaker have an important role in chip
curvature, and consequently in segmentation and
fragmentation.
a) b) c) D (Avg 75%)
tc =675µs tc =864µs t c=1287µs
Possible fracture Real fracture
Rake face
ap Cutting edge
Flank face
d) tc =2889µs
e)
Fig. 4. Chip morphology evolution regarding to damage
variable D (SDEG) and experimental comparison
VC = 200 m/min and f = 0.4 mm/rev
5 CONCLUSIONS
The global aim of this contribution concerns the
comprehension physical phenomena accompanying
the chip formation according to the variation of feed
rate and cutting velocity for an aluminium alloy
referenced as A2024-T351.
Two approaches, experimental and numerical are
proposed. The experimental one focuses on physical
quantity such as periodicity of chip segmentation
and fragmentation. A methodology measurement of
these quantities in term of frequency is proposed
based on the cutting force signal analysis. From the
numerical point of view, presentation of Abaqus
/Explicit based methodology explains an original
optimised approach for building an orthogonal
cutting model with realistic chip formation. This
model is based on a coupling between damage and
fracture energy. The numerical results show the
effect of the chip curl-up and its self contact on the
initiation of segmentation. This remark was
confirmed by high sampling videos.
Eventually, from a numerical point of view, the
combination of the fracture energy coupling model
to the Johnson-Cook law gives good results
compared to the experimental ones. Values of
fracture toughness Kic are often given in material
databases and they can be exploited to contribute to
the elaboration of a predictive cutting model.
REFERENCES
1. S. Belhadi, T. Mabrouki, J.-F. Rigal, L. Boulanouar,
Experimental and numerical study of chip formation
during a straight turning of hardened AISI 4340 steel.
Journal of Engineering Manufacture, IMechE, PEP,
London, Vol. 219, 2005, pp. 515-524
2. F. Lapujoulade, G. Coffignal, J. Pimont. Cutting forces
evaluation during high speed milling, 2ème Conf. Int.
IDMME’98. Compiègne – France. 27-29 May 1998.
3. L. R. Castro, P. Viéville, P. Lipinski. Correction of
dynamic effects on force measurements made with
piezoelectric dynamometers, International Journal of
Machine Tools & Manufacture 46 (2006) 1707–1715.
4. K. Li, X.-L. Gao, J.W. Sutherland, Finite element
simulation of the orthogonal metal cutting process for
qualitative understanding of the effects of crater wear on
the chip formation process, J. Mater. Proc. Techn. 127/3
(2002) 309–324.
5. G.R. Johnson, W.H. Cook, Fracture characteristics of
three metals subjected to various strains, strain rates,
temperatures and pressures, Eng. Fract. Mech. 21/1
(1985) 31–48.
6. X. Teng, T. Wierzbicki, Evaluation of six fracture
models in high velocity perforation Engineering
Fracture Mechanics 73 (2006) 1653–1678.
7. www.knovel.com
8. A. Hillerborg, M. Modeer, and P. E. Petersson, Analysis
of Crack Formation and Crack Growth in Concrete by
Means of Fracture Mechanics and Finite Elements,
Cement and Concrete Research, vol. 6, pp. 773–782,
1976.
Criterion to evaluate diffusive wear in 3D simulations when turning AISI
1045 steel
A. Attanasio1, E. Ceretti1, C. Giardini2, L. Filice3, D. Umbrello3
1
University of Brescia, Dept. of Mechanical and Industrial Engineering, 38 via Branze, 25123 Brescia, Italy
URL: www.ing.unibs.it/tecmec e-mail: aldo.attanasio@unibs.it; elisabetta.ceretti@unibs.it
2
University of Bergamo, Dept. of Design and Technologies, 5 Via Marconi – 20044 Dalmine (BG), Italy
URL: www.unibg.it e-mail: claudio.giardini@unibg.it
3
University of Calabria, Dept. of Mechanical Engineering, 44/C Ponte Pietro Bucci – 87036 Rende (CS) Italy
URL: tsl.unical.it e-mail: l.filice@unical.it; d.umbrello@unical.it

ABSTRACT: This paper deals with the simulation of cutting processes. In particular, 3D FE simulations were
carried out in order to estimate tool wear development during turning operations. An analytical model taking
into account diffusive wear mechanism has been implemented in a proper software subroutine for the tool
wear evaluation. Preliminarily conducted experimental tests allowed the analytical expression coefficients
evaluation. The adopted wear model gave the possibility of correctly defining the tool wear during both the
(initial) transient phase and the steady-state of the analyzed turning operation. A comparison between
experimental and simulative results showed a general good agreement. The final goal of the research is to
identify a FE based strategy to be reliably used to forecast tool geometry and cutting parameters influences on
tool wear.

Keywords: cutting operation, 3D FEM simulation, tool wear, analytical model.

software could help in the prediction of the crater

1 INTRODUCTION
Tool wear is widely considered as one of the most
important aspects causing poor quality of worked
piece in cutting. Consequently tool wear prediction
and tool substitution policy are regarded as very
important tasks in order to maximize tool
performance and minimize cutting costs. Many
researches have been carried out on tool wear whose
results are reported in a wide number of scientific
papers [1-3]. In these papers different wear
analytical descriptions, according to the different
wear mechanisms, are considered referring to
empirical methods or experimental studies. The
different mechanisms affecting the tool wear are:
material adhesion, abrasion, erosion, diffusion,
corrosion and fracture. Research papers have shown
how these are generally present in combination,
even if only one or few of them result to be
dominant as the cutting conditions, and consequently
contact pressure and temperature, and the workpiece
and tool materials change.
Even if, the implementation of such models in a FE
and flank tool wear, only few research papers are
available on this topic and, what is more, are mainly
focused on 2D problems. Of course, 3D simulations
would be more interesting since they permit to study
all the occurring complex phenomena without heavy
geometrical simplifications.
Thanks to the increased hardware and software
efficiency, now 3D models can be effectively used
to simulate actual machining processes despite of the
calculation time (almost 10 times the 2D one).
This paper reports the results of a preliminary study
on the possibility of simulating non-orthogonal
cutting operations considering the tool wear too, by
means of 3D FE models, with the aim to supply
results more closer to the industrial needs.
A diffusive wear model implemented in the 3D FE
software has been defined starting from a series of
bi-dimensional Arbitrarian Lagrangian Eulerian
(ALE) simulations coupled with orthogonal cutting
experimental tests. The final validation of the whole
simulation strategy was carried out through a series
of three dimensional experimental tests.
2 TOOL WEAR MODEL 3 3D EXPERIMENTAL TESTS

According to technical literature, several wear Cylindrical bars with a diameter of 100 mm were cut
mechanisms can be defined. Among them, the using an uncoated ISO P40 tool with tool nose
Takeyama and Murata model [3] is able to properly radius of 0.8mm, rake angle γ=0°, clearance angle
take into account abrasion and diffusion wear. As far α=6° and inclination angle λ=7°. Furthermore, the
as mechanical abrasion is concerned, it is tool was positioned into the holder in order to obtain
proportional to cutting distance and strictly related to the angles χ and χ’ of 90° and 30°, respectively.
hardness and distribution of the abrasive particles. Several levels were selected for each parameter,
Whereas the diffusion term is a physic-chemical namely 150, 160 and 190 m/min (cutting speed),
contribution associated with temperature. 0.17, 0.18 and 0.25 mm/rev (feed rate), while the
Although wear model depends upon the cutting depth of cut was fixed to 1.5 mm. The operations
conditions, the diffusive term is the predominant one were conducted in dry conditions and both flank and
for uncoated tools under standard working crater wear were measured at different cutting times,
conditions. In their study Takeyama and Murata [3] using an optical microscope (50X) equipped with a
demonstrated that abrasion rate of a tool is almost motorized faceplate. Table 1 reports tool wear data
independent on temperature, if the latter is lower for the 3D experiments.
than the critical temperature of the tool material.
Therefore at high temperatures the modified Table 1. Measured wear data.
Takeyama and Murata model is suitable: Test VC f Time KM KT
number [m/min] [mm/rev] [min] [mm] [mm]
∂w E (1) 1 0.389 0.027
= D ⋅ exp(− )
∂t RT 1 150 0.17
2 0.476 0.038
4 0.554 0.074
being D a material constant, E the activation energy 6 0.497 0.095
(75,35 kJ/mol), R (8,314 kJ/mol K) the gas constant 1 0.548 0.020
and T the local temperature in K. Thus, the 2 160 0.25
2 0.558 0.048
calibration of parameter D, as a function of both 4 0.686 0.099
6 0.693 0.110
cutting time and temperature, assumes dramatic
1 0.475 0.042
importance. This task was carried out utilizing the 2 0.467 0.076
ALE approach to predict the temperature 3 190 0.18
4 0.576 0.092
distribution in the tool and an inverse approach was 6 0.656 0.132
used to calculate the optimal D value for each
experimental test condition [4].
As results, it is possible to state that the dependence 4 3D FE SIMULATIONS
of D on the cutting time is negligible for normal
cutting durations (apart the first minutes) while it is 4.1 The model
very sensitive with temperature ( Figure 1).
The 3D model, implemented in Deform 3D v6.1, is
D(T ) = 1.6735⋅10− 9 ⋅ T 3 − 5.5373⋅10− 6 ⋅ T 2 + 6.1161⋅10− 3 ⋅ T − 2.251
reported in Figure 2 where it is possible to see the
workpiece with the growing chip and the tool. The
tool, a rigid object meshed with more than 100,000
elements, is oriented according to the cutting angles
set in experimental test and reported in Table 2 and
it moves along a linear direction. The workpiece,
considered as a rigid-plastic object meshed with
more that 45,000 elements, is fully constrained on
the lower and lateral sides so it cannot move. On the
same faces thermal boundary conditions are set so to
Fig. 1. Evolution of parameter D vs. temperature. simulate the heat diffusion. The thermal exchange
For this reason, the third order polynomial law between tool and chip is regulated by heat diffusion
(correlation index=94.85%) reported in Figure 1 relationship whose characteristic parameters are
was derived in order to consider D variation in the reported in Table 2. The friction was modelled
equation 1. considering a shear factor equal to 0.82.
Table2. Geometry and materials data. thermal steady state. In this way the thermo-
Parameter Values mechanical steady state is achieved in a shorter
Tool
simulation time. After these two simulation phases,
γ angle 0°
λ angle 7° an additional Lagrangian simulation is run for a
χ angle 90° constant simulation time ∆t in order to identify the
Material WC correct temperature and force distributions. At this
Conductivity [W/m K] 69 point it is possible to run the implemented wear
Heat Capacity [J/m3 K] 3.8·106 calculation subroutine and, as a consequence, to
Emissivity 0.45
Workpiece update the geometry and the worn tool mesh. The
Material AISI 1045 (Oxley equation) mesh updating procedure is repeated until the total
Conductivity [W/m K] Function of temperature cutting time (tend) is reached (loop in Figure 3).
Heat Capacity [J/m3 K] Function of temperature
Emissivity 0.75 4.2 Mesh updating subroutine
Shear friction factor (m) 0.82
The subroutine for tool mesh and geometry updating
The adopted simulation strategy is schematized in calculates the wear rate according to the analytical
the flow chart of Figure 3 and it is an evolution of a model described in paragraph 2 for each node of the
simulative strategy presented in previous works [5]. tool mesh boundary. Thus the mesh nodes
movement direction is identified and finally the tool
Tool geometry and mesh are updated (Figure 4).

Tool Wear rate

Calculation

Chip
Identification of 3D
Movement Direction

Tool Mesh and

Workpiece Geometry Update

Fig. 2. The implemented 3D FE model. Fig. 4. Tool mesh and geometry updating subroutine.

Incremental Force Steady In particular, Figure 5 illustrates how the subroutine

Lagrangian Model State algorithm calculates the node movement direction.
For each node the procedure finds the connected
Machining Steady- Temperature elements and for each connected element the values
State (ALE) Steady State
of the three versor components are determined. The
Force and node movement direction is obtained as vectorial
Incremental
Lagrangian Model
Temperature sum of all the vectors of the connected elements.
Steady State

Time=Fixed ∆t Node to move

N=Σni
Wear Calculation Subroutine
n
No
Time=tend? N
n
Yes

End
Initial
Fig. 3. Simulation strategy. Connected
geometry
element
The simulation starts with an updated Lagrangian Worn
formulation for reaching the mechanical steady state, geometry
followed by an Eulerian formulation for reaching the Fig. 5. Algorithm for node movement direction identification.
5 RESULTS the tool wears.

Once the 3D simulative strategy for tool geometry

KM FEM KM Exp
updating was implemented, the attention was
focused on crater wear prediction as the cutting
parameters change. Figure 6 shows the worn tool
geometry as calculated by the above described
procedure. In particular, a section of the tool is
reported and the simulated crater wear after 6 KT FEM
minutes of cut for test number 3 is presented
together with its characteristic parameters (KT and KT Exp
KM). Table 3 reports the FEM results obtained for
the analysed cases.
A-A
KM
Fig. 7. Development of experimental and simulative crater
wear (KT and KM) as a function of cutting time for test 2.
KT

6 CONCLUSIONS

The paper reports the results obtained simulating the

tool wear development using a 3D FE model. In
particular, a new subroutine for tool wear calculation
and tool mesh and geometry updating was
A A implemented in a commercial 3D code. The
Fig. 6. Worn tool and crater wear parameters (KT-KM). obtained results showed a good agreement with the
Table 3. Simulative results.
experimental tests proving the reliability of the
Test VC f Time KM KT developed procedure.
number [m/min] [mm/rev] [min] [mm] [mm] For the future, new efforts will be spent to
1 0.540 0.023 implement a more comprehensive wear model
2 0.540 0.040 taking into account also the abrasive effect.
1 150 0.17
4 0.565 0.076
6 0.488 0.091
1 0.611 0.025 REFERENCES
2 0.609 0.049
2 160 0.25 1. V.P. Astakhov, The assessment of cutting tool wear, Int.
4 0.596 0.093
6 0.597 0.116 J. Mach. Tools Manuf, 44, (2004) 637-647.
1 0.524 0.069 2. Y.C. Yen, J. Sohner, B. Lilly, T. Altan, Estimation of
2 0.547 0.082 tool wear in orthogonal cutting using the finite element
3 190 0.18 analysis, J. Mater. Proc. Technol., 146, (2004) 82-91.
4 0.507 0.099
6 0.563 0.120 3. H. Takeyama and T. Murata, Basic investigations on tool
wear, Trans. of ASME, J. Eng. Ind., 85, (1963) 33-38.
The comparison between experimental and 4. R. Calzavarini, L. Filice, L. Settineri, D. Umbrello, F.
Micari, Experimental Campaign to Assess Numerical
simulative results in terms of KT and KM parameters Simulation of Tool Wear in Orthogonal Cutting, In:
for test 2 are reported in Figure 7. The agreement is Proceedings of the High Speed Manufacturing, Senlis,
clear so allowing to validate the wear model and the France, (2007), CD support.
simulative procedure. Good results were obtained 5. E. Ceretti, C. Giardini, A. Attanasio, L. Filice, S. Rizzuti,
for all the other tests, too. The higher differences Diffusion wear Modeling in 3D cutting process, In:
Proceedings of the 10th CIRP Int. Workshop on Modeling
between FE and experimental results were obtained of Machining Operations, eds, F. Micari and L. Filice,
for cases 1 and 3 at the beginning of the cutting test. Reggio Calabria , Italy, (2007) 287-292.
These differences decrease as the cut continues and
Surface roughness modelling in finish face milling under MQL and dry
cutting conditions
C. Bruni, L. d’Apolito, A.Forcellese, F. Gabrielli, M. Simoncini
Department of Mechanics, Università Politecnica delle Marche, 60131 Ancona, Italy
URL: www.dipmec.univpm.it e-mail: c.bruni@univpm.it; l.dapolito@univpm.it; a.forcellese@univpm.it;
f.gabrielli@univpm.it; m.simoncini@univpm.it

ABSTRACT: The effect of lubrication-cooling condition on surface roughness in finish face milling
operations has been widely investigated. Different cutting speeds and lubrication cooling conditions (dry, wet
and MQL), in finish face milling of AISI 420 B stainless steel, have been considered. The evolution of the
surface finish and tool wear with cutting time have been monitored. Analytical and artificial neural network-
models, able to predict the surface roughness under different machining conditions, have been proposed.

Key words: Dry cutting, MQL, Finish face milling, Modelling, Surface roughness

1 INTRODUCTION
The reduced utilization of cooling lubricants, in
order to improve environmental protection, safety of
machining processes and to decrease time and costs
related to the number of machining operations, can
be pursued performing machining processes with the
MQL (minimum quantity of lubricant) technique or
without any cutting fluid (dry cutting) [1]. Such
approaches can allow the obtaining of the product
specifications, in terms of surface roughness and
dimensional accuracy, by the shortening
conventional process cycles (i.e. avoiding grinding).
The effect of the lubrication-cooling condition on
the surface quality of the machined part strongly
depends on the type of machining operation to be
performed (e.g. turning, milling, etc.), as well as on
the process parameters to be used. In particular, in
face milling operations cutting takes place with high
frequency tooth impacts, depending on the cutting
speed, and discontinuously due to the presence of
several teeth; for such reasons dry and MQL face
milling can be performed over a wide field of
workpiece materials [2-4], once that the proper
cutting materials and tool coatings, with improved
performances, and machining parameters are
considered [4-7].
A very useful tool for industrial finish machining
applications is represented by the availability of
models able to predict surface roughness (Ra) as a
function of lubrication cooling technique, cutting
parameters, etc. In this way, the knowledge of the
surface roughness levels can be used in the design
stage of machining operations. A review of
predictive models and related approaches has been
reported in [8,9], also under dry machining [10].
Among them statistical (MRA) and artificial neural
network (ANN) modelling approaches are the most
used.
In this framework, the present work aims at building
predictive models of surface roughness including,
among the input parameters, also the lubrication
cooling condition. The present paper represents the
first step of such investigation and focuses on the
study in depth of the effect of different lubrication-
cooling conditions and cutting speed on the surface
roughness in finish face milling operations. The
machining tests have been performed at different
cutting conditions on AISI 420B stainless steel.
Analytical and non analytical models, relating
surface roughness with process parameters and
lubrication cooling condition, are proposed.
2 EXPERIMENTAL AND MODELLING the coefficients ai (i=1,..9) represent the regression
coefficients. The values of such coefficient are
2.1 Experimental summarised in table 1.
Concerning the ANN-based approach, a multi-layer
Finish face milling tests were performed on blocks feed forward artificial neural network, using the
(width:32 mm; length along the feed direction: 345 back-propagation algorithm, was built. Nine inputs
mm; height: 130 mm) of 420B stainless steel under were used: Vc, t, LC, Vc2, t2, LC2, Vct, LCt,VcLC. The
wet, dry and MQL conditions. The tests under MQL output of the ANN was the Ra value. Different
condition were performed using a system based on network configurations were considered; the final
the use of a pneumatic pump delivering a minimal one consisted of one hidden layer with nine hidden
quantity of lubricant (20 ml/h) along a capillary tube neurons. The topology and training parameters for
fitted inside length of the air line to the nozzle head. the developed artificial neural network-based models
At this point the lubricant droplet is introduced into are shown in table 2.
the air stream and transported to the cutting edge.
The tool holder was characterised by a diameter Table1. Regression coefficients.
(THD) of 63 mm. Five inserts in cemented carbide Coefficient Value
(R245 12 T3 E-ML) [11] with two layer coatings a0 1.056
a1 5.65E-04
(TiN and TiAlN) were mounted on the tool holder a2 -1.04E-02
with the axial rake angle of 23° [7]. The milling a3 3.23E-02
experiments were carried out with only one tooth– a4 1.41E-05
workpiece contact each time. a5 3.12E-05
The cutting parameters were selected by considering a6 1.00
a7 2.21E-06
that finish face milling can be used as an operation a8 -1.23E-02
alternative to grinding. Therefore, according to the a9 6.9E-05
tool manufacturer recommendations [11], cutting
speed (Vc) was varied between 120 and 180 m/min. Table2. Topology and training parameters for ANN.
A depth of cut of 0.2 mm and a feed of 0.14 Number of input nodes 9
Number of output nodes 1 (Ra)
mm/tooth was used. The effect of the feed variation
Number of hidden layers 1
was not taken into account for its negligible effect Number of hidden nodes 9
on surface roughness, due to the geometry of the Activation function input-hidden layers Sigmoid
insert used [11]. The wear criterion and the approach Activation function output-hidden layers Linear
followed for tool wear and surface roughness Distribution of weights Gaussian
Momentum coefficient 0.1
evaluation are reported in [7].
Learning coefficient 0.9
2.2 Modelling approach 3 RESULTS AND DISCUSSION
The surface roughness Ra was modelled using the 3.1 Experimental
multiple regression analysis (MRA) and artificial
neural network (ANN) approaches. In both the The surface roughness, plotted vs. time under
cases, the surface roughness Ra was related to the different conditions, in terms of cutting speed and
cutting speed (Vc), cutting time (t) and lubrication lubrication-cooling technique, is reported in Figure
cooling condition. When the MRA approach is 1. For each cutting speed investigated, Ra tends to
concerned a second (polynomial) order regression decrease with increasing cutting time under wet
model was used according to the following cutting, as shown by other authors [3], whilst a slight
formulation: increase can be detected under dry cutting. When the
MQL condition is considered, it can be observed that
2
Ra = a0 + a1t + a2Vc + a3 LC + a4t 2 + a5Vc + a6 LC 2 + the Ra vs. cutting time curves assume values similar
to, or lower than, those obtained under wet cutting.
+ a7Vc t + a8 LCt + a9Vc LC (1)
Moreover, the VB values detected under MQL
condition are slightly lower than those observed
where, LC represents a constant value which takes under wet and dry conditions, especially at the
into account the lubrication cooling condition and highest cutting speed investigated.
 0.45 As shown by the authors in a previous work [7] and
by other researchers [3] the mean tool-chip interface
0.4 a)
temperature detected under dry cutting is higher than
0.35 that observed under wet machining.This could be
responsible for the increase in VB but, on the other
0.3
hand, also for the workpiece material softening. In
the experimental conditions of the present
Ra [µm]

0.25
investigation, the latter effect should prevail on the
0.2
former, at least at lowest cutting speed investigated.
0.15 Vc = 120 m/min The interesting results obtained under MQL
Vc = 150 m/min conditions, in terms of Ra and VB, can be attributed
0.1
Vc = 180 m/min to the beneficial effect of the aerosol that produces a
0.05 cooling of the insert allowing at the same time the
0 material softening due to temperature increase in the
0 10 20 30 40 50 60 70 80 90
deforming zone, however, such aspect needs to be
Cutting time, min
further investigated.
0.12
0.45
b) 0.1
0.4

0.35
0.08
0.3
VB, mm
Ra [µm]

0.25 0.06

0.2
0.04 DRY - Vc = 120 m/min
WET - Vc = 120 m/min
0.15 MQL - Vc = 120 m/min
Vc = 120 m/min DRY - Vc = 180 m/min
0.1 Vc = 150 m/min 0.02
WET - Vc = 180 m/min
Vc = 180 m/min MQL - Vc = 180 m/min
0.05
0
0 0 20 40 60 80 100
0 10 20 30 40 50 60 70 80 90 Cutting time, min
Cutting time,min
Fig. 2. VB vs. cutting time, under wet, dry and MQL condition
0.45 at Vc=120 and 180 m/min.
c)
0.4 3.2 Modelling
0.35
The effectiveness of both the modelling approaches
0.3 in predicting Ra has been checked using surface
roughness vs. cutting time curves not used in the
0.25
Ra, µm

building of the model. Figure 3 shows the

0.2 comparison between experimental Ra vs. cutting
time curve, obtained at 150 m/min under wet
0.15
Vc = 120 m/min
condition, and the ones predicted using MRA and
0.1 ANN models.
Vc = 150 m/min
Both the MRA and ANN models, under the
0.05 Vc = 180 m/min
experimental and modelling conditions of the
0 present investigation, allow to predict Ra vs. cutting
0 10 20 30 40 50 60 70 80 90 time curves, when the lubrication cooling condition
Cutting time, min
is considered as a input variable.
Fig. 1. Surface roughness vs. cutting time under wet (a), dry
(b) and MQL condition (c) at Vc=120, 150 and 180 m/min.
 Eng. M. Pieralisi of the Università Politecnica delle Marche for
0.45
his help in performing experimental work.
0.4

0.35 REFERENCES

0.3 1. A.E. Diniz, J.R. Ferreira and F.T. Filho, Influence of

refrigeration/lubrication condition on SAE 52100
Ra, mm

0.25 hardened steel turning at several cutting speeds.

International Journal of Machine Tools and Manufacture
0.2 43 (2003) 317-326.
2. K. Weinert, I. Inasaki, J.W. Sutherland and
0.15 Wakabayashi, Dry machining and minimum quantity
EXP
0.1 lubrication. Annals of the CIRP. 53/2 (2004) 1-27.
MRA
3. J.M. Vieira, A.R. Machado and E.O. Ezugwu,
0.05 ANN Performance of cutting fluids during face milling of
steels, Journal of Material Processing Technology, 116
0 (2001) 244-251.
0 10 20 30 40 50 60 70 80 4. N. Liu, C. Han, H. Yang, Y. Xu, M. Shi, S. Chao, F. Xie,
Cutting time, min The milling performances of TiC-based cermet tools
with TiN nanopowders addition against normalized
medium carbon steel AISI1045, Wear, 258 (2005) 1688-
Fig. 3. Comparison between experimental Ra vs. cutting time 1695.
curve obtained at 150 m/min – wet and those predicted by 5. M. Rahman, A. S. Kumar, M.U. Salam, Experimental
MRA and ANN. evaluation on the effect of minimal quantities of
lubricant in milling, International Journal of Machine
Tools & Manufacture, 42 (2002) 539-547.
4 CONCLUSION 6. Y.S. Liao, H.M. Lin, Y.C. Chen, Feasibility study of the
minimum quantity lubrication in high-speed end milling
Machining tests have been performed under of NAK80 hardened steel by coated carbide tool.
different cutting conditions on AISI 420B stainless International Journal of Machine Tools & Manufacture,
47 (2007) 1667-1676.
steel. MRA and ANN models, relating surface 7. C. Bruni, L. d’Apolito, A. Forcellese, F. Gabrielli,
roughness with parameters and lubrication cooling L.Gobbi, M. Simoncini, Finish milling operations of
condition, have been proposed. AISI 420 B stainless steel under dry and wet conditions,
The MQL lubrication cooling technique provides, In: Proc. of the 10th CIRP International Workshop on
under the experimental conditions of the present Modelling of Machining Operations, Eds, F. Micari and
L. Filice, Reggio Calabria, Italy (2007) 275-279.
research, very low values of Ra and VB, especially 8. B. Ozcelik and M. Bayramoglu, The statistical modelling
at higher cutting speeds. When the modelling stage of surface roughness in high-speed flat end milling.
is concerned, both the MRA and ANN models can International Journal of Machine Tools & Manufacture,
be used in predicting Ra values. Of course, the 46 (2006) 1395-1402.
prediction capability of both the models can be 9. P.G. Bernardos, G.-C. Vosniakos, Predicting surface
roughness in machining: a review. International Journal
improved with increasing the number of the of Machine Tools & Manufacture, 43 (2003) 833-844.
experimental curves to be used in the building stage 10. M.-Y. Wang, H.-Y. Chang, Experimental study of
and in the validation one. surface roughness in slot end milling AL2014-T6.
International Journal of Machine Tools & Manufacture,
44 (2004) 51-57.
ACKNOWLEDGEMENTS 11. Sandvik Coromant, 2007, Insert Technical Data
(http://www.coromant.sandvik.com).
The research reported in this paper was performed within the
project CIPE 20/2004 – Marche Region. Authors wish to thank
Assessment of material models through simple machining tests
F. Chinesta 1, L. Filice 2, F. Micari 3, S. Rizzuti 2, D. Umbrello 2
1
LMSP UMR 8106 CNRS-ENSAM-ESEM – F75013, 151 Boulevard de l'Hopital, Paris - France
URL: www.ensam.fr e-mail: francisco.chinesta@paris.ensam.fr
2
Dept. of Mechanical Engineering, University of Calabria – 87036, Rende - Italy
URL: www.unical.it e-mail: l.filice@unical.it; stefania.rizzuti@unical.it
3
Dept. of Technology, Production and Management Eng., University of Palermo –90128, Palermo - Italy
URL: www.unipa.it e-mail: micari@dtpm.unipa.it

ABSTRACT: The accuracy of the results obtained from FEM simulation of machining operations depends on
the accuracy of input data. Among these, the flow stress data of the workpiece are extremely important
together with the friction along the tool-chip interface. In this study, an identification procedure for the
determination of material parameters that are used for the FEM simulation of machining processes is
proposed. The procedure is based on the coupling of a numerical identification procedure and Arbitrarian
Lagrangian Eulerian (ALE) Finite Element simulations of turning operations. An experimental campaign was
developed in order to calibrate the model and to validate the procedure. The basic idea is to utilize only
machining tests to properly define the material behaviour of the workpiece material, taking also into account
the thermal phenomena involved in the process. The preliminary results of this analysis are discussed in the
paper.

Key words: Material modelling, Inverse Approach, Machining, Finite Element Simulation

behaviour modelling is usually regarded as a known

1 INTRODUCTION
Finite Element Method (FEM) based simulation is a
very useful tool for obtaining relevant information in
machining; the latter a usually difficult to
experimentally acquire. Some examples are strain,
strain rate, stress and temperature in both tool and
machined workpiece. Anyway, the reliability of the
results obtained from FEM simulation depends on
the accuracy of the input values. For these reasons,
numerical simulation of cutting process is still a very
hard challenge since a lot of staffs have to be taken
into account. Although sometimes a strong
geometrical simplification is done studying
orthogonal cutting conditions (thus reducing the
computational complexity), other aspects must be
properly modelled in order to supply reliable result
as simulation output. Among them, material
behaviour, friction, energy conversion into heat and
its propagation, definition of material damage
criteria are just few examples that engage a number
of scientists all over the World. Actually, material
issue since several laws are easily recognizable in
literature [1-4], almost for the most diffused
materials. On the other hand, comparing the
different laws, it is possible to recognize also
relevant discrepancies that suggest to investigate
more deeply in that direction [5]. Traditionally the
studies were mainly oriented toward friction
modelling [6], damage criteria development [7],
extension to 3D geometries but, nevertheless these
huge efforts, some results are today still
unsatisfactory. For instance, it is well known in the
cutting community that the simple ratio between
cutting and thrust force is not accurately calculated.
At the same time, the thermal field is usually not
well calculated, especially using Lagrangian
formulation. This relevant inaccuracy may be due,
also, to the material behaviour laws which are
usually obtained by processes completely different
respect to cutting. In fact, how it is well-known, to
be useful in metal cutting simulation, flow stress
data must be obtained at high strain rates (up to 106
s-1), temperatures (up to 1000°C) and strain (up to
4). These data are impossible to reliably obtain with
conventional tensile and compression tests. Often
high speed Hopkinson’s bar compression tests are
used but these tests require much effort and
expensive equipment, and can provide data only for
limited strain rates (up to 104 s-1) and strain (up to
1.0 or less). Taking into account all these
considerations, a new strategy is proposed in this
work, based on the use of some simple measures
obtained during cutting process, i.e. cutting force,
thrust force and temperature in the tool. In this
preliminary approach, for sake of simplicity, the
problem was decomposed taking into account the
reduced sensitivity of some variables on the material
law. The latter was then obtained by an inverse
approach using the results of a Lagrangian-Eulerian
simulation and utilizing an optimization algorithm.
The validity of the simplified material law was
determined and its validity was verified in other
cases. A satisfactory agreement between
experimental and numerical data was found. The
results will be presented in the following.
The workpiece material was an AISI 1045 steel,
while the cutting tool was an uncoated carbide ISO
P30, with a rake angle of 0° and a relief angle of 6°.
The tests were executed without any lubricant at the
tool–chip interface, with the values of cutting speed
and feed rate reported in Table 1.
Table 1. Cutting conditions
Test case 1 2 3
Cutting speed Vc [m/min] 100 150 100
Feed f [mm/rev] 0.1 0.1 0.15
During the tests cutting and thrust forces were
measured by using a piezoelectric dynamometer
while an optical microscope was used to estimate the
chip thickness. The temperature was measured using
an embedded thermocouple, close to the cutting
edge, at a distance of 0.6 mm from the rake face and
0.35 mm from the tool tip (Figure 2).

2 THE PROPOSED PROCEDURE

2.1 Experimental tests

As above introduced, an experimental investigation Fig. 2. Thermocouple position

was preliminarily carried out in order to acquire
reliable data to develop the procedure and to set-up Table 2 shows the average experimental results in
the material model. In particular, cutting processes terms of cutting and thrust forces, chip contact
were run on a lathe, reproducing orthogonal cutting length, chip thickness and shear angle.
conditions (Figure 1).
Table 2. Experimental results
Test case 1 2 3
Cutting force Fc [N] 745 715 1027
Thrust force Ft [N] 600 522 749
Chip thickness [mm] 0.288 0.220 0.355
Shear angle [deg] 19.0 24.4 23.0
Temperature [°C] 542 567 596

2.2 Numerical procedure

In the present study the FEM commercial software

Fig. 1.The experimental set-up.
“DEFORM-2D” was used to simulate orthogonal
cutting. The reference cutting condition used was
identified in test case 1 of Table 1 and Table 2.
Workpiece and tool sizes in simulation model were
set large enough so that the predicted results were
not sensitive to the boundary conditions. Figure 3
shows the FE model with the thermocouple socket
position.
 contrary, it’s strongly related to friction coefficient.
Thus, in this phase no thermal aspects were
considered and the h value was set equal to zero. In
particular the constant shear friction model with a
coefficient equal to 0.98 permitted to obtain the
minimum shear angle error.
Subsequently, an Eulerian analysis was run taking
into account the Lagrangian step outputs. It’s worth
noticing, in fact, that both the deformed mesh and
the velocity distributions were assumed as input for
the Eulerian analysis. In this case the workpiece
material behaviour was described by means of a
simple law (Eq. 3) taking into account strain-rate
and temperature:
. m

Fig. 3. FE model σ = C ⋅ ε ⋅ T −γ (3)

An Updated-Lagrangian simulation was executed where C, m and γ are constants.

utilizing a well-known power law for the workpiece
material. In particular, in this first case the employed
model was the one proposed by Oxley [1], in which
the effective flow stress is computed as an
exponential function of the effective strain ε (Eq. 1):
σ = σ 1ε n 1
(1)
These constants were found utilizing an inverse
procedure based on the optimization algorithm of
Newton-Raphson. In particular, the three unknown
parameters were determined minimizing the
objective function E, representing the difference
between the experimental and corresponding
computed cutting and thrust forces and temperature:

being σ1 the strength coefficient and n1 the strain- E= ((Fc

num
− Fc ) + (F
exp 2
t
num
− Ft ) + (T
exp 2 num
− T exp ) ) (4)
2

hardening exponent. Both σ1 and n1 vary with

velocity-modified temperature Tmod, which couples where the numerical values derive from the Eulerian
. finite element simulation.
the effects of strain rate ε and the temperature T1. Of course, Eq. 3 introduces a strong simplification in
The concept of velocity-modified temperature was the analysis, partially justified by the fact that a
proposed by MacGregor and Fisher [9] and is sensitivity analysis shown that the strain influence is
defined as: lower than the others. In particular, in Eulerian
⎛ frame all the data are not related with strain.
ε ⎞⎟
.
⎜
= T1 ⎜1 − ν log . ⎟ Finally, it was possible to identify the following
Tmod
⎜ ε 0 ⎟⎠ material law:
⎝ (2)
. 0.006
. σ = 970 ⋅ ε ⋅ T − 0.0225 (5)
where ν and ε 0 are constants that depend on the
workpiece material and are readily available [1]. The The application of this simple power law in Eulerian
step of the turning process provided the results
relationship between Tmod and σ1 and the
reported in Table 3 which also reports the prediction
relationship between Tmod and n1 have been
accuracy.
experimentally established for carbon steels using
high speed compression tests [1]. Table 3. Results for Vc=100 m/min and a f=0.1 mm/rev
The aim of this step was to find the friction model Fc [N] eFc% Ft [N] eFt% T eT%
745.2 0.03 600.2 0.03 542.6 0.11
and the friction value in order to obtain a numerical
shear angle as nearest as possible to the Subsequently, the law (Eq. 5) was used to predict the
experimental one. In fact, an important assumption cutting and thrust forces such as the temperature in
is here done, according with previous results [5]: the two new cases, with different cutting conditions.
chip-tool geometry is not relevantly dependent on Also in these cases the procedure was the previously
the material law and thermal distribution. On the illustrated one. A Lagrangian simulation was
performed, utilizing the Oxley’s law to describe the
AISI 1045 behaviour. The coefficient was
determined in order to approach the experimental
geometry. Also in these cases, the constant shear
model and the friction coefficient equal to 0.98
allowed to minimize the errors. The relative errors
for the shear angle were about 5% and 9%
respectively. Table 4 reports the predicted variable
at the end of the Eulerian simulation for a cutting
speed of 150 m/min and feed equal to 0.1 mm/rev.
Table 4. Results for Vc=150 m/min and a f=0.1 mm/rev
Fc [N] eFc% Ft [N] eFt% T eT%
694.4 -2.9 570.9 9.4 515.9 -9.0
Forces, temperature and relative errors for a cutting
speed of 100 m/min and a feed rate of 0.15 mm/rev
are summarized in Table 5.
Table 5. Results for Vc=100 m/min and a f=0.15 mm/rev
Fc [N] eFc% Ft [N] eFt% T eT%
1045.8 1.8 821.7 9.7 575.7 -3.4
3 SOME CONSIDERATIONS
The preliminary proposed procedure provided
satisfactory results, as it appears analyzing the data
reported in the Tables 3-5. In fact, the average error
is less than 10% and the researchers involved in this
field surely agree to consider this discrepancy
absolutely acceptable. Of course the calibration and
validation procedures are not so robust at this state
but the actual aim of this work is to propose a
breaking procedure which starts from a different
point of view respect to the traditional approaches.
Its main point of strength is the possibility to
simulate the process in Eulerian environment,
without knowing a priori the material flow rule. The
latter can be directly derived by using simple steps if
some mechanical and thermal data are available.
In the opinion of the authors this way could
constitute a point of discontinuity with the past since
material is characterised in its real processing
conditions thus it is intrinsically more robust and
physically consistent. Temperature, strain and,
mainly, strain rate are of the same order of the
process ones, at difference of any other
characterisation test.
On the other hand, it is clear that substantial
improvements will be necessary in terms of
robustness and completeness.
4 CONCLUSIONS
The preliminary results of a study concerning
material modelling through simple machining tests
were presented. Nevertheless the strong
simplification and the rough calibration and
validation approach, the performance of the
proposed procedure was verified comparing the
numerical and experimental data.
In the future, the methodology has to be surely
refined supplying a powerful procedure able to allow
the simple calculation of the material constitutive
low just measuring few experimental data.
ACKNOWLEDGEMENTS
The authors gratefully thank prof. Luca Settineri of ISTEC-
CNR for his support in developing the experiments.
REFERENCES
1. P.L.-B. Oxley, Mechanics of Machining – An Analytical
Approach to Assessing Machinability, Ellis Horwood
Limited, London (1989).
2. G.-R. Johnson, W.-H. Cook, A Constitutive Model and
Data for Metals Subjected to Large Strains, High Strain
Rates and High Temperatures, Proceedings of the Int.
Symposium on Ballistics, Netherlands, (1983) 541-547.
3. P. Sartkulvanich, F. Koppa, T. Altan, Determination of
Flow Stress for Metal Cutting Simulation – a Progress Re-
port, Journal of Materials Processing Technology, 146,
(2004) 61-71.
4. T. Shirakashi, K. Maekawa, E. Usui, Flow stress of low
carbon steel at high temperature and strain rate (Parts I-
II), Bulletin of the Japan Society of Precision
Engineering, 16, (1983) 161-172.
5. L. Filice, F. Micari, S. Rizzuti, D. Umbrello, Dependance
of Machining Simulation Effectiveness on Material and
Friction Modelling. That’s Why Industrial Application of
Machining Simulation is still so far, Proceedings of the
10th Int. Workshop on Modelling of Machining
Operations, Reggio Calabria, (2007) 181-187.
6. T.H.C. Childs, Friction Modelling in Metal Cutting, Wear,
260 (3), (2006) 310-318.
7. J. Hua, R. Shivpuri, Prediction of Chip Morphology and
Segmentation during the Machining of Titanium Alloys,
Journal of Materials Processing Technology, 150 (1-2),
(2004) 124-133.
8. N. Ahmed, A.V. Mitrofanov, V.I. Babitsky, V.V.
Silberschmidt, 3D Finite Element Analysis of
Ultrasonically Assisted Turning, Computational Materials
Science, 39 (1), (2007) 149-154.
9. C.-W. Mac Gregor, J.-C. Fisher, ‘A Velocity-Modified
Temperature for the Plastic Flow of Metals’, ASME J.
Appl. Mech., 13, (1946) 211–218.
Characterization of the frictional properties of various coatings at the
tool/chip/workpiece interfaces in dry machining of AISI 4140 steel
C.Claudin1, J.Rech1, W.Grzesik2, S.Zalisz2
1
Ecole Nationale d’Ingénieurs de Saint-Etienne, LTDS, 58 Rue Jean Parot – 42000 Saint-Etienne - France
e-mail: joel.rech@enise.fr
2
Opole University of Technology, Department of Manufacturing Engineering and Production Automation,
Poland
e-mail: w.grzesik@po.opole.pl

ABSTRACT: This work deals with the development of a new tribometer designed for the characterization of
the frictional properties at the tool/chip/workpiece interfaces in cutting processes. The average contact
pressure can be selected up to 3 GPa under sliding velocities reaching 300 m/min. Moreover the system
provides a continuously regenerated contact surface. Under such severe conditions, which are not reachable
with conventional tribometers, the apparent friction coefficient is quantified in parallel to the heat flux
transmitted to the pin. This new system has been applied to the characterization of the frictional properties
during the dry machining of a AISI 4140 treated steel with various coated carbide substrates. The coatings
investigated in this work are: TiN, TiAlN, AlCrN coatings deposited by PVD. A large range of sliding
velocities and pressures have been investigated in dry conditions. It has been shown that the coating is the
most influential parameter, followed by the sliding velocity and finally the pressure. As far as the friction
coefficient is concerned, the superiority of AlCrN and TiAlN coatings in dry cutting has been revealed.

Key words: Friction; tribology; cutting; coating; adhesion.

machining, usual cutting conditions lead to severe

1 INTRODUCTION
In the context of a global competition, companies
are compelled to improve their productivities. As a
consequence they have to optimize their production
processes including cutting processes. In order to
achieve this aim, industry adopts very high cutting
regimes (high cutting speeds and feed rates) and dry
conditions. Under severe conditions, caused by a
high cutting regime, the mechanical stresses and
temperatures at tool-chip interface and around the
cutting edge can be critically high, resulting in an
excessive tool wear or even a premature tool failure.
Finite element methods are even more applied in
order to optimize cutting tool design. However the
‘cutting’ scientific community regrets an important
lack of fundamental understanding of the frictional
phenomena occurring at the tool-chip interface
(‘secondary shear zone’ in Fig. 1) and at the tool-
workpiece interface (‘rubbing zone’ in Fig. 1) in
order to achieve and accurate modelling by means of
such finite element methods [1]. In the case of steel
tribological conditions: high velocities (60-600
m/min), high temperatures (up to 1000°C), high
pressures (up to 2 GPa) since [2-4].
In order to investigate the tribological phenomena at
these interfaces, scientists use two approaches. The
first approach consists in using the cutting process
itself [5-6] but it is very difficult to discriminate the
proportion among friction on interfaces and
workmaterial shearing in the primary shear zone.
The second approach consists in using laboratory
tests. They enable to master much more accurately
the contact conditions and to modify these
conditions as desired. Several frictional set-ups
already exist but the conditions (temperature,
pressure) are not relevant to those observed in
cutting processes [7-8]. Moreover, during a cutting
operation, the chip flows on the rake face and never
comes back. In parallel, the cut surface rubs the
flank face and is never more in contact with the tool
(Fig. 1). Specialists in tribology use the terminology
“open tribo-system”. On the contrary, in standard
tests such as pin-on-disc systems, the pin always
rubs on the same track. This configuration is a so-
called “closed tribo-system”. Both configurations
lead to fully different tribological results. So a new
configuration of frictional tests has to be designed.
The new system presented below has been applied in
order to characterize the frictional influence of
various coatings deposited on carbide tools during
the high speed dry cutting of a treated AISI 4140
steel (300 HB):
- TiN (PVD - cathodic arc - ~2µm - monolayer)
- TiAlN (PVD - cathodic arc - ~2µm - monolayer)
- AlCrN (PVD - cathodic arc - ~2µm - monolayer)
2 EXPERIMENTAL SET-UP
The principle of this tribometer is based on the
principle proposed by Hedenqvist et al. [9]. A pin,
having a spherical geometry at its extremity, rubs on
a refreshed surface in rotation. This sphere-plane
contact configuration has been chosen in order to
reach relevant contact pressures. Several diameters
can be used in order to vary contact pressure. It is
made of cemented carbide with a similar grade to the
one used for cutting tools (90% WC – 10%Co –
average grain size ~ 0.8 µm.
The bar is made of the workmaterial to be machined
with the relevant microstructural state. A feed
provided to the pin enables a helical movement. As a
consequence, the contact surface between the pin
and the sample is continuously regenerated. This
experimental set-up, installed on a lathe, can lead to
sufficient sliding velocities (several hundreds of
m/min). After having made a friction test from one
side of the bar to the other side, the surface has to be
regenerated. The regeneration is rather easy since it
is just necessary to make a longitudinal turning
operation. A belt finishing operation is also
performed in order to obtain a very low surface
roughness. So it becomes simple to start again
making friction tests.
The pin is maintained by an instrumented pin-holder
able to provide data about the instantaneous heat
flux entering into the pin. This specific equipment is
described in the references [10-11].
The pin-holder is fixed onto a dynamometer in order
to provide the apparent normal and tangential force
(macroscopic forces).
2.1 Testing conditions
Testing conditions have to be chosen in accordance
with the frictional conditions estimated along the
tool/chip/workpiece interfaces. For instance, the
machining of a AISI 4140 steel with a TiN coated
carbide tool in dry cutting conditions is commonly
performed at a cutting speed in the range Vc~120-
200 m/min. It is well known that the macroscopic
sliding velocity Vs at the flank face (tool-workpiece
interface) is almost equal to the cutting speed Vc. On
the contrary, it is well known that the chip section is
much larger than the input section (= the feed
multiplied by the depth of cut: f.ap, in the case of
orthogonal cutting). The ratio between the chip
thickness and the feed is usually called the chip
compression ratio β. In the case of such steels, the
chip compression ratio β is usually around 2. These
values have been confirmed by experimental
measurements of chips after standard turning
operation in the same configuration (same couple of
materials) [12]. So, from a macroscopic point of
view, the average sliding velocity on the tool rake
face should be 2 times lower than the cutting speed ~
60 m/min.
3‐ Pin :
Carbide +
coating
1‐Workpiece : 
AISI4140
4‐Pneumatic jack
2‐Cutting tool 
refreshing
the surface
6‐Dynamometer 5‐Pin holder
Heat Flux
meas.
Forces
meas.
Fn
7‐Thermistor
Ft
Fig. 1. Principle of the tribometer.
As a consequence, the characterization of the
frictional properties at the tool/chip/workpiece
interface necessitates performing friction tests in the
following conditions: sliding velocity: 60-200
m/min, pressure: 1-3 GPa.
In this work, five sliding velocities have been
selected: 60-90-120-150-200 m/min. A normal force
equal to about 1000 N has been applied onto pins
Φ9mm in order to have an average contact pressure
around 2600 MPa. Contact pressures are obtained by
the ratio between the force F and the semi-circle
contact surface.
3 EXPERIMENTAL RESULTS
The first output data provided by this set-up is the
ratio between the tangential and the normal force.
This coefficient can be defined as an apparent
friction coefficient:
µapp=Ft/Fn (1)
The term ‘apparent’ is used since it is of evidence
that, under such high contact pressure, an intense
plastification of the steel occurs. As a consequence,
the tangential force originates from two phenomena:
adhesion and plastic deformation. It is impossible to
discriminate easily the proportion of both
phenomena. A numerical model of this friction test
under such extreme conditions has been developed
by [8] in order to perform this work. It has been
revealed that, in the case of treated AISI4140 steel,
the adhesive friction coefficient can be estimated
approximately by Eq. (2) for any testing conditions:
µadh ~ 0.8*µapp (2)
So, from a qualitative point of view, the analysis of
the results obtained for the apparent friction
coefficient µapp is sufficient to bring conclusions
about the influence of coatings on the intensity of
adhesion.
The evolution of µapp versus sliding velocity is
plotted in Fig. 2 for the three contact pressures. It
shows that friction is strongly influenced by sliding
velocity. This confirms that it is not relevant to
consider the friction coefficient as independent from
cutting speed. It is also clear that over 200 m/min,
the slope becomes much smaller which indicates a
modification of the frictional behaviour. The system
also confirms the great improvement of the friction
coefficient with a coating deposited on carbide pins,
compared to uncoated pins. This result was expected
since several papers have already shown the
reduction of adhesion induced by coatings [2;6].
Among the investigated coatings, TiAlN and AlCrN
coatings exhibit the best behaviour. TiN coating lead
to friction coefficients very close to TiAlN and
AlCrN.
The second output data provided by the set-up is the
heat flux transmitted to the pin φpin. The evolution of
heat flux versus sliding velocity is plotted in Fig. 3.
It appears that heat flux increases with the velocity,
which is coherent with the fact that more energy is
produced and has to be dissipated. It appears that
TiN coating enable a higher transfer of heat flux to
pins which is correlated to its higher friction
coefficient.
For each curve, two part can be distinguished can be
distinguished. For uncoated pins, the modification of
the behaviour starts since 100 m/min. This
modification is clearly explained by the rapid wear
of uncoated pins. This shows that uncoated tools
should be used with a local sliding velocity lower
than 100 m/min when machining treated AISI4140
in dry conditions.
Workmat.: AISI4140 ( 300 HB)
Pins: carbide
Pin’s diam.: 9 mm
Normal force: ~1000 N
Dry
Pressure ~ 2600 MPa
Fig. 2. Influence of sliding velocity for various coatings.
Workmat.: AISI4140 ( 300 HB)
Pins: carbide
Pin’s diam.: 9 mm
Normal force: ~ 1000 N
Dry
Fig. 3. Evolution of the heat flux transmitted to pins.
Concerning coated pins, below 200 m/min, the slope
is smaller, whereas it increases significantly over
this value. The modification of the frictional
behaviour over 200 m/min is not fully clarified for
the time being, since no wear has ever been
observed for any sliding velocity. Some ideas can be
proposed: the frictional behaviour may be modified
over a certain value (activated by a certain level of
temperature for example), the heat transfer may be
no more possible over a certain level (saturation). A
more detailed analysis at the local scale is necessary
to understand such phenomena. However it is
possible to predict that this increase of heat flux may
lead to a dramatic of the temperature in the
corresponding cutting, which enable us to assume a
increase of wear rate.
The third output of such tests concerns the
observation of pins after each test. Observations
reveal clearly that coated pins have an adhesive
layer for low sliding velocities. This layer tends to
disappear when increasing sliding velocity, which is
coherent with the significant reduction of the slope
in Fig. 2. This phenomenon is schematized in Fig 4
in the case of TiN coating,
Material flow
TiN
Adhesion (layer)
Vs = 60 m/min
TiN
Adhesion (traces)
Vs = 200 m/min
Fig. 4. Influence of sliding velocity on adhesion
4 CONCLUSIONS
This paper has presented the development of a new
tribometer dedicated to the characterization of
friction coefficients along the tool/chip/workpiece
contact in cutting. This system enables to reach
pressure and velocities relevant with cutting
applications. This system has been applied to the
case of the machining of a treated AISI4140 steel in
dry conditions with various uncoated and coated
carbide tools: TiN, TiAlN AlCrN.
The set-up has shown that the presence of a coating
is the predominant parameter on friction coefficient.
The sliding velocity of the workmaterial around the
cutting tool (rake face and flank face) is the second
most important factor. Higher sliding velocities lead
to lower friction coefficient and lower adhesion on
pins, but to more intense heat fluxes transmitted to
tools. So a compromise has to be chosen between
adhesion and heat flux generation.
TiAlN and AlCrN coatings have similar frictional
properties and exhibit the best frictional properties in
dry conditions. The TiN coating has a somewhat
worse behaviour Uncoated tools lead to much higher
friction coefficient than any coated tools.
Based on the macroscopic informations delivered by
this set-up, a perspective of this work should consist
in analyzing the frictional phenomena at the local
scale and to bring adhesive friction models usable in
analytical or numerical cutting models.
REFERENCES
1. T. Özel, The influence of friction models on finite
element simulations of machining, International Journal
of Machine Tools and Manufacture 46 (2006) 518-530.
2. J. Rech, A multi-view approach to the tribological
characterization of cutting tool coatings for steels in high
velocity dry turning, International Journal for Machining
and Machinability of Materials 1 (2006) 27-44.
3. E.M. Trent, Metal Cutting, Butterworth Heinemann,
1991, ISBN 0-7506-1068-9.
4. D. Buryta, R. Sowerby, I. Yellowley, Stress distributions
on the rake face during orthogonal machining,
International Journal of Machine Tools and Manufacture
34 (5) (1994) 721-739.
5. W. Grzesik, An integrated approach to evaluating the
tribo-contact for coated cutting inserts, Wear 240 (2000)
9-18.
6. J.Rech, Influence of cutting tool coatings on the
tribological phenomena at the tool-chip interface in
orthogonal dry cutting, Surface and Coatings
Technology, 200 (2006) 5132-5139.
7. W. Grzesik, Z. Zalisz, P. Nieslony, Friction and wear
testing of multilayer coatings on carbide substrates for
dry machining applications, Surface and Coatings
Technology 155 (2002) 37–45.
8. F.Zemzemi, J.Rech, W.Ben Salem, P.Kapsa, A.Dogui,
Development of a friction model for the tool-chip-
workpiece interface during dry machining of AISI4142
steel with TiN coated carbide cutting tools, International
Journal for Machining and Machinability of Materials
Vol.2, Is.3-4 (2007) 361-367.
9. P. Hedenquist, M. Olsson,’Sliding wear testing of coated
cutting tool materials‘, Tribology International 23 (3)
(1991) 143–150.
10. F.Zemzemi, W.Bensalem, J.Rech, A.Dogui, P.Kapsa,
New tribometer designed for the characterization of the
friction properties at the tool/chip/workpiece interfaces
in machining, Tribotest, (accepted for publication)
11. A.Kusiak, J.L.Battaglia, J.Rech, Tool coatings influence
on the heat transfer in the tool during machining, Surface
and Coatings Technology 195 (2005) 29-40
12. F.Zemzemi, PhD Thesis, Ecole Centrale de Lyon, dec.
2007.
Machining and surface integrity of polymeric materials
C. Fetecau1, F. Stan1, A. Munteanu1, V. Popa1
1
Dunarea de Jos University of Galati, Faculty of Mechanical Engineering
Domneasca 47 Street, Galati, 800 008, Romania
URL: www.ugal.ro, e-mail: catalin.fetecau@ugal.ro

ABSTRACT: In this paper, measurements of cutting force and temperature developed during orthogonal
cutting at different cutting speeds, feed rates and depths of cuts were carried out. During the experiments the
cutting force was measured using Hottinger strain gages, while the temperature distributions were measured
using an infrared Thermo Vision A20 camera. The results showed that the cutting speed had the most
influence on the cutting force and temperature distribution.

Key words: Thermoplastic material, Cutting force, Temperature distribution

machining will raise the temperature in the cutting

1 INTRODUCTION
In the last years there is an increasingly range of
application of plastic materials. Examples are cars,
mobile telephones, etc. Plastic materials are usually
manufactured by various moulding processes such
as injection moulding, extrusion or compression
moulding. However, machining of polymeric
materials becomes necessary for small quantity
production when the cost of moulding tools and dies
is very high, or for customer-specific when a
complex and accurate shape is required.
The machining characteristic of plastics depends
primarily on their mechanical, thermal and
rheological properties [1, 2, 3]. The main factors that
influence the quality of a machined component are:
temperature stability, slide straightness, tool design
and machining conditions such as rake angle, tip
radius, depth of cut, feed rate and cutting speed, and
the physical and chemical properties. For example,
when cutting at low speed, the viscous part of the
polymer is applied and a smooth and ductile chip
can be obtained. By increasing the speed, the elastic
part of the material behaviour will become more
important resulting in a more brittle process [3].
Also, temperature plays a vital role in the machining
of plastic material as it directly influences the rate of
tool wear and the final surface finish of a workpiece
material. The heat generated during the polymer
zone. If the temperature in the cutting zone exceeds
the glass transitions temperature of a polymer, a
good surface finish can be achieved and the
materials would be removed in a ductile manner. On
the other hand, due to the increase of the
temperature in the cutting zone the mobility of the
materials molecular chains will increase and thus
reduce the shear stress and tensile strength [3, 4].
During the plastic deformation of the polymers the
physical aging which cam lead to an increase in
initial yield stress must be taken into account.
Although some efforts have been made to
understand the machinability of the polymeric
materials [3, 4], any evaluation of machining
characteristics must allow for the particular
properties of the material being used. Each category
of plastic material demands different tool geometries
and cutting conditions for a great acting and
productivity.
In this paper, the cutting force and temperature
distribution during the orthogonal cutting of three
thermoplastic materials (CESTILENE HD 500,
ERTALYTE and ERTALON 6-SA) are investigated
for a range of cutting speeds, feed rates and depths
of cuts. The cutting force during machining is
measured based on Hottinger strain gages while the
temperature distributions are investigated using
infrared thermography. Also, the surface finish is
analyzed by the surface image analysis.
2 EXPERIMENT The strain gage has the following characteristics:
measuring grid length of 10 mm, nominal resistance
2.1 Experimental set-up and worpiece materials of 120 Ω, and maximum perm. of effective bridge
supply voltage of 4 V.
Orthogonal cutting was performed using a CNC The force signals were captured and processed using
machine at a series of cutting speeds, cutting depths, the data acquisition system based on hardware
and feed rates at a fixed rake angle of +10°. The SPIDER 8 which consists of eight digital amplifiers
cutting parameters are given in Table 1. operation in frequency of 4.8 kHz.
The calibration was made by applying weights and
Table 1. Cutting conditions
Cutting speed, v m/min
the strain values were recorded for each weight. The
23.864, 47.728, 95.456, 149.15 calibration curve was obtained to convert the output
Depth of cut, t mm readings into cutting force values.
0.5, 1.0, 1.5, 2.0 An infrared ThermoVision A20-M camera was used
Feed rate, s mm/rev for the thermal experiments (fig. 1). This camera has
0.024, 0.125, 0.208, 0.302
a temperature sensitivity of 0.10°C in a range from -
20°C up to +900°C. It produces high-resolution
The experimental set-up used to measure the cutting
images (160x120 pixels) offering more than 19,000
force and the temperature distribution is
individual measurement points per image at 50 Hz.
schematically presented in figure 1.
The cutting experiments are conducted for three
materials: CESTILENE HD 500, ERTALYTE and
Work piece
ERTALON 6 SA which are used extensively in
mechanical, chemical and electrical applications.
These materials offer an optimal combination of
Strain gage F mechanical strength, stiffness, toughness,
IR Camera
44 27
mechanical damping properties and wear and
abrasion resistance.
Strain gage

Data Acquisition 2.2 Force measurements

System

Computer
Fig. 1. Experimental set-up
The cutting tool was tungsten carbide cemented K20
(according to ISO 2004) without chip breaker. The
geometry of the cutting tool is presented in figure 2.
Fig. 2. Geometry of the cutting tool
The cutting force during machining is measured
based on Hottinger strain gages. The gage is
installed on the cutting tool at the point and direction
where the strain is to be measured (figure 1).
The relationship between the cutting force and the
cutting parameter were examined for CESTILENE,
ERTALYTE and ERTALON 6 SA, respectively.
It should be noted that the chips formed during the
machining are discontinuous types in scrap form.
Figure 4 shows the variation of the main cutting
force depending on the cutting speed for two
different depths of cuts and four different feed rates.
Generally, the results show that increasing cutting
speed, feed rate and depth of cut resulted in increase
the main component of the cutting force.
For CESTILENE, at lower cutting speed, when the
feed rate is increased to 0.302 mm/rev from 0.024
m/rev, the cutting force increases between about
122% and 434 % for 0.5 and 2 mm depths of cuts,
respectively.
At high cutting speed, increases in the feed rates
resulted in 176% increases in cutting force at 0.5
mm depth of cut while at 2 mm depth of cut the
cutting force increased by 327%. The same trends
were observed during the machining of ERTALYT
and ERTALON. However, the average cutting force
is slightly higher for the machining of ERTAON.
 50
G s=0.024 mm/rev C s=0.208 mm/rev Through the analysis of the graphs presented in
E s=0.125 mm/rev S s=0.302 mm/rev
figure 4, we can see that the feed rate has a less
Cutting force [N]

40
30
S influence on the cutting force during the machining
S
S C
with cutting speed in the range of 23.864-
20 S C C
E E
C
E E
G 149.15m/min.
10 G G G
If the cutting force is examined for different feed
0
0 20 40 60 80 100 120 140 160 rates, it is seen from figure 5 that the cutting forces
Cutting speed [m/min] t=0.5 mm increase when the feed rate is increased.
100
Cutting force [N]

S CESTILENE S CESTILENE
80 S
S G
D ERTALYTE I ERTALYTE
S
60 C 50 C ERTALON E ERTALON

Cutting force [N]

C E
40 C
E
C
E
E
40 I v=149.15 m/min
E
I S
20 G G
G 30 E
S D v=23.864 m/min
G I
E C
0 20 S
D
C
D
C
G
G
0 20 40 60 80 100 120 140 160 I
E
S
G
10 DC
G
Cutting speed [m/min] t=2 mm
0
a) CESTILENE HD 500 0 0.1 0.2 0.3 0.4 0.5
Feed rate [m/rev] t=0.5 mm
G s=0.024 mm/rev C s=0.208 mm/rev
50 E s=0.125 mm/rev S s=0.302 mm/rev
100
Cutting force [N]

Cutting force [N]

S
40 I
S v=149.15 m/min
E
S C 80
S
30 S C
60 I
I
E
D
C v=23.864 m/min
G
C E E S
E
20 C
E E
40 I
S
D
C
G
G D
C
G
G
10 G G
20
E
D
S
C
G
0 0
0 20 40 60 80 100 120 140 160 0 0.1 0.2 0.3 0.4 0.5
Cutting speed [m/min] t=0.5 mm Feed rate [mm/rev] t=2 mm
100
Cutting force [N]

S S
Fig. 5. Variation of cutting force versus feed rate
80 S
S C
60 C E When the cutting speed was increased to 149.15
40 C
E
C
E
E
G
G m/min from 23.864 m/min at a 0.302 mm/rev feed
G
20 G
rate, an increase of about 50-60% in the cutting
0 force was observed at 0.5 mm depth of cut while an
0 20 40 60 80 100 120 140 160 increase of only 30-35% in the cutting force was
Cutting speed [m/min] t=2 mm
observed at 2 mm depth of cut.
b) ERTALYTE For each combination between cutting speed, feed
50
G s=0.024 mm/rev C s=0.208 mm/rev rate and depth of cut, scanning electron microscopy
E s=0.125 mm/rev S s=0.302 mm/rev
(SEM) analysis was performed using a scanning
Cutting force [N]

40 S
electronic microscope system Quanta 200. The study
30 S C
S S C E of the SEM images reveals a change in the topology
20 C E
C
E E
G of the surface; however, the modification can be
10 G G G
localized at the surface, allowing surface properties
0
0 20 40 60 80 100 120 140 160 to be tailored without altering the bulk properties.
Cutting speed [m/min] t=0.5 mm
100 a b
Micro cuttings Micro cuttings
Cutting force [N]

S
80 S
S
S
60 C
C
E
C E
40 C
E E
G G
20 G G
0
0 20 40 60 80 100 120 140 160
Cutting speed [m/min] t=2 mm
c) ERTALON 6 SA Fig. 6. SEM imagines for CESTILENE at the cutting speed of
23.864 m/min and a) feed rate 0.024 mm/rev b) 0.302 mm/rev
Fig. 4. Variation of cutting force versus cutting speed
2.3 Thermal aspects
For the investigation of the influence of cutting
speed on the temperature distribution, cutting
experiments were performed at three different
cutting speeds (23.864, 95.456, and 149.15 m/min),
two different depths of cuts (0.5 and 2 mm) and two
different feed rates (0.024 and 0.302 mm/rev).
Figure 7 shows a typical image obtained with the IR
camera for machining of CESTILENE at cutting
speed of 149.15 m/min, feed rate 0.024 mm/rev and
depth of cut 2 mm. The temperature was measured
at the interface workpiece-cutting tool (see figure 1).
The increase in the temperature with the increase of
the cutting speed is not uniform and was found to
change between 7 % and 32.5% at 0.5 mm depth of
cut. Further increases in the depth of cut lead to an
increase of low order (3.5 - 20%) in the temperature.
At the low cutting speed and 0.5 mm depth of cut,
increases in the feed rate resulted in ~ 32.5%
increases in the temperature for CESTILENE and
only ~ 10% for ERTALYT and ERTALON, while at
the high speed, increase in the feed rate resulted in ~
16.4% increases in the temperature for CESTILENE
and only ~ 7% for ERTALYT and ERTALON. The
results reveal that the feed rate has a much more
influence on the temperature than the depth of cut.

3 CONCLUSIONS

In this paper, the cutting force and temperature

Fig. 7. The IR image captured during the machining of
CESTILENE
Figure 8 shows the temperature measured when
machining at different cutting speeds, two different
feed rates (0.024 and 0.302 mm/rev) and two
different depths of cuts (0.5 and 2 mm).
From these graphs it can be seen that an increase in
cutting speed in the range of 23.864- 149.15 m/min
results in a temperature rise in the cutting zone.
G CESTILENE, t=0.5 mm
100
Temperature [ o C]
distributions depending on the cutting parameters
were investigated. If the variation of the cutting
force is interpreted in terms of cutting speed and
depth of cut, it is seen that the increase in cutting
speed and feed rate caused an increase in cutting
force. When the temperatures are examined in terms
of cutting speed it is seen that the increase in cutting
speed caused an increase in the temperature. The
increase in the temperature with the increase of the
cutting speed is not uniform in the range of 23.864-
149.15 m/min. Surface analysis showed that the
topography and the surface finish undergo changes;
however the surface roughness can be improved by
increasing the cutting speed.
S t=2 mm

I ERTALYTE, t=0.5 mm D t=2 mm

80 C ERTALON, t=0.5mm E t=2 mm ACKNOWLEDGEMENTS

E
D
60 E
D C
E
C
C
I
I
S
G
The authors would like to thank the ANCS for financial
40 D
I
S
S
G support within the scope of the project CEEX M1 62/2006.
G
20
0 REFERENCES
0 20 40 60 80 100 120 140 160
Cutting speed [m/min] s = 0.024 mm/rev
1. Alauddin, M., Choudhury, I., El Baradie, MA., Hashmi,
G CESTILENE, t=0.5 mm S t=2 mm MS., “Plastic and their machining: A review” Journal of
I ERTALYTE, t=0.5 mm D t=2 mm Materials Processing Technology, 54 (1995) 40-46.
100
Temperature [o C]

C ERTALON, t=0.5mm E t=2 mm

80 2. Roy, PK., Basu, SK., “Evaluation of processing factors

E
E
D
C on turning of thermoplastics” Polymer Engineering and
60 D
C I
E
C
D I
S
S
G Science, 17:10 (1977) 751-757.
40 S
I
G
G
3. Xiao KQ., Zhang L., “The role of viscous deformation in
20
the machining of polymers, International Journal of
0 Mechanical Sciences 44 (2002) 2317-2336.
0 20 40 60 80 100 120 140 160
Cutting speed [m/min] s= 0.302 mm/rev 4. Smith EF., Single-point turning amorphous polymers,
MS Thesis, North Carolina State University, NC (1989).
Fig.8. Temperature variation as a function of cutting speed
Cutting applied as test method for acquisition of flow stress data at high
strains and strain rates
S.M. Ganesan1, Y.A. Khan2, H.Valberg3, P. T. Moe4, A. Willa Hansen5
1,2,3,4,5
Department of Engineering Design and Materials, Norwegian University of Science and Technology,
Richard Birkelands vei 2B, 7491 Trondheim, Norway
URL: www.immtek.ntnu.no

E-mail: ganesan.s.m@ntnu.no1; yawar.abbas@ntnu.no2; henry.valberg@ntnu.no3;

Per.T.Moe@ntnu.no4; Arnfinn.Willa.Hansen@ntnu.no5

ABSTRACT: In rapid metal forming processes, very high strain and strain rates are experienced. It is
therefore important to determine the flow stress of metals under such conditions. In the cutting process, it is
well known that high strains occur in a localized shear zone formed as the shearing punch penetrates down
into the material to be cut. In rapid cutting, very high strain rates are also experienced in the shear zone.
Cutting was therefore investigated to see if the process is suitable as technological test for acquisition of flow
stress data for metals. It is shown how approximate RT flow stress data can be obtained in partial cutting tests
if the flow stress of the sheet obeys the power-law.

Key words: Shearing, Cutting, Blanking, Technological test, Flow stress, High strain, High strain rate

resulting flow stress agrees well with same flow

1 INTRODUCTION stress determined in tensile and compression testing.

Metal cutting distinguishes itself from other metal

processing methods due to its extreme deformation 2 PRINCIPLE OF SHEARING
conditions in a small volume, i.e. intense strain in a
thin layer for a short duration [1]. In metal forming
and in cutting, it is important to predict the flow
stress of the metal with high accuracy, if the
processes should be modeled with good accuracy
using FEM-analysis
Generally the flow stress of a metal depends on
many processing factors like strain, strain rate,
microstructure, temperature, etc. But in many cold
forming operations, the flow stress of a metal is
mainly dependent on the strain, while the influence
of strain rate and temperature is rather modest.
Only a small fraction of the work, required to
Fig.1. Force components in shearing operation
deform the metal is stored in the microstructure. The
rest is converted into heat causing adiabatic shear Force components in shearing operations are shown
bands that could lead to localized softening of the in Fig.1. Forces Fv and Fv’ caused by compressive
metal [2]. Since the strain found in the cutting stresses, act at small distances away from the cutting
process is rather large, it is necessary that the strain edge. The distance between these forces, l, creates a
in the tests is of the same order, to predict flow moment which tilts or bends the specimen. It is
stress to apply in finite element analysis [3]. In counter-balanced by horizontal forces FH and FH’.
cutting, there can also be very high strain rates, The horizontal forces cause frictional forces on the
reaching up to 10 3 - 10 6 [4]. side surfaces of the tooling, µFH and µFH’. Fv and
In this work a procedure is presented for obtaining Fv’ also cause frictional forces µFv and µFv’ [5].
the flow stress of power-law strain-hardening Fig.1 gives an overall idea about forces acting on
materials in a simple partial cutting operation. The sheet and tool during cutting.
procedure is applied to measure the flow stress of a
soft annealed Al alloy at room temperature. The

1
 3 EXPERIMENTAL SET UP
(a) (b)
Fig.2a) and b) show the Karl Kaesler Coburg
eccentric press and its punch portion used to carry
out our experiments. It has maximum capacity of
1400KN load and 60mm stroke.

3.1 Velocity characteristics

Velocity profile of the punch with respect to its

Fig. 2. (a) Eccentric hydraulic press with its (b) punch and die
Forward Stroke (FS) is shown in Fig. 3. This profile
reveals that the punch has a range of velocities in its
forward stroke, i.e. highest velocity at starting of FS,
which gradually drops down to zero at end of the FS,
where after the Back Stroke (BS) immediately starts.
So specimen with high thickness is subjected to
higher cutting speed during early stages of cutting
compared to specimen with low thickness, see Fig.4.

4 EXPERIMENTAL DETAILS

Two sets of specimens were used in the experiments.

Fig.3. Velocity profile of the punch with respect to its
Forward Stroke (FS)
Table.1 reveals properties of the specimens. While
the specimen in Exp.1 and 2 consisted of a hot-
extruded strip, in Exp.4 to 7 it consisted of a cast
billet, from which discs of different thickness were
cut. In all cases material was of the alloy AA6082.
The bottom die was placed at a certain height below
the bottom position of the upper punch to get a
partial cut, see Fig.5. The upper punch moved
backwards before the cut had been completed. In
this way we got specimen samples without complete
separation of slug from specimen, what is good for
study of shearing conditions in cutting.
All experiments were conducted at room
temperature with a lubricant consisting of MoS2
paste that was applied on both sides of the specimen.
Deformation load and ram displacement were
recorded.

Fig.4. Starting velocities of the punch depend on specimen
thickness
Table 1. Properties of the specimen used in this work
5 EXPERIMENTAL RESULT
The load-stroke curves obtained for specimens of
different thickness are shown in Fig.6

Exp. Thickness Specimen Punch

no (mm) shape penetration
(mm)
1-3 3.3 Square plate 1.20
4 3.3 Round plate 1.20
5 5.2 Round plate 2.90
6 7.5 Round plate 4.64
7 10.3 Round plate 7.40 Fig.5 Photograph showing partial cutting of two samples.
Exp.no.1 at top and Exp.no.5 at bottom

2
 160000 shown in Fig.7 as a family of curves.
10.3
140000

120000

7.5
100000
Load (N)

80000 5.2

60000
3.3
40000

20000

0
0 2 4 6 8 10
Stroke (mm)

Fig. 7 Load vs. stroke curves for different values of strength

Fig.6 Load-stroke curves of specimen at different thickness coefficient (K) and strain hardening exponent (n).

In Exp.no.1 to no.3 load stroke curves were equal.
Hence the reproducibility in our experiments was
good. This load-stroke curve is shown in Fig.10.
The curves rise steep initially with increasing punch
movement, reaching a maximum. Then there is a
drop in load as the punch penetrates further. They
agree well with load stroke curves reported
elsewhere.
6 FEM ANALYSIS
Isothermal FEM-models of cutting corresponding to
the experiments were built using the DEFORM 2D®
program to study conditions in the experiments. The
specimens were modelled either as plastic or elastic
plastic with flow stress defined by the well-known
power-law [6],
σ = Kε n
A series of FEM-simulations were also conducted to
study how the load stroke curve is predicted to
depend on the two material parameters K and n, i.e.
the strength coefficient and the strain hardening
exponent. In these simulations, all cutting
parameters were kept constant except for the above
mentioned parameters, which were varied
systematically as shown by the legend given in
Fig.7. The following conditions were kept constant:
sheet as plastic, dies as rigid: t=6.5mm,
Rblank=67.5mm, Rpunch=33mm, Rdie 33.13mm,
v=130mm/s, μ=0.6.
While considering these curves in more detail, we
found that independent of the K-value of the sheet,
i.e. whether it was 100, 222 or 300, the peak values
of the curves were at the same stroke length for one
particular value of n. But when the n-value increased
the peak was shifted towards higher stroke value.
On the other hand, when considering a constant
value of n there was a shift of the peak value
towards higher level of maximum punch load.
Hence for a power-law material, the simulation
results indicate that stroke length at peak value of
the graph is related only to n, i.e. it is independent of
K. Based on this observation the graph shown in
Fig.8 was made on basis of the same simulated data.
This is a graph from which one can obtain the strain
hardening exponent if one knows the stroke length
of a particular unknown power-law material, if it is
present as sheet in same thickness as that applied in
(1) the FEM-analysis. In a corresponding way the curve
shown in Fig.9 was made from the simulations. This
graph is used to find the K after the n-value has been
obtained from the first graph.

7 PROCEDURE TO OBTAIN POWER LAW

CONSTANTS FROM CUTTING TESTS

Different load-stroke graphs obtained in the

simulations for the different material parameters K
and n were plotted into the same diagram and is Fig. 8 Graph obtain from simulation to deduce n-value

3

Fig.9 Punch load at peak value vs. n-value. K-value is
deduced from curve.
8 APPLICATION OF PROCEDURE ON
PARTICULAR MATERIAL
The mentioned procedure for determination of K and
n was then applied on the strip material investigated
in Exp.1-3. The measured load-stroke curve of this
metal is shown in Fig.10. A simulation model was
composed for this case with input data
corresponding to experimental data. Simulations
were run for different K and n values and
corresponding graphs as those in Fig.8 and Fig.9
were made such that they were valid for the actual
cutting conditions in those tests.
When this procedure was used we obtained the
following data for the alloy: K = 185MPa and
n = 0.13 . This agrees reasonably well with flow
stress data obtained for the same material in tensile
and compression testing. Finally, as Fig.10 depicts, a
final FEM-simulation of the experiment with this
flow stress data, gave good agreement between the
load-stroke curve in simulation and experiment.
60000
50000
Simulated
40000
Load (N)
30000
Experimental
20000
10000
0 stress in the primary shear zone’, Journal of Material
0 0.2 0.4 0.6 0.8
Displacement (mm)
Fig.10 Load vs. Stroke curves of experiment and simulation in
Exp. no.1.
9 DISCUSSION
The partial cutting test reported here is a very cheap
and rapid test for testing of material from sheets or
plates. Flat pieces can be tested very quickly without
cumbersome and expensive preparation of test
specimens. Data treatment when the procedure
proposed by us is applied for determination of the K
and n values of power-law materials is also easy and
rapid.
In our first trial using this new method we got
reasonable agreement between flow stress data
obtained in the partial cutting test and traditional
recordings of flow stress data for the same material
obtained in tensile and compression tests.
In future, more work will be conducted to map the
accuracy of the technique we propose and to study
how thermal and strain rate effects influence the
result.
10 CONCLUSION
The partial cutting test method proposed by us can
be applied for rapid approximate determination of
flow stress data for sheet and plate materials. When
applying the test method, sample preparation is very
simple and cheap. In our first application of the test
method for acquisition of flow stress data, we
determined the strength coefficient and the strain
hardening exponent of a material with reasonable
accuracy.
ACKNOWLEDGEMENT
Acknowledgements are due to The Research Council of
Norway for financial support through project no.168188/140.
REFERENCES
1. S.N. Nasser, Y.F.Li and J.B. Isaacs, ‘Experimental/
computational evaluation of flow stress at high strain
rates with application to adiabatic shear banding’,
Mechanics of Materials 17 (1994), 111-134.
2. S.P.F.C. Jaspers and J.H. Dautzenberg, ‘Material
behaviour in metal cutting: strain, strain rates and
temperatures in chip formation’, Journal of Materials
Processing Technology 121 (2002), 123-135.
3. M.N. Bassim and N. Panic, ‘High strain rate effects on
the strain of alloy steels’, Journal of Material Processing
Technology 92-93 (1999), 481-485.
4. S.F.P.C. Jaspers and J.H. Dautzenberg, ‘Material
behaviour in conditions similar to metal cutting: flow
1 1.2 1.4 Processing Technology 122 (2002), 322-330.
5. Kurt Lange et.al, ‘Hand book of Metal Forming’, Part4:
Blanking, piercing, slug production, page. 24.6
6. G.E.Dieter, ‘Mechanical Metallurgy’, Part4: Plastic
Forming of Metals, Chapter 20: Sheet-Metal Forming,
657-659.
4
Laser-assisted machining of Inconel 718 with carbide and ceramic inserts
G. Germain1, J.L. Lebrun1, T. Braham-Bouchnak1, D. Bellett1, S. Auger2
1
ENSAM CER d’Angers, LPMI/EPPM-EA1427, 2 bd de Ronceray 49035 Angers, France
URL: www.ensam.eu e-mail: guenael.germain@angers.ensam.fr
2
CETIM site de Senlis, . 52 Avenue Félix-Louat 60304 Senlis, France
URL: www.cetim.fr e-mail: stephane.auger@cetim.fr

ABSTRACT: Laser assisted machining (LAM) can improve the machinability of materials by locally heating
the material prior to its removal. The work presented here is a study of the laser assisted machining of Inconel
718 (NiCr19FeNb at 46 HRc) with carbide and ceramic insert. The tests have shown a reduction in the cutting
force, and have highlighted the impact of laser assistance on the integrity surface (roughness, appearance,
residual stress) and the tool life.

Key words: LAM, Inconel 718, superallows, machinability, tool life, tool wear, surface integrity

Laser Beam
Faisceau Laser

1 PRESENTATION

1.1 Introduction
Outil de coupe
Cutting tool
The use of high strength materials, like titanium
alloys and tool steels, is becoming increasingly
Pièce
common in industry. However, these materials are Workpiece

difficult to machine because they maintain their

mechanical properties even at high temperatures.
Hence, conventional machining (CM) of these Fig. 1: LAM configuration (turning operation)
materials is slow and inefficient because only slow
For most materials a drop in the tensile strength or
cutting speeds can be used. In order to increase
hardness occurs near 500°C. To be effective the
productivity certain types of ‘assistance’ can be used
cutting tool must thus operate in the zone where the
to facilitate the cut. It has already been shown that
temperature remains higher than this value.
LAM makes it possible to machine high strength
2000
metals like bearing steel [1], Ti6Al4V [2, 3], boron
Résistance à la traction (MPa)
Ultimate tensile strength (MPa)

1800 Ti6Al4V
[4], metal matrix composites [5] and ceramics [6-7]. 1600
1400
For these materials, it improves workability by 1200
NiCr19FeNb -Inconel 718-
decreasing the cutting forces and by increasing the 1000 35NiCrMo4

tool life. This process is currently the only process 800

Si3N4
600
able to machine very hard materials. 400
200

1.2 Principle of LAM 0

0 200 400 600 800 1000 1200 1400
Température (C°)

LAM is a high temperature cutting process using a
laser beam as the heat source (Figure 1). The
principle of the process is to reduce the cutting force
necessary to machine the material by increasing the
temperature to the point where the strength of the
material is reduced (Figure 2). Indeed at high
temperature, the specific cutting energy is weak,
which improves workability. Figure 2 shows the
characteristic evolution of the ultimate tensile
strength of various materials with regard to
temperature [3].
Fig. 2: Sensitivity of σUTS to the temperature for various
materials [3]
The effect of the angle of incidence has been
investigated by Germain et al [8-9]. In particular, it
has been shown that the laser beam absorption is the
same for all angles of incidence less than 40°
relative to the surface normal. For all the work
reported in this article, the angle of incidence was
set equal to 20° relative to the surface normal in
order to prevent the laser nozzle from hitting the
tool.
1.3 Experimental equipment
A numerically controlled lathe (REALMECA RT-5)
equipped with a 2.5 kW ROFIN YAG laser was used
in this work. The laser nozzle can be controlled using
three translations and two rotations. A numerical
control system (NUM 1060) allows the control of the
seven independent degrees of freedom. The high-
power laser beam is delivered through a fibre optic
cable to the lathe chamber, where it is focused on the
workpiece surface. During machining, a gas under
pressure (air) is forced through the laser nozzle to
protect the focusing lens from being damaged by
chips. Three components of the cutting force are
measured by a Kistler piezoelectric dynamometer.
of the tool wear, the three components of the cutting
force, and the surface roughness, were measured for
each test. The tool life and residual surface stresses
were also determined.
2 RESULTS OBTAINED WITH THE CARBIDE
INSERT
2.1 Cutting force (carbide insert)
The results show a decrease magnitude of the cutting
force with the laser power (Figure.3). Specifically,
there is a decrease of 6.5% with a laser power of
1165 W and 10.8% with a power of 1975 W.

1800
1.4 Material investigated 1700
1600
1500

Force intensity (N)

1400
The material under investigation is the nickel alloy 1300
Axiale force
NiCr19FeNb which is known commercially as 1200
1100 Tangentiale force

Inconel 718. It has been structurally hardened via a 1000

900
Radiale force
Cutting force (norme)

heat treatment consisting of quenching (from 940 to 800

700
1010 ° C) and structural hardening (720 ° C for 8 hr 600
500
followed by 620 ° C for 8 h). The microstructure is 400
0 500 1000 1500 2000
austenitic with a grain size index of G=10. The Laser power (W)
hardness of the material is uniform and has an Fig 3: Evolution of the three components and the magnitude of
average value of 482 HV (46 HRc). the cutting force as a function of the laser power

1.5 Cutting parameters used with the carbide insert 2.2 Surface Integrity (carbide insert)

Tests were carried out with the cutting parameters
recommended by the CETIM for this type of insert
and with several laser Powers: 0 Watts (conventional
machining), 1165 and 1975 Watts. The cutting
parameters recommended are: Carbide tool insert
KC5525 (Kennametal) ref. CNMG 120412 RP;
advance, f = 0.2 mm/rev; cutting depth, ap = 2.5
mm; cutting speed, Vc = 30 m/min; without
lubrication. Material removal rate of 15 cm3.min-1. A
new cutting edge was used for each test. The laser
beam was focused on the chamfered cut surface
approximately 5 mm from the cutting tool.
The three components of the cutting force, the
surface integrity (roughness, residual stresses), and
The impact of laser assistance on the surface
integrity has been quantified by the evolution of
several roughness criteria and the residual stresses
on the cut surface. The three surface roughness
criteria used are: the arithmetic average (Ra), the
maximum height peak/valley (Rmax) and the
average height peak/valley (Rz). The following
figure (Figure 4) shows that laser assistance can
influence the surface roughness of the workpiece. A
high laser power results in a slight improvement of
the roughness criteria.
Ra
6
Rmax
5 Rz
Roughness (µm)

4
the tool life were measured for each test.
3

1.6 Parameters used with the ceramic insert 2

Similarly, the cutting parameters used for the 0

Without 1165 W 1975 W
ceramic insert are those recommended by the assistance
Laser power (W)
CETIM: CC670 ceramic insert (Sandvik) ref. RNGN
Fig 4: Evolution of the surface roughness as a function of laser
090300; advance, f = 0.18 mm/rev; cutting depth, ap power
= 1.5 mm; cutting speed, Vc = 220 m/min; without
lubrication. Material removal rate of 59.4 cm3.min-1. The residual stresses were determined by X-ray
The tests were carried out without laser assistance diffraction. The analysis was carried out for the test
and with an assistance of 1500 Watts. The evolution without assistance (conventional machining - CM),
and with the maximum laser power. The values results in an increase of the radial and axial
obtained are summarised in the following table components of the cutting force, but not the
(Table1). tangential component.
Table 1. Residual Stresses values on the cut surface 3.2 Surface Integrity (ceramic insert)
LAM
Stress (MPa) CM
1975 W The surface integrity was quantified by the
Axial direction 740 ± 25 520 ± 30
measurement of surface roughness (three criteria:
Tangential direction 435 ± 45 440 ± 45
Ra, Rmax and Rz), by the determination of surface
residual stresses and by examination of the cut
The table shows that the residual stresses on the surface with a scanning electron microscope (SEM).
surface are positive (tensile stresses) for the two The three roughness criteria show the same trend.
tests. The laser assistance decreases the residual Consequently only the Ra criterion is presented in
stress in the axial direction. figure 6.
3,5

2.3 Tool wear (carbide insert) 3,0

LAM (1500W)
Without Laser (CM)

Roughness, Ra (µm)
2,5
These initial results demonstrate an improvement in
2,0
the machinability of Inconel 718 with laser
assistance. However, the improvement is modest and 1,5

there is a significant drop in tool life with the use of 1,0

laser assistance. Indeed, the tool life is only 50sec 0,5

with the laser assistance, compared to approximately 0,0

0 200 400 600 800 1000 1200 1400 1600 1800
2min 50sec in conventional machining. In both Length machined (m)
cases, the insert deteriorates very quickly with an Fig 6: Evolution of the surface roughness (Ra), in CM and
almost instant failure of the cutting edge. LAM, as a function of the distance machined
It can be seen that the evolutions of Ra, in CM and
3 RESULTS OBTAINED WITH THE CERAMIC LAM, are similar up to a machined distance of
INSERT approximately 800 m. After this distance, the
roughness is lower and less dispersed for LAM. For
3.1 Cutting force (ceramic insert) conventional machining after approximately 800 m,
the roughness increase significantly and the data is
The figure below (Figure 5) shows the evolution of widely scattered (scratches due to chips).
the magnitude of the cutting force as a function of
the distance machined for both conventional The analysis of residual stress was performed using
machining and LAM (1500 Watts). The cutting a PROTO X-ray diffraction machine. The following
force in LAM is lower than in CM. Depending on table (Table 2) summarises the values of residual
tool wear, the reduction in the cutting force in UAL stress at the surface for the two types of machining.
is between 40% and 20%.
Table 2: Surface residual stresses
LAM
Stress (MPa) CM
2400
1500 W
Axial direction 160 ± 35 260 ± 35
Cutting force (N)

1900

Tangential direction 270 ± 30 555 ± 45

1400

LAM (1500 W)
900 Without Laser (CM) These values indicate that these cutting parameters
400
result in positive or tensile residual stresses in both
0 200 400 600 800 1000 1200 1400 1600 1800 directions. In addition, LAM generates higher
Length machined (m)

Fig 5: Evolution of the magnitude of the cutting force, in CM

stresses due to thermal effects of laser heating.
and in LAM, as a function of the distance machined These values confirm the results obtained in other
materials [9] and those found during the tests with
Inserts have machined respectively 1500m in LAM carbide inserts.
and 1650m in conventional machining prior to the
collapse of the cutting edge. The beginning of the The machined surfaces were observed with a
deteriorated area appears more rapidly in LAM scanning electron microscope. The figure 7 below
(1200 m) than CM (1400 m). This deterioration shows pictures of the machined surfaces in CM (a)
and LAM (b).
(a)
Fig 7: SEM photos of the machined surfaces in (a) CM and (b)
LAM.
These photos (taken at the same magnification)
demonstrate that the surface quality is considerably
higher in LAM than in CM. In CM, the surface
appears to be torn and dark spots, evenly distributed
over the machined area, can be observed.
Electron Diffusion Spectrometer (EDS) analyses
were conducted to determine the nature of these dark
spots. The EDS analyses show that all of the dark
spots are caused by pollution of the surface by the
tool insert. The presence of these elements (silicon,
nitrogen, oxygen and aluminium) corresponds to a
material deposit from the tool insert. This
phenomenon is only visible on the conventional
machined specimens. The machined parts in LAM
demonstrate no such pollution.
3.3 Tool wear (ceramic insert)
The evolution of the wear on the clearance face, Vb,
is shown in figure 8, for the two types of machining,
as a function of the machined distance. The
evolution of the clearance wear is different for
machining with and without laser assistance. In
conventional machining clearance wear increases
approximately linearly, while for LAM three areas
of change are visible.
0,6
LAM (1500W)
Without Laser (CM)
0,5
Wear, Vb (mm)
0,4
Vb = 0.3 mm
0,3
0,2
0,1
0
0 500 1000
Lenght machined (m)
Fig 8: Evolution of wear (Vb) as a function of the machined
distance
In LAM, at the beginning of machining, up to about
400 m, the wear increases rapidly (run-in). Then
between about 400 and 1000 m, the wear is
relatively stable in LAM. By contrast, from 1000 m
the degradation of the tool is greater in LAM with a
very rapid increase in clearance wear.
For a maximum wear criterion set at Vb = 0.3 mm
(ISO standard), the tool life is 4min 25sec in CM
and 5min 40sec in LAM.
4 CONCLUSIONS
This study, on the laser assisted machining of
Inconel 718 has highlighted significant differences
(b) between the use of carbide inserts and ceramic
inserts. In addition, comparisons with conventional
machining were also conducted. Tests have shown
that, no matter which insert is used, LAM
significantly reduces the cutting force (up to 40%).
The integrity of the machined surface, in terms of
roughness, is not improved with the use of ceramic
inserts in LAM compared to CM. However, this
gain is not visible with carbide inserts. Ceramic
inserts allow a very good performance during the
laser assisted machining, unlike carbide inserts. In
fact, the life of a carbide insert in LAM is
considerably lower than its life in CM, whereas the
life of ceramic inserts in LAM increase by 25%
compared to the life in CM.
REFERENCES
[1] Kainth G.S., Dey B.K., Experimental investigation into tool
life and temperature during hot machining of EN-24 steels,
Bulletin Cercle Etudes des Métaux 14 n°11 (1980) p22.1-22.7
[2] Kitagawa T., Katsuhiro K., Kudo A., Plasma hot machining
for high hardness metals, Bulletin Japon Society of Precision
Engineering Vol.22 n°2 (1988) p145-151
[3] Lesourd B., Etude et modélisation des mécanismes de
formation de bandes de cisaillement intense en coupe des
métaux. Application au tournage assisté laser de l’alliage de
titane TA6V, Thèse Ecole Centrale de Nantes (1996)
[4] Malot T., Usinage assisté par laser du bore, Thèse
Université de Bourgogne (2001)
[5] Wang Y., Yang L.J., Wang N.J., An investigation of laser-
assisted machining of Al2O3 particule reinforced aluminium
matrix composite, Materials Processing Technologies 129
(2002) p268-272
[6] Rebro P.A., Shin Y.C., Incropera F.P, Design of operating
conditions for crackfree laser-assisted machining of mullite
ceramic, International Journal of Machine Tools &
Manufacture 44 (2004) p667-694.
[7] Melhaoui A., Contribution à l’étude de l’usure d’outil de
coupe en usinage asssité par laser et à l’usinabilité d’une
céramique à base d’oxyde de zinc, Thèse de l’ECP (1997)
1500 2000 [8] Germain G., Lebrun J-L., Robert P., Dal Santo P., Poitou
A., Experimental and numerical approaches of Laser assisted
turning, IJFP Vol. 8 - Special Issue 2005, Multiscale
Simulations and Experiments to optimize Material Forming
Processes (2005) p347-361
[9] Germain G., Morel F., Lebrun J-L., Morel A., Huneau B.,
Effect of laser assistance machining on residual stress and
fatigue strength for a bearing steel (100Cr6) and a titanium
alloy (Ti 6Al 4V), Materials Science Forum 524-525, (2006)
p559-574
Impact of cutting conditions on mechanical, microstructural and
electrochemical state in super finish turning of copper
J. Gravier1,2, V. Vignal1, S. Bissey-Breton2, J. Farré2
1
Institut Carnot de Bourgogne, UMR 5209 CNRS-Université de Bourgogne, BP 47870, 21078 Dijon, France
URL: http://icb.u-bourgogne.fr e-mail: jerome.gravier@u-bourgogne.fr; vvignal@u-bourgogne.fr
2
CEA, Centre de Valduc, DFTN, 21120 Is-sur-Tille, France

ABSTRACT: The complex mechanisms acting during cutting affect the mechanical, microstructural and
topographical state of the surface and the near-surface layer of metals. Such changes are important in
controlling the corrosion behaviour of these workpieces in the presence of an aggressive environment. This
study presents an investigation of the influence of cutting parameters on surface and subsurface
characteristics. Electrochemical behaviour of the machined surfaces is analysed using local techniques.

Key words: Finish turning of copper, Residual stress, Roughness, Local electrochemical behaviour

(Oxygen Free High Conductivity). Samples were

1 INTRODUCTION
Precision of surface generated by machining and
surface quality are related to a complex interaction
between the material surface and the cutting tool
generating heat transfer and high deformation. Those
modifications induce residual stresses at the surface
of machined components and modify significantly
the microstructure and the texture in a small volume
close to the surface. Those alterations of the material
state have significant consequences on the material
lifetime by influencing corrosion resistance. In a
previous work, the authors have shown that the
material properties of pure copper (crystallographic
texture, surface stress field, microstructure and
corrosion behaviour) are altered by super finish
machining processes [1]. In the present paper, the
impact of cutting parameters on the surface and
subsurface characteristics of machined copper
samples after super finish turning was studied by
combining XRD analysis, surface observations,
numerical analysis (Principal Component Analysis
PCA and analysis of variance ANOVA) and the
local electrochemical technique.
2 EXPERIMENTAL PROCEDURE
2.1 Samples and surface preparation
Experiments were conducted on pure copper OHFC
machined from rolled bar (cylinder of 80 mm
diameter and 20 mm height). Prior machining was
performed to get smooth surfaces and to ensure that
the depth of cut will be constant during machining.
Note that between these two operations, samples are
not removed from the chuck.
2.2 Surface observations
Surface observations were performed on the
machined surfaces using an optical-reflecting
profilometer AltiSurf 500. Line scans were plotted
along the x3-axis. Average peak-to-valley roughness
values RZ were calculated for each sample. Cross-
section surfaces were observed using optical
microscopy after polishing and chemical etching.
2.3 XRD Measurements
The crystallographic orientation of machined
surfaces was determined using an X-ray (diameter of
about 500 µm) goniometer (D8 Discover with a
detector from Bruker AXS) on the {220}, {111} and
{200} planes. The texture index was calculated from
pole figures using LaboTex software [2]. Residual
stresses along the x1-axis (σcircumferential) and x3-axis
(σaxial) were determined on {311} planes using the
sin²ψ method. The X-ray elastic constants were ½ S2
= 1.22 MPa-1 and S1 = -2.164 MPa-1. The surface
quadratic stress was then calculated:
σ quadratic = σ axial 2 + σ circonferential 2 (1) of the three most influent cutting parameters (tool
nose radius, feed rate and lubrication). Then new
machining tests have been performed defining a
2.4 Local electrochemical measurements complete design of experiment L8 (23) at 2 levels for
each factor. The other cutting parameters are
The local electrochemical behaviour of machined
constant at the lower conditions in Table 1. ANOVA
surfaces (samples 17-24) was studied using the
[6] has been used to quantify the relationships
electrochemical microcell technique [3]. The
between the cutting parameters and the surface and
potentiodynamic polarization curves were plotted at
subsurface characteristics of machined samples.
a scan rate of 1 mV/s in 1M NaClO4, pH=6, at 25°C.
The potential range considered was from -700
mV/SCE to the pitting potential. Note that no pre- 4 EXPERIMENTAL RESULTS
polarisation was applied before these measurements.
All potentials were measured versus a Saturated 4.1 Analysis of the orthogonal plan of L16 (28)
Calomel Electrode (SCE) and the counter electrode
was platinum wire. Numerical values of the surface and subsurface
characteristics are reported in Table 2.
3 METHODOLOGY USED TO ANALYSE THE Table2. Surface and subsurface characteristics of samples 1-16.
EXPERIMENTAL RESULTS at (µm) RZ (µm) F2 (a.u.) σquadratic (MPa)
1 3.6 ± 1.3 1.2 3.4 104 ± 19
2 7.5 ± 5.9 7.5 5.6 157 ± 24
The coordinates used in this study are given in 3 19.9 ± 4.7 13.6 16.4 86 ± 40
Figure 1. 4 5.7 ± 4.8 2.3 4.7 169 ± 27
x1
x2
x1 5 8.6 ± 2.9 6.1 2.4 76 ± 22
Machined
surface Machined Cross 6 14.9 ± 5.9 9.3 3.8 103 ± 24
surface x3 section
x3
surface 7 9.6 ± 4.5 8.1 2.7 86 ± 24
x2 8 18.5 ± 5.9 11.1 6.2 158 ± 28
x2 9 8.9 ± 3.7 3.8 3.3 107 ± 24
x1 circumferential
Cross
section
10 9.8 ± 5.1 7 4.5 204 ± 25
x2 radial
x3 axial
x3
x1 surface
11 13.9 ± 4.0 8.1 7.7 142 ± 72
12 6.5 ± 2.6 3.7 4.7 98 ± 44
Fig. 1. Specimen coordinates and surfaces notation. 13 13.7 ± 4.6 9.6 7.6 102 ± 23
14 7.0 ± 2.8 6.2 3.4 112 ± 21
The influence of eight different parameters has first 15 9.5 ± 4.2 8.5 2.3 103 ± 23
been investigated, namely the machine tool quality 16 22.7 ± 9.5 14.8 5.1 114 ± 48
(MT), the tool nose radius (rε), the edge sharpness
(ES), the tool holder length (TH), the cutting speed PCA was performed from these data and correlations
(VC), the depth of cut (ap), the feed rate (f) and the between cutting parameters and surface and
lubrication (lub). Numerical values of these cutting subsurface characteristics are given in absolute
parameters have been defined to create an values in figure 2(a). According to the results and
orthogonal design of experiments L16 (28) [4] with the significance level, the most significant cutting
16 degrees of freedom and 8 factors at 2 levels parameters are feed rate, lubrication, edge sharpness
(levels are reported in Table 1). and tool nose radius. The edge sharpness which is
very difficult to measure accurately was neglected in
Table1. Levels of the orthogonal plan L16 (28). the following.
MT
rε
ES TH
VC ap f
lub Interactions between surface and subsurface
(mm) (m/min) (mm) (mm/rev)
characteristics are shown in figure 2(b) and reveal an
low 1 0.8 New Short 138 0.05 0.05 Yes
high 2 0.4 Used Long 86 0.3 0.2 No interesting interaction between the affected layer
thickness and the roughness. This interaction
Surface and subsurface characteristics have been displays the powerful opportunity to measure the
determined for each sample: affected layer thickness affected layer thickness using non destructive
at, roughness RZ and XRD analysis (texture index methods.
F2, and surface stresses σquadratic). PCA was carried
out from these data [5] and gives the correlation 4.2 Analysis of the complete plan of L8 (23)
between the cutting parameters and the surface and
subsurface characteristics. PCA allows the selection Most influent cutting parameters (feed rate,
lubrication and tool nose radius) are considered to the value of C is set at –1 with lubrication and +1
built a complete plan of L8 (23), using the without lubrication. From the data in Table 4, the
experimental conditions reported in Table 3. following equations were determined by a least
square regression:
a) co
rre
0,7
thicknessthickness
affected affected a_t Roughness Rz texture index F2 quadratic stress
lat
io
0,6 at = 6.1×A+165.6×B+1.6×C-180.2×AB+0.6×AC
n
fa 0,5
ct
or
s
0,4 -2.0×BC+17.5×ABC-0.3 (2)
in
ab 0,3
so
lut 0,2
RZ = 3.2×A+87.6×B+1.8×C-76.1×AB-2.0×AC
e
va 0,1
lu
es
0,0
-10.1×BC+11.5×ABC-1.4 (3)
significant machine tool tool edge edge tool holder cutting depth of cut feed rate lubrication
radius sharpness

cutting parameters
speed
F2 = -3.0×A+43.0×B-2.9×C-2.5×AB+6.8×AC
b) 1,0 affected thickness
thickness affected a_t
roughness Rz
+103.0×BC-152.5×ABC+3.7 (4)
0,8
texture index F2
int
quadratic stress
er 0,6
ac
tio
n 0,4
σquadratic = -35×A+925×B+151×C+137×AB
fa
ct
or 0,2
s
0,0
-193×AC-2083×BC+2783×ABC+51 (5)
The proposed methodology associated to Equations
-0,2
significant affected
thickness
thickness
affected
a_t roughness
roughnessRz
Rz texture index F2 quadratic
quadratic
stress
stress

Fig. 2. PCA analysis: a) impact of the cutting parameters on the (2-5) provides a useful tool to predict machined
surface and subsurface characteristics in absolute value; b) surfaces properties evolution depending on the
interactions between the surface and subsurface characteristics.
cutting parameters. However, these equations can
Note that the other cutting parameters are only be used in case of super finish turning of pure
maintained at the lower level (Table 1). Numerical copper when using similar cutting conditions.
values of the surface and subsurface characteristics Further investigations have to be performed to
are reported in Table 4. It can be noticed that confirm the predictive ability of Equations (2-5).
samples 20, 21, 22 and 23 show a similar texture
which can be compared to a shear-type texture [7].
5 ELECTROCHEMICAL BEHAVIOUR OF THE
3
Table3. Cutting parameters of the complete plan L8 (2 ). MACHINED SURFACES
rε (mm) f (mm/rev) lub
17 0.8 0.05 Yes 5.1 Effect of the texture on the anodic processes
18 0.4 0.05 Yes
19 0.8 0.1 Yes
20 0.8 0.05 No Polarisation curves (Figure 3) obtained on samples
21 0.4 0.1 No 21, 22 and 23 were compared to quantify the texture
22 0.8 0.1 No
23 0.4 0.05 No
impact. The current density measured in the passive
24 0.4 0.1 Yes domain on sample 23 is significantly lower than on
the polished surface. This shows that the shear-type
Table4. Surface and subsurface characteristics samples 17-24. texture tends to stabilize the material by inhibiting
at (µm) RZ (µm) F2 (a.u.) σquadratic (MPa)
17 3.0 ± 2.3 2.38 1.7 72 ± 23
anodic reactions. Comparing the behaviour of
18 4.8 ± 3.9 2.06 2.6 62 ± 21 samples 21 and 23 demonstrates that the behaviour
19 3.5 ± 2.6 3.76 4.7 116 ± 23 of machined surfaces become heterogeneous with
20 8.3 ± 5.2 2.66 4.9 79 ± 24
21 13.8 ± 13.7 6.04 10.7 112 ± 74
increasing roughness and confirms that roughness
22 10 ± 5.4 3.95 6 138 ± 22 has negative impact on the electrochemical
23 8.9 ± 7.1 3.46 6.5 112 ± 22 behaviour of samples.
24 9.2 ± 4.7 5.19 2.6 159 ± 26
5.2 Effect of the lubrication on the anodic processes
ANOVA technique was then used to determine
quantitatively the relationships between these three Sample 24 exhibits the highest roughness value RZ
cutting parameters and the surface and subsurface (Table 4). Therefore, lubricant (4% oil in water) may
characteristics. Each characteristic is defined as a be trapped in the micro-cavities generated at the
combination of the three cutting parameters (tool surface during machining. After this operation, a
nose radius rε, feed rate f and lubrication, noted A, B heterogeneous layer composed of residual oil is
and C, respectively), the double interactions (AB, AC visible, confirming the previous assumption.
and BC) and the triple interaction (ABC). Note that
 10000 10000
The current density at the steady state was plotted

Current density (µA/cm²)

Current density (µA/cm²)

1000 1000

100 100 versus quadratic stress (figure 4(b)). It appears that

10

1
Cu17
polished
10

1
Cu18
polished
the current density in the passive range increases
0,1
-600 -400 -200 0 200 400
0,1
-600 -400 -200 0 200 400
linearly with quadratic stress.
Potential (mV / SCE) Potential (mV / SCE)

10000 10000
Current density (µA/cm²)

Current density (µA/cm²)

1000 1000

100 100 6 CONCLUSIONS

10 10
Cu19 Cu20
polished polished
1 1

0,1
-600 -400 -200 0 200 400
0,1
-600 -400 -200 0 200 400
The following conclusions can be drawn:
Potential (mV / SCE) Potential (mV / SCE)

10000 10000
• numerical analysis has shown that the most
influent cutting parameters in this case are feed
Current density (µA/cm²)
Current density (µA/cm²)

1000 1000

100

10
Cu21, ridge
100

10
Cu22, ridge
Cu22, groove
rate, lubrication and tool nose radius;
Cu21, groove polished

0,1
1 polished

0,1
1
• correlations between surface and subsurface
-600 -400 -200 0 200 400
Potential (mV / SCE)
-600 -400 -200 0 200
Potential (mV / SCE)
400
characteristics provide an interesting interaction
10000 10000
between the affected layer thickness and the
Current density (µA/cm²)

Current density (µA/cm²)

1000 1000

100 100
roughness;
10

1
Cu23
polished
10

1
Cu24, ridge
Cu24, groove
polished
• quantitative equations were determined from the
0,1
-600 -400 -200 0 200 400
0,1
-600 -400 -200 0 200 400
analysis of the complete plan L8 (23);
Potentiel (mV / SCE) Potential (mV / SCE)

• the local electrochemical behaviour of machined

Fig. 3. Potentiodynamic curves determined on the machined samples has shown that the texture tends to
surfaces of samples 17-24 with 75 µm microcapillaries.
stabilize the material in the presence of an
Current densities determined in the passive range are aggressive solution;
generally very low, but some sites exhibit different • in the passive domain, the current density
current densities confirming the heterogeneous layer increases with increasing quadratic stress and
of residual oil. The beneficial effects of lubrication decreases strongly on rough surfaces which were
were not counter-balanced by quadratic stress. lubricated during machining.
Indeed, the current density on sample 24 is very low
whereas the quadratic stress is around 159 MPa.
ACKNOWLEDGEMENTS
5.3 Effect of the quadratic stress on the anodic The authors would like to thank the Conseil Régional de
processes Bourgogne (France) for financial support.

Samples 17-20 have low roughness and low texture

REFERENCES
index and therefore, they were used to determine the
quadratic stress influence on the electrochemical 1. N. Mary, V. Vignal, S. Bissey-Breton, J. Farré, Impact
behaviour of machined surfaces. For these samples, of machining on the texture, surface stress field,
the current density in the passive domain was microstructure and corrosion behaviour of metals,
Advances in Materials Science, vol. 7, No. 2 (12), (2007)
systematically higher than on the polished sample. 129-140.
To provide a quantitative relationship between the 2. H.J. Bunge, Texture analysis in materials science,
quadratic stress and the current density, Butterworths, London, (1982) 1-41.
potentiostatic tests were performed in 1M NaClO4 3. H. Böhni, T. Suter, A. Schreyer, Micro- and
for 300 seconds, as shown in figure 4(a). Nanotechniques to study localized corrosion,
Electrochimica Acta vol. 40, No. 10, (1994) 1361-68.
4. G. Taguchi, S. Konishi, Taguchi methods, Orthogonal
a) 4 Cu17 Cu19 b) 4
arrays and linear graphs, Tool for Quality Engineering,
Current density (µA/cm²)

Cu18 Cu20
Current density (µA/cm²)

19
3 3
American Supplier Institute, (1987) 35-38.
2 2
20
5. J.D. Carroll, P.E. Green, A. Chaturvedi, Mathematical
1 1
18
17
tools for multivariate analysis, Academic Press, New
0 0

0 50 100 150 200 250 300 60 80 100 120

York (1997).
Time (sec) Quadratic stress (MPa) 6. H.R. Lindman, Analysis of variance in complex
experimental designs, W. H. Freeman & Co (1974).
Fig. 4. a) Potentiostatic curves at -100 mV/SCE on samples 17- 7. M. Haouaoui, K. T. Hartwig, E. A. Payzant, Effect of
20 with 75 µm microcapillaries; b) Relation between tensile strain path on texture and annealing microstructure
stress and the current density. development in bulk pure copper processed by simple
shear, Acta Materialia 53, (2005) 801-810.
Deformation conditions in conventional shearing dependent on
geometrical parameters of tooling
Y. A. Khan1, S.M. Ganesan2, H. Valberg3 , P.T. Moe4 , A. W. Hansen5

1,2,3,4,5
Department of Engineering Design and Materials, Norwegian University of Science and Technology,
Richard Birkelands vei 2B, 7491 Trondheim, Norway
URL: www.immtek.ntnu.no

E-mail: yawar.abbas@ntnu.no1; ganesan.s.m@ntnu.no2; henry.valberg@ntnu.no3;

Per.T.Moe@ntnu.no4; Arnfinn.Willa.Hansen@ntnu.no5

ABSTRACT: Shearing conditions in sheet metal blanking depend strongly on the geometry of the shearing
dies applied. A study is reported here to reveal how tooling geometry affects the conditions in the shearing
process. A FEM-analysis was performed to know about different geometrical conditions of the tooling which
influence the conditions in the shearing process. An experiment has also been conducted to validate the results
from the FEM-analysis. In the test, a visco-plastic technique was applied to map the real deformations.
Further, the mechanics of the shearing process is discussed elaborately on the basis of the above investigation.
From the FEM-analysis results it is concluded on; what was the shearing conditions in shear zones present in
the experiment.

Keywords: Shearing, Deformations, Shearing conditions, Strain rate, Strain

helped to evaluate the influence of, especially the

1. INTRODUCTION
The shearing process has a vital position in both
heavy as well as light metal processing industries.
Clearance and shape of shearing edge are two main
parameters of major concern in the shearing
operation. It is a fact that the quality of the blank
surface is affected by the magnitude of the clearance
between punch and die. Literature claims that large
clearances cause extensive plastic deformations
adjacent the shear region and punch corner.
Generally clearance values lie in between 4-12 % of
the blank thickness. With increased ductility of the
materials, the clearance usually decreases [1].
Finite element method is an excellent choice for
analysing the effect of various process parameters on
deformation in metal forming. It eliminates time
consuming experiments to optimise the process
parameters. In this work, Deform 2D® was used to
analyse the deformation conditions in terms of the
nature of localised shearing in the shear bands for
different clearances between punch and die in a
particular shearing test.
Thickness of sheets and punch diameter can have
very wide variations. Clearance will affect precision
of the sheared shape and the appearance of the
different shear zones [2]. The present study has
clearance on the deformation conditions in the shear
bands, i.e. strain and strain rates in the layer where
the metal finally gets separated in shearing.
From the literature, it is known that smaller
clearances and higher shearing rates improve the
quality of sheared parts. In our work special efforts
are put into investigating the nature of the shear
bands for different magnitude of clearances.
Simulation results mimic very well the shear band
formation in the experiment, by giving insight into
the mechanics of the formation of the shear bands.
2. SHEAR BAND FORMATION
Shear bands are narrow bands where localised
deformations take place at macroscopic level
irrespective of the nature of the micro-level
crystallographic slip bands [3]. Thermally activated
shear bands, called adiabatic shear bands can also
form depending on the rate of shearing. At high
strain rates, heat generated due to severe plastic
deformation and boundary friction, gets less time to
flow into neighbouring regions, and could promote
the formation of adiabatic shear bands [4, 5]. It is
often claimed that certain layers within the shear
bands experience dynamic recrystallization. This is
likely since temperature and strain rates can go very
1
high in these layers. The highly sheared material in
this localised region is commonly referred to as
shear deformed bands. It is also mentioned that there
could be thermo-mechanical instability, associated
with abrupt heating in such localised regions, due to
plastic deformation at high strain rate.
Investigation on the shearing of metallic parts has
given considerable attention to the shapes of the
product and the sheared off edge face. Previous
work has dealt with various materials and tool
conditions. In our study, precise experimental
analysis of flow of material has been carried out and
comparison is made with computational studies.
In experiments the periphery of the specimen were
held together with the bottom die by use of a simple
blank holder system, in corresponding manner as in
fine blanking. By this way the degree of tilting of the
sheet outside the blanking area was minimised. This
also contributes to create high hydrostatic pressure,
which puts constraints on the degree of radial flow
of material. Altogether, this should delay initiation
of fracture and promote large shear deformations
[3].
Punch
5.76 mm
2.62 mm
Die
Fig.1. Schematic diagram showing side view of specimen with
varying thickness and experimental setup for shear test
3. EXPERIMENTAL PROCEDURE
The specimen used in the study was a hot extruded
Al-strip with grid stripes inserted into the material
during the extrusion process. It was intentionally
ground to make an oblique specimen, so that there
was variation in specimen thickness around the
circumference of it, see Fig.1. Approximate
maximum and minimum value of the thickness were
6.6 and 2.0 mm on both extreme sides of the
specimen. Correspondingly in the region where the
punch shears the material these dimensions were as
shown in Fig.1. The material of the specimen and
the experimental procedures used were the same as
those reported in another of our papers [6] at this
conference. The approach was to create a partial cut,
i.e. the specimen to be sheared was positioned so
that the punch would reach its lowest position before
the specimen had been sheared fully across its whole
thickness, see Fig.3.
4. SIMULATION
Isothermal FEM-models of shearing corresponding
to the experiments were built using the DEFORM
2D® program to create axisymmetric models. The
purpose was to create models which allow us to
study the in-situ conditions inside shear bands
during shearing, i.e. the plastic deformations taking
place in these shear bands. The Deform-code uses
the Von Mises yield criterion so analysis assumes
this criterion to be valid. All tools were modeled
rigid and the workpiece as rigid-plastic with flow
stress defined by the well-known power-law,
σ = Kε n (1)
While working with the option of movement of the
top die in Deform, velocity was used being a
function of the stroke of the shearing punch. This
was required since the experiment was conducted in
a mechanical press (described in [6]) where punch
velocities varies in a fixed manner throughout the
stroke. Measured values of the characteristic
movement of the press have been obtained; they are
as shown in fig.2. This velocity profile was used as
input data in simulation to reproduce with good
accuracy the conditions in the experiment.
Friction between workpiece and die was modelled as
Coulomb-friction. The coefficient of friction was set
equal 0.2 which is expected to be close experimental
conditions, since the experiments were lubricated.
The shearing edges were modelled slightly rounded.
5. EXPERIMENTAL AND SIMULATION
RESULTS
Fig.3 reveals the internal stripe pattern in a diametric
cross-section cut through the specimen after end of
the shear test.
70
60
Thickest cut
50
Punch velocity (mm/s)
40
30
20
10
0
0 0.5 1 1.5 2 2.5 3 3.5
Stroke (mm)
Fig.2 Velocity vs. stroke used as input data for simulations of
shearing of the thickest portion of the specimen
2
Fig.3 Deformed internal stripe pattern of partially cut specimen
Fig.4 Magnified section of shear zone with grid in thick section
of specimen
The flow behaviour of the material in the shearing
experiments can easily be interpreted by looking at
the deformations present in the stripe pattern, as will
be discussed below.
The stripe pattern in the thickest part of the
specimen which was subjected to the largest total
degree of shearing has been shown in magnified
view in Fig.4. From this figure it is understood that
shearing at this location along the circumference of
the specimen occurred with negative clearance.
The internal stripe pattern in Fig.4 clearly depicts the
extension of the shear band inside the specimen, i.e.
the shear zone formed at this location in the
experiment.
A corresponding grid pattern was then made in the
same region inside the specimen in the FEM-model.
The appearance of the pattern at the same instant,
that is when the velocity of the punch has dropped
down to zero, and the punch is about to be pulled in
reverse by the press, is shown in Fig.5. As this figure
shows the simulated net deformations are very close
those in the experiment with one exception, i.e. in
the experiment the blank would move a larger
distance (denoted by a in the figure) towards the left
hand side. It is not obvious why this difference
occur, but one reason might be that we have tried to
FEM-model what is in reality a complex 3-D type of
experiment by a simple 2D-model.
Fig.5 Shear zone as predicted in the FEM-model
In spite of this difference in experimental and FEM-
predicted deformations in the shear zone, we can
conclude that the FEM-model is able to reproduce
the main features of deformations in the shear zone
in the experiment.
6. SHEAR ZONE APPEARANCE DEPENDENT
ON CLEARANCES
As discussed above, our analysis showed that our
simulation model was able to describe the plastic
deformations which occurred in the thickest part of
the specimen with acceptable accuracy. Hence we
decided to use the FEM-code to map what effects
are predicted in simulation if the clearance is
changed from the negative value in our experiment,
to zero clearance and finally to a positive clearance
of same magnitude as used in the experiment.
This was done by making two new simulation
models, which were identical to the first simulation
model used to mimic the experiment, with the
exception that punch diameter now was decreased in
relation to that of the first model, to obtain different
clearances. Resulting deformed grid patterns in these
two new simulations are shown Fig.6.
a b
Fig.6 FEM-predicted shear zones in case of (a) positive and (b)
zero clearances in the cut
3
 zero clearance, and the localised highest shear
deformations of the band therefore become less.
Very high strain rates are also encountered in the
shear layer investigated by us. During shearing, the
simulation model predicts effective strain rate in the
range of 20-30s-1 in case of zero clearance.

REFERENCES

1. Miguel Vaz Jr. and José Divo Bressan. ''A

a) b)
Fig.7 FEM-predicted shear zones in case of a) positive and b)
zero clearance in the cut
The effective strain values in the shear zones as
predicted by DEFORM for the two different
clearances values simulated by us is shown in Fig.7.
As this figure depicts the effective strain in the most
intense shear layer is much higher in case of zero
clearance compared to what it is for the larger
clearance.
7. DISCUSSION
It is not straightforward to know upto what extent
FEM-simulations are able to predict the conditions
in shear zones occurring in metal forming. In our
investigation, however, we have used an internal
stripe pattern inside the metal, to check how well
overall deformations in a shear band are reproduced
in a FEM-simulation model. Since we got good
agreement between simulated and experimental
deformations in the shear band investigated area in
our study, we feel confident that our simulation
model is describing the conditions in this strongly
deforming layer rather accurately.
computational approach to blanking processes
Journal of Materials Processing
Technology, Volumes 125-126, 9 September
2002, Pages 206-212
2. Zafer Tekiner, Muammer Nalbant and Hakan Gürün.
''An experimental study for the effect of different
clearances on burr, smooth-sheared and blanking force
on aluminium sheet metal''
Materials & Design, Volume 27, Issue 10, 2006, Pages
1134-1138
3. Manabu Gotoh, Minoru Yamashita.''A study of high-
rate shearing of commercially pure aluminium sheet ''
Journal of Materials Processing Technology, Volume
110, Issue 3, 2 April 2001, Pages 253-264.
4. Z. H. Chen, L. C. Chan, T. C. Lee and C. Y. Tang. ''An
investigation on the formation and propagation of shear
band in fine-blanking process'' Journal of Materials
Processing Technology, Volume 138, Issues 1-3, 20
July 2003, Pages 610-614
5. X. Teng, T. Wierzbicki And H. Couque ''On the
transition from adiabatic shear banding to fracture''
Mechanics of Materials, Volume 39, Issue 2, February
2007, Pages 107-125.
6. S.M.Ganesan, Y.A. Khan, H.Valberg, P. T. Moe, A.
Willa Hansen ''Cutting applied as test method for
acquisition of flow stress data at high strains and strain
rates'' ESAFORM 2008.

8. CONCLUSION

Since we applied an internal grid pattern to check

the deforming behaviour we can conclude that the
deformations in the shear band formed in our
shearing experiment are well reproduced in our
simulation model.
From the simulations we find that effective strain
values larger than 1.5 will have been created in the
most intense shear layer of the specimen when
approximately 2/3rd of its cross-section has been
sheared off. This is if shearing is done with zero
clearance of tooling.
In case of a positive or negative clearance
deformations spread into a broader layer than for

4
Improving numerical simulation of metal forming processes using
adaptive remeshing technique
C. Labergere1, A. Rassineux2, K. Saanouni1
1
University of technology of Troyes, ICD/LASMIS FRE CNRS 2848, 12 rue Marie Curie BP2060 10010
Troyes
e-mail: carl.labergere@utt.fr; khemais.saanouni@utt.fr
2
University of technology of Compiegne, Laboratoire Roberval, UMR CNRS 6253, centre de Recherche de
Royallieu, BP 20529, 60205 Compiegne
e-mail: alain.rassineux@utc.com

ABSTRACT
Numerical simulation by finite element of forming processes requires ductile damage. In this context,
problems of meshing and remeshing often impede the convergence of the processes and frequent remeshing
must be performed in order to avoid severe mesh distortions while adapting the mesh size to the solution
(plasticity, damage, temperature). Solutions to this problems are proposed in this paper. Geometrically and
physically error estimators have also been developed in order to locate areas where a remeshing is needed.
This adaptive scheme has been coupled to ABAQUS/Explicit solver

Key words: Adaptative remeshnig, error estimation, damage, FEA simulation, blanking

discretization by refining or coarsening the

1 INTRODUCTION
Numerical simulation by the FEM of various metal
cutting processes such as blanking or machining by
chip formation requires advanced modelling together
with adaptive analysis. Computational methods for
the virtual prediction of manufacturing processes
while taking into account strong coupling between
various thermo-mechanical phenomena including
ductile damage have been presented including some
works ([1], [2]). To avoid large mesh distortions due
to large plastic deformation, the mesh must be
adapted. Mesh density must be controlled by the
gradient of thermo mechanical field (plastic strain,
damage, temperature for instance) in order to
increase computation efficiency together with the
quality of the physical solution. In our approach,
macroscopic cracks are described by the fully
damaged elements deletion and a mesh adaptation is
also required. This mesh adaptation can be based on
some adhoc error indicators defined either by the
physical fields (solution) or the geometry of the
surfaces into contact.
In this work, a fully adaptive 2D numerical
methodology is proposed in order to simulate
various metal forming processes using advanced
constitutive equations together with an adaptive
meshing procedure. The numerical aspects related to
the resolution strategy of the initial and boundary
value problem including the adaptive time and space
discretization steps. This adaptive procedure is
validated by working out various sheet and bulk
metal forming. Some perspectives related to the
generalization of this adaptive procedure to the 3D
case are given.
2 MODELIZATION OF DUCTILE DAMAGE
One of the goal of this work is to propose FEM
simulation using advanced constitutive equations
accounting for non linear isotropic and kinematic
hardenings strongly coupled with ductile isotropic
damage. The coupling between the ductile damage
and the elastoplastic constitutive equations is
formulated in the framework of the thermodynamics
of irreversible processes together with the
Continuum Damage Mechanics (CDM) theory.
Further details to the behaviour model can be found
in the following reference (see [1], [2], [7]).
3 COMPUTATION OF GRADIENT-HESSIAN
FIELDS AND DIFFUSE DATA TRANSFER
As the domain geometry is time dependent,
emphasis must be given to the transfer of the
different fields during the process. In order to
recover the different field, we use a 2nd order
diffuse interpolation model [6]. The same model is
used to compute gradient and hessian needed to
compute error estimators when necessary.
For a 2D problem, all fields can be seen as a Monge
patch
z = f ( x, y ) =< 1, x, y, x 2 , xy, y 2 > α = PT α (1)
in which x and y denotes the coordinates of the
nodes of the structure. A surface can be thereafter
built by diffuse approximation from the data of the
3D mesh. The third coordinate z is given by the
value of each field. First and second order
derivatives can be computed from the set of point on
the basis of diffuse approximation. In this context,
moving least squares proved to provide a very good
accuracy of the derivatives while providing a
continuous model.
The Diffuse Interpolation is given by the
minimization of a criterion based on the
interpolation of the set of neighboring nodes and can
be written as
i=n
( )
2
J x (α ) = ∑ w( xi , x) × pT ( xi − x )α − zi
i =1
(2)
The minimization of quadratic criterion Jx(α)
leads to a 6×6 system and to the determination of the
surface coefficients. The contribution of each nodal
value to the approximation is influenced by a
weighting function w(xi,x) such that w(xi,.)>0 inside
the domain of influence of the node i and w(xi,.)=0
otherwise, providing a local character to the
approximation.
4 ERROR ESTIMATION
When dealing with non-linear behavior, both
material and geometrical one, it is essential to
capture the phenomenon with accuracy. We have
chosen an error estimator based on both spatial and
time value and hessian of the plastic strain.
Some authors [3], [4] have proposed error estimators
based on the hessian of the fields (plastic strain) and
on local deviation of the solution surface using
tangent plane. We experienced that the estimation of
the derivatives had to be weighted by the value itself
especially when damage occurs. Otherwise, the
method proved to overestimate the number of
elements.
Basically, the structure can be decomposed into 3
types of areas depending on the linear, non-linear
and damaged status.
• If the behavior is linear (geometric or
material), element size can be maximized.
• In order to capture non linearities such as
plasticity, it is essential to control the
derivatives of the plastic strain. We propose
an error estimator based on the Hessian of
the plastic strain. In order to compute the
Hessian matrix, a second order moving least
square model (diffuse approximation) is used
as explained in 3. 2nd order derivatives can be
obtained very easily from the polynomial
expression.
• We have also tested a physical error
predictor based on analytical functions which
depend both on the cumulated plastic stain
“p” and the damage “D”
Mesh size hp calculated by the cumulated plastic
strain (p):
( )( )
p ≥ p max ⇒ h p = h pmin − h pmax 1 − exp ( −ω ( p − p max ) ) + h pmax (3)
h pmax − h max
p < p max ⇒ h p = p − h max (4)
p max
When p<pmax we consider that the strain
distribution is homogeneous, h pmax and h pmin are mesh
size parameters. hmax is the maximum size of the
mesh.
Mesh size hd based on damage variable D:
D min ≤ D ≤ D max ⇒ h d = h endo
min − h
p
( ) DD + hp (5)
max
D max < D ⇒ h d = h endo (6)
min
where Dmin and Dmax are damage bound, h endo min
denotes the smallest element size. Generally
h max < h pmax < h pmin < h min
Local mesh size is estimated at each gauss point of a
element by the following expression:
&
h = h d − (h d − h p ) exp( − D) (7)
where D & is the damage velocity.
Before the first load step, the initial mesh geometry
can be optimized. Previous work has been done on
this topic. Some authors propose to adapt the mesh
with respect to the curvature of the contour. We
have chosen to limit the adaptation at the first step to
contact detection. The effect of curvature is detected
thereafter by the mechanical problem itself.
We noticed that whenever elements are fully
damaged, element size inside the whole structure
can be driven by a single error estimate based on the
value of the damage field. In the model we use,
cracks are represented by the removal of quasi-
totally damaged elements. Therefore, the smallest
size must be imposed to the damaged elements. The
size of the elements close to the crack is thereafter
small enough to capture with accuracy the effect of property is that the curve is contained in the control
plasticity and damage. polygon of its points (the convex hull).
Once the densities of the optimal mesh have been The evaluation of a point on a curve is given
computed, a new mesh which respects the prescribed thereafter given by de Casteljau's algorithm
density is generated. The aim of such a procedure is
to improve the accuracy of the numerical solution
while reducing the number of elements.

5 THE MESHING TECHNIQUE

In a 2D context, the difficulties of remeshing with (a) (b)

respect to a size map have been solved [6] and no
emphasis is given to this subject in this paper. The
main difficulty is to find a good compromise
between element gradation and the evolution of the
physical phenomenon. In the context of 2D
adaptation, a whole remeshing of the structure can
be performed at each step since its cost is much less
than the cost of the analysis itself.
We use here an advancing front mesh generator
Fig. 1. (a) Nodal generation by quadtree (detail) (b) Smoothing
along the crack and mesh coarsening
6 ADAPTATIVE MESHING PROCEDURE
The different steps of the FEM procedure using
adaptive remeshing is illustrated in figure 2:
Initialisation
([σ], [ε],Tc=0s)

based on a Delaunay criterion. Nodes are created on Tc=Tc+∆T

Subdivision of time in load

a prior approach using a quadtree structure based on sequences ∆Tc=TStep/nb_sequence
Interpolation and

the decomposition of an initial cell triangle no

transfert of mechanical
fields
Tc<Tstep
whenever triangles elements are used as shown in End of
computatio
yes
Generation of the new
mesh
figure 1a. When quads are requested, the initial cell Computation
Update correction

is a square. In a final step, local remeshing Abaqus/Explicit+Vumat

Adaptation of
sequences ∆Tc
load
Error estimation

topological operators are applied. Theses procedures Recovery of mechanicals fields and update yes

are known as edge splitting edge collapsing, edge

of the solution
-Displacement of part’s nodes
-Displacement of the tools
-Physicals values of Gauss points

swapping vertex removing and nodal shifting. The no Elimination of

damaged elements
Nbre of damaged
only geometrical data required by any automatic Damaged no
element<Kill_ma

mesh generator is limited to the discretized frontiers elements

yes Multiplication of the load sequence

(contours here). The prescribed element size is given ∆Tc= α∆Tc

by information computed at the integration point of

the elements. Therefore, all size information can be
represented by a punctual field.
In this spirit, whenever contact between bodies
(segment intersection) is detected, the element size
can be easily translated into a number of size
information points. The first step of the remeshing
process consist of an adaptation of the contour itself
with respect to the prescribed density.
During the remeshing process, some vertices of the
geometry must be kept and therefore these singular
points on the contours must be identified. These
points denoted as sharp corners can be either given
by the user or automatically detected by an angle
criterion. These point will be kept throughout the
whole process. The frontiers lines along the crack
are represented by Bezier curves. The choice of
Bezier curves can be justified as follows. In our
approach, the crack is represented by eliminated
elements. Therefore, the crack has a “sawtooth”
shape and in order to coarsen the mesh along the
crack, a smoothed representation of the crack has to
be used as shown in figure 1b. The main interesting
Fig. 2. flowchart of the adaptation meshing procedure
7 APPLICATION TO A TENSILE TEST
We simulate a tensile test in which the gage length
geometry of the specimen are 30x18x1mm. The
model is meshed with a plane strain triangular
element CPE3. The material parameter is copper
with the following properties : E=110GPa, ν=0.35,
σy=195MPa, Q=700MPa, b=4.2, S=35MPa, s=1.5,
β=3, Y0=0
This example illustrates the efficiency of the
meshing adaptation procedure with respect to both
plastic strain and damage. We clearly notice on
figures a-d, the mesh refinement with respect to the
cumulated plastic strain. Figure 3c shows two shears
bands while in figure 3d only one shear band
develops a crack.
 d=2mm (a) d=14,9mm (b)

d=15,6mm (c) d=18mm (d)

Fig. 4. Distribution of the cumulated plastic strain for different
displacements d d=0.33mm (c) d=1 mm (d)
Fig. 6. Distribution of the cumulated plastic strain for different
Figure 5 shows a good correlation between punch displacement d
numerical and experimental load-displacement curve

9 CONCLUSION

An adaptive remeshing procedure and error

indicators for the simulation of damage have been
presented here. Most of the tools used in this
approach can be extended to 3D problems.

REFERENCES
Fig. 5. Comparison between experimental load-displacement
and numerical load-displacement 1. K. Saanouni, Y. Hammi, Numerical
simulation of damage in metal forming
process, in Continuous Damage and
8 THE BLANKING PROCESS Fracture, Editor A. Benallal, Elsevier, ISBN
2-84299-247-4, (2000), pp 353-363
The mesh adaptation procedure is used on a 2. N. Belamri, A. Cherouat, K. Saanouni, P.
blanking operation of a cylindrical sheet. This Autessere, Numerical simulation 3D of sheet
process schematized in figure 8. The material is metal slitting process,. 4th International
copper previously used in chapter 7. The sheet is Conference on Fracture and Damage
meshed with quadrangular bilinear elements CAX4R Mechanics , 12-14 July 2005, Mallorca,
(axisymetric configuration). The friction between Spain p137-143
the tools and the sheet is modelized by the classic 3. H. Borouchaki, P. Laug, A. Cherouat and K.
Coulomb model with a friction coefficient η=0,15. Saanouni, «Adaptive remeshing in large
Fig. 6. Tool and sheet geometry of the blanking operation
plastic strain with damage», IJNME, vol. 63,
The error estimate parameters is given in table 1 2005, p. 1-36.
hmax h pmin h pmax pmax ω h endo
min
Kill_max 4. R. Boussetta, T. Coupez, L. Fourment,
0.8 0.07 0.2 0.35 20 0.03 40 Adaptive remeshing based on a posteriori
Tab. 1. Physical error estimate parameters error estimation for forging simulation,
Comp. Meth. Appl. Mech. Engrg., Vol. 195,
Figures 6 (a-d) show various sequences of mesh (2006), pp:6626-6645
adaptation during the blanking process. Figure 6d 5. A. Rassineux, «An automatic mesh generator
illustrates the fracture topography with different for planar domains», StruCome (1991), p.
zones (convex zone, sheared zone, fracture zone and 519-531.
the bur). 6. A. Rassineux, P. Breitkopf, P. Villon,
«Simultaneous surface and tetrahedron mesh
adaptation using meshfree techniques»,
IJNME, vol. 57, (2003), p. 371-389.
7. K. Saanouni, J-.L. Chaboche, Application to
Metal Forming, Computational Damage
Mechanics, Numerical and Computational
methods. (Editors: R. de Borst, H. A. Mang),
d=0.21mm (a) d=0.29mm (b) in Comprehencive Structural Integrity,
Edited by I. Milne, R.O. Ritchie and B.
Karihaloo, ISBN: 0-08-043749-4, Vol 3,
n 7, 2003
Interaction between machining and new fixturing principles for
aerospace structures
J. Leopold 1*, A. Poppitz 1, M. Klärner 2, A.-K.Schmidt1,J. Berger 1
1
Fraunhofer Institute IWU – Reichenhainer Str. 88; 09126 Chemnitz, Germany
URL: http://www.fraunmhofer.iwu.de e-mail: juergen.leopold@iwu.fraunhofer.de;
2
Professorship Plastics Processing Engineering; Chemnitz University of Technology - Reichenhainer Str.
70; 09126 Chemnitz, Germany
URL: http://www.tu-chemnitz.de e-mail:matthias.klärner@mb.tu-chemnitz.de;

ABSTRACT: Automated assembly is generally confined to mass production environments such as the
manufacture of cars and white goods. Even in this environment high-level automated assembly is restricted to
the OEMs where production volumes are high and flexibility and the ability to quickly reconfigure systems
are not major drivers.
In the aerospace industry the problem is further complicated by the move to thin walled monolithic parts and
the increasing use of composite structures. Monolithic structures have been introduced to reduce the costs of
assembling large numbers of components. Although the benefit of using monolithic parts is a large reduction
in overall manufacturing costs the downside is a more difficult component to handle and assemble. In
addition, there are thin walled components with sometimes-internal stresses and in our case made of nickel
based alloys, which only can be cut with difficulties. Machining the flexible structure and maintaining close
tolerance is difficult, transferring to assemble is difficult as well.
In this paper a theoretical analysis regarding static, dynamic and thermal behaviour of an aerospace part
(nozzle-guide-vans) starting from macro scale effects up to micro scale domain mechanics evaluation is going
to be presented. This analysis allows to evaluate different effects related to anisotropy, non-linearity and
damping properties, which can influence the final results of analysis.
The influence of different grain orientations of the grinding process to the mechanical properties in surfaces
has been included in the investigations.

Key words: Fixturing, Machining, Aerospace structures, Finite element

accurate and automatic fixing systems in industry.

1 INTRODUCTION
The reduction of overall-costs and lead-times is one
of the main goals in today’s manufacturing. At the
same time innovative materials including titanium
alloys, composites and metal-composite sandwiches
are used for assembling lighter and more and more
complex parts.
In manufacturing processes, fixturing is used for
accurate locating and tight holding of the workpiece.
The ability to establish and secure desired positions
is significantly assessing the effectiveness of a
fixture system. Further on, product differentiation
increases the need of highly efficient, flexible,
Especially in the aerospace and automotive sector
the number of low volume, high value and difficult-
to-handle products is increasing.
Specialized fixtures only are economic within a
mass production and cannot be reconfigured easily.
Otherwise the design and construction of modular
fixtures are very time-consuming and expensive,
with problems in repeatability and positioning
accuracy. Furthermore, quality and precision cannot
be controlled and influenced actively during the
manufacturing process and during
loading/unloading.
Thus there is a need of active, efficient, flexible and
precise fixturing systems, which are fields of
research in many sectors of industry nowadays.
As a basis of all development processes, the forces
acting upon the fixing components during the
manufacturing process have to be analyzed. In
industry, the knowledge about fixturing systems is
less advanced than about machine tools. An iterative
improvement of fixture prototype designs is still
common and causes high development costs.
Consequently an effective modeling and simulation
method for the prediction of the workpiece-fixture
behaviour during machining is required. Therefore,
the finite element method (FEM) is an adequate tool,
which is used for simulations [LCK+07].
In this context, this paper pays particular attention
on the interaction between machining and clamping
forces on the work of piece stability and deformation
based on a new fixturing principle for aerospace
structures, investigated on a Nozzle Guide Vane
(figure 1).
500
Force-X (N), Force-Y (N)

400

300

200
Force-X (N)
Force-Y (N)
100
Residual Stress 2
Leopold / Poppitz

0
0 5E-05 0.0001 0.00015 0.0002 0.00025 0.0003
Time (s)

Figure 2: Temperature distribution for the second cut

and cutting forces
Figure 1: Nozzle Guide Vane

3 MODELLING AND SIMULATION OF

WORKPIECE DEFORMATION
2 GRINDING SIMULATIONS
First, a complex model was developed, which based
The machining of the contours of the Nozzle Guide
on the CAD one (figure 3).
Vanes is done by 5 axis grinding machines.
However, to simplify the FE-simulation 2D-models
were used. As simulation software AdvantEdgeTM
was available. The grain machining that is typical
for grinding was replaced by single grain machining
with a strongly negative rake angle, like shown in
figure 2. As workpiece material Rene95 was used.
Parallel to the simulation grinding tests were done
with different plate models. The dynamic force
components include among other things the input for
the dynamic calculations of the workpiece clamping.
Figure 3: Complex model
4 CONCLUSIONS
It is known that fixturing and robotic grasping are
important manipulations which contribute greatly to
the production quality, cycle time, cost in
manufacturing, assembly and other industrial fields.
The goals of both are to immobilize kinematically an
object by means of a suitable set of contacts.
However, from the viewpoint of forces there are
essential differences between fixturing and robotic
grasping. The ultimate difference lies in the fact that
all of the fixture locators are passive elements and
only clamps can be considered to be active whereas
all of the robot fingers are active end-effectors. The
robotic fixturing focuses on the localization
accuracy of workpiece besides fixturing closure and
stability, whereas the robotic grasping emphasizes
the grasping closure and stability.
In this research, the authors employed the available
methods to address a closed loop system for the
space industry.
ACKNOWLEDGEMENTS
This research work is supported in parts by grants from the
European Union within the Integrated Project AFFIX (IP
NMP2-CT2006-026670-AFFIX), which is gratefully
acknowledged
clamping forces, the part deformation and new
simulation for fixturing principles of aeronautic
structures compared with principles of a smart
grinding fixture controller. The FEM allows to
analyse workpiece deformation, influence of
location errors and fixture deformation. In addition
to previously published papers, the acting grinding
forces (simulation and experiments) are included in
the simulation.
The reported work is part of an ongoing project
focusing on high precision applications in the aero-
REFERENCES
1. J. Leopold, D. Clauß, M. Klärner, M. Baldoli, A. Merlo,
M. Gimenez, B. Larranga, Investigations to New
Fixturing Principles for Aerospace Structures,
Proceedings of the APT’07: International Conference on
Applied Production Technology, pp. 173-189, 09/2007
Form Error Prediction of Gearcases’ Face Milling

J.V. Le Lan1 , A. Larue2 , P. Lorong2 , G. Coffignal2

1
Renault - API CTR B02 1 60; 67 rue des bons raisins; 92508 Rueil Malmaison; France
URL: www.renault.com e-mail: jean-vincent.le-lan@renault.com
2
ENSAM ParisTech - LMSP - 151 Boulevard de l’Hpital; 75013 Paris; France
URL: www.paris.ensam.fr/lmsp e-mail: arnaud.larue@paris.ensam.fr
philippe.lorong@paris.ensam.fr
gerard.coffignal@paris.ensam.fr

ABSTRACT: Reducing process development time and costs of new parts implies an increasing need of re-
liable process simulation. In order to guarantee efficiency to machining simulations, Renault has chosen four
criteria that must be respected by such numerical methods: Accuracy, Computation time, Robustness and Easy
use. A new numerical method adapted for gearcase milling is presented in this paper. This method is based on
the modal behaviour of the workpiece and the tool to provide a form error prediction. This paper uses an illus-
trative gearcase milling operation to present the method. Predicted results and production reality are showing a
good agreement.

KEYWORDS: Milling, Vibrations, FEM

1 INTRODUCTION the system.

Machining simulation has started in the early 80’s
at Renault. First computations were focusing on the
effect of clamping forces on form error of machined
parts. Then, cutting forces have been tacken into ac-
count after collaborative work with academics [6].
The goal is to compute the form error of the whole
machined surface. A macroscopic point of view is
thus adopted to describe the cutting operation.
A static method has been developped in collaboration
with ULg then improved as presented in Masset’s the-
sis [4]. Unfortunatelly, this method is no efficient any
more when the workpiece vibrates. It is the case of
gearcases face milling.
Macroscopic methods for milling that allow both
workpiece and tool to vibrate are already available.
For example, the work done by Coffignal et al. [3]
may be exploited. Such methods are using a geomet-
rical model and a mechanical one. The geometrical
model is used to compute the quantity of removed
material. The mechanical model is used to compute
deflections of both the tool and the workpiece. The
geometrical model is updated using the mechanical
results. This implies the use of a loop in order to
search for the dynamical equilibrium configuration of
This procedure is time-consuming and, in the case of
the simplified methods used at Renault, it is assumed
that the cutting force is not affected by the relative
displacement of the tool and workpiece neither by the
influence of the vibration on the history of the mate-
rial removal. This implies that no regenerative effects
may be tacken into account, then chatter prediction
has to be treated separatelly. Methods for chatter pre-
diction are proposed by authors [2].
2 THE OPERATION
The studied operation is the face milling of an alu-
minium gearcase. The clamping of the part is done
using nippers. This clamping design induce no part
deformation. Figure 1 shows the finite element model
of the part. Arrows are representing the clamping.
The machined surface is colored in blue.
The mill follows a complex trajectory around the ma-
chined surface. Its diameter is 63 mm, it wears 4
inserts and the rake angle is 90o . The rotation speed
is 20 000 RP M , the feed is 256 mm/min and the
depth of cut is 1 ± 0.5 mm.

1

Figure 1: Finite element model of the studied gearcase.
3 OPERATION MODEL
In this study, the mill is assumed to be rigid. In order
to take the dynamic behaviour of the workpiece into
account, the following equation is used :
M q̈(t) + C q̇(t) + K q(t) = Q(t) (1)
Where M , C and K are respectively the mass,
damping and stiffness matrices. Column q(t) con-
tains generalized displacements of the finite element
model. Column Q(t) contains generalized loads, in
this case, the cutting forces.
3.1 Integration of equation (1)
Matrices of the equation (1) are big so that the inte-
gration of the whole system could be very long. In
order to reduce the sizes of the involved matrices, the
system is expressed into the dual space of the eigen
modes. Calling Ψ the matrix that contains the whole
modal basis of the system, q(t) may be projected in
the modal space :
q(t) = Ψ y(t) (2)
In order to reduce size of involved matrices and
columns, a modal truncature is done. Hats denote
truncated modes and indice n denotes kept modes.
q(t) = Ψn y n (t) + Ψ̂ ŷ(t) (3)
| {z } | {z }
q n (t) q̂(t)
2
Two more diagonal matrices may be defined here :
ΨT M Ψ = I (4)
T
Ψ KΨ = Λ (5)
Matrix I is the identity. The following assumption
is done on the damping matrix :
ΨT C Ψ = D (6)
Where matrix D is diagonal. Combining definitions
(4), (5) and (6) with equations (1) and (3) and assum-
ing that truncated modes have only a static contribu-
tion to q(t), the following system can be written :
I n ÿ n (t) + Dn ẏ n (t) + Λn y n (t) = ΨTn Q(t)
(
−1 T (7)
q̂(t) = Ψ̂ Λ̂ Ψ̂ Q(t)
In this last expression, one may identify a resid-
ual compliance matrix Ŝ corresponding to the static
compliance of the system contained in the truncated
modes.
−1 T
Ŝ = Ψ̂ Λ̂ Ψ̂
(8)
= S − Ψn Λ−1
n
ΨTn
Where matrix S is the compliance matrix of the whole
mechanical system. q̂(t) becomes the static correc-
tion to q n (t). Finaly, computed q(t) has a good dy-
namic reliability in the frequency range of the selected
modes as well as a good static behaviour.
Integration of equations (7) is done using a Newmark
algorithm very quickly since involved matrices are
relativelly small sized.
3.2 Load modeling
The cutting force is assumed to be punctual (applied
on the nominal diameter of the mill). This is an hard
assumption for two reasons. First, the loading moves
along the time so that it is never applied on mesh
nodes. This imply over compliance of the finite el-
ement model that may have to be corrected as ex-
plained in [1] in case of low dynamic contribution.
The second reason is the fact that in reality the load-
ing is more likely a pressure applied on the intersec-
tion surface between the cutting insert and the part.
Many works focused on fine loading models such as
[5].
The cutting force is computed using cutting force
laws. These laws are indentified experimentally us-
ing the protocol described in figure 2.
Figure 2: Cutting force laws identification protocol.
It does not depend on previous computation steps
so that no regenerative effects are taken into acount.
3.3 Rear insert cutting
Since the mill follows a complex trajectory, no tilt an-
gle is applied to it. This increases risks to have rear
inserts to cut the part again (figure 3).
Figure 3: Back cutting.
When back cutting occurs, the corresponding
depth of cut is so low that there could be cutting re-
fusal and a high deterioration rate of the cutting insert.
Since the chip is very small, it is assumed that the rear
cutting applies no loads on the part. Its model simply
combines old results and instantaneous surface dis-
placement to know wether the cut exists or not. If
3
the rear inserts cut, the corresponding material is re-
moved.
4 APPLICATION
4.1 Model reduction
The finite element model used in this case for the
gear case contains more than 1 000 000 DOF. A static
reduction is used to reduce the size of the stiffness
matrix to the 5000 DOF of the machined surface.
This static model is enhanced with 27 modes up to
4kHz expressed on the same machined DOF set.
4.2 Test Results
In order to compare computed results to physical
parts, 10 gearcases have been machined on a flexi-
ble machine tool of the production line in Cléon fac-
tory, in France. Inserts height have been controlled
with a micrometer precision using optical devices.
Machined parts have then been measured on a 3D au-
tomated measuring device on 70 measurement points.
4.3 Computation Results
At each time step, the whole column q(t) is avail-
able. It is used to compute the displacement of the
force application point that is the only stored data.
At the end of the machining computation, all stored
displacements are used to compute the form error.
Observations of the displacements show that mode 8
is the most excited vibration mode. Its shape is pre-
sented in figure 4.
Figure 4: 8th Eigen mode of the clamped part (1373 Hz). Color
axis corresponds to displacements in the axis normal to the
figure
 Figure 5 shows the computed form error and fig-
ure 6 presents the form error measured on a represen-
tative part extracted from measurements.
This result shows good accuracy with measurements
on 80% of the surface. This result is showing good
agreement with the shape of mode 8 as well. Sectors
1 and 10 are showing poor results that may be due to
the fact that the machine tool is not modelized in this
computation. More precisely, since mode 8 implies
no dynamic compliance in this area, a static effect
may be missed due to the lack of machine tool model.
Figure 5: Computed form error result.
Figure 6: Measured form error result. This measure corresponds
to a representative part extracted from a 10 parts set.
4
5 CONCLUSION
The presented method shows good results in this case
of gearcase face milling. The model can be extended
to the machine in order to take its deflection into ac-
count. The whole form error computation is done us-
ing matlab. It needs 90 minutes on a laptop with a
pentium M 1.7 GHz, 1 Go RAM, which allows us to
expect dramatically reduced computation time in the
next evolutions.
This method is now used at Renault with strong bene-
fits to predict form errors for new industrials projects.
ACKNOWLEDGEMENT
Authors would like to thank Mickael Masson for his help and
organisation of the gearcases machining tests.
REFERENCES
[1] J.V. Le Lan. Etude de méthodes simplifiées pour la simu-
lation de l’usinage à l’échelle macroscopique. PhD thesis,
ENSAM ParisTech, CER de Paris, 2007. Available online
on pastel.paristech.org.
[2] J.V. Le Lan, A. Marty, and J.F. Debongnie. Providing sta-
bility maps for milling operations. International Journal of
Machine Tool and Manufacture, pages 1493–1496, 2006.
[3] Ph. Lorong, J. Yvonnet, G. Coffignal, and S. Cohen. Contri-
bution of computational mechanics in numerical simulation
of machining and blanking : State of the art. Archives of
Computational Methods in Engineering, 13, 2006.
[4] L. Masset. Analyse de gammes d’usinage par la méthode
des éléments finis. PhD thesis, Université de Liège, Faculté
des sciences appliquées, 2004.
[5] E. Ozturk, E. Ozlu, and E. Budak. Modeling dynamics and
stability of 5-axis milling processes. In 10th CIRP Inter-
national Workshop on Modeling of Machining Operations,
2007.
[6] Herbert Schultz and Klaus Bimschas. Optimization of preci-
sion machining by simulation of the cutting process. Annals
of the CIRP, 42/1/93, 1993.
High speed broaching of hard machining materials
V.F. Makarov, D. I. Tokarev, V.R. Tyktamishev
Perm State Technical University, – Komsomolski prospect, 29a, Perm, Russia, 614600
URL: www.pstu.ru e-mail: tms3@pstu.ru

ABSTRACT: Materials have been broaching on various speeds of cutting are presented. Comparative
research of influence the traditional broaching on low speeds of cutting (V=2 m/min) by the broaches from
high speed steel R18 and the high-speed broaching (V=26 m/min) by the hard-alloy broaches VK8 on the
change of intensity of wear of the broaches and formation of key parameters of quality of a superficial layer – a
roughness and fatigue durability. Introduction of the high-speed broaching promotes the receipt of higher and
stable parameters of quality of a superficial layer, increase of fatigue durability, reliability and a resource of
gas-turbine engine parts and the minimal values of the tool wear intensity.

Key words :Broaching, Optimization, Quality, Superficial Layer, fatigue durability.

1 INTRODUCTION
(2) to study the influence of speed of cutting on the
It is known, that the speed of cutting at the broaching change of intensity of wear of the broaches
usually makes 0,5-6 m/min. This speed is 10-15 times
less than the speed of cutting at turning and milling of (3) to show the influence of speed of cutting on the
the same steels and alloys. As tool materials for the change the parameters of quality of a superficial layer
broaches from high speed steels is usually used. and fatigue durability.
Broaching proceeds under high temperature
conditions and exceeding of this speed of broaching 2 DETERMINIATION OPTIMUM SPEEDS OF
leads to severe wear of tools. At turning and milling CUTTING
the cutters and mills from hard alloys are applied. Theoretically and practically it is proved, that at the
Long life of these tools is higher at 5-10 times than broaching there is an optimum condition of a zone of
tools from high speed steels. The advent of hard cutting at optimum temperature of cutting. At this
alloyed broaches gives rise to higher production rate temperature optimum conditions of formation of
of broaching hard machining materials. But for its required parameters of quality of the processed
realization theoretical and experimental surface and the minimal values of intensity of wear of
investigations of high speed broaching are needed. the tool are observed [6].
Such investigations (laboratorial and industrial) were
carried out at Perm Motor Plant. 30 marks of various
heat resistant steels and alloys in a range of speeds
of cutting 1, 5 – 60 m/ min were broached. For
investigations modernized broaching machines with
specially developed fast cutting and hard alloyed
broaches were used.
On the basis of obtained results a high speed
broaching method is developed and investigated and
optimization of its parameters and tool is carried out.
It is known, that witch increase the speed of cutting
the temperature and force of cutting change too [1-
5]. This changes leads to change the parameters of
quality of a superficial layer. It is therefore the main
Fig. 1: Influence of temperature of mechanical tests T on
objectives of this paper are: durability σв and plastic (δ,ψ) properties of the heat resisting
(1) to determine optimum speeds of cutting alloy EI787VD.
The analysis of results of research of change of
plasticity (δ,ψ) of the heat resisting alloy EI787VD
from temperature of mechanical tests T (see Figure 1)
and results of research of change of intensity of wear
of the broaches hiw and temperatures of cutting T
from speed of cutting V and feeds on tooth Sz (see
Figure 2) has shown coincidence of temperature of a
failure of plasticity Tfp to optimum temperature of
cutting To=700°C. It allows to explain extreme
character of dependence of wear of tooth’s from
speed of cutting hiw=f(V) at the broaching respective
alteration of plastic properties of materials at rise in
temperature of cutting.
Fig. 2: Influence of speed of cutting V on change of
temperature of cutting T (a) and intensity of wear of the
broaches h iw (b) at the broaching of the heat resisting alloy
EI787VD with various feeds on tooth Sz.
It is established, that for different values of rise on
tooth Sz=0,02; 0,06; 0,10 mm/tooth, minima of
intensity of wear of the broaches hiw is observed at
corresponding optimum speeds of cutting Vo=30; 22;
15 m/min to which there corresponds the same
optimum temperature of contact To=700°C (see
Figure 2). At the broaching of the alloy EI437BUVD
at the same rises on a tooth, to optimum speeds of
cutting 35; 28; 25 m/min correspond To=720°C. At
the broaching of the alloy EP109VD at the same rises
on a tooth, to optimum speeds of cutting 20; 15; 12
m/min correspond To=820°C.
From here it is possible to draw a conclusion, that at
the broaching the multi-section block of broaches, in
process of reduction of size of feed by tooth Sz, the
speed of the broaching is necessary to increasing up
to optimum sizes. In result productivity of the
broaching and long life of the broaches raises in some
times.
3 RESEARCH OF FORCES OF CUTTING
Change of force of cutting changes character of
contact pressure in a zone of cutting. It influences
character of friction, wear and destruction of tooth’s
of a broaches, on process of formation of a shaving,
on formation of key parameters of quality of a
superficial layer and fatigue durability of the
processed details.
Research of forces of cutting is carried out by the
example of alloy EI787VD. At increase in speed of
the broaching with 1 up to 60 m/min axial Pz and
radial Py forces of cutting are reduced up to the
certain value with the subsequent stabilization or
some increase at speeds of 40-60 m/min.
Fig. 3: Influence of speed of the broaching on change of
tangential and radial making forces of cutting at the broaching
of the heat resisting alloy EI787VD (width - 5 mm).
Comparison of schedules in figure 3 with schedules adverse speeds (Figure 4) is accompanied by
in figures 1, 2 shows, that the minimal value of forces substantial growth of height of microroughnesses.
of cutting corr and the minimal values of intensity of
wear of the tool are observed esponds to optimum
speed of cutting Vo. At optimum speed of cutting Vo,
temperature of contact To coincides with a
temperature zone of "a failure of plasticity" materials,
i.e. with the lowered durability and the increased
fragility. At increase or reduction of speed of cutting
respective alteration of temperature of cutting results
in increase of plastic properties of a material and,
finally, to increase of forces of cutting.
Thus it is established, that the minimal value of forces
of cutting Pz and Py corresponds to speed of cutting
which we name optimum Vo. This speed corresponds
also to the minimal value of intensity of wear of the
broaches It is interesting to note also, that values axial
Pz and radial Pу making forces of cutting appreciably
depend on physical and chemical properties of
metals, for example, from the maintenance of
strengthening γ′-phase. With increase in the
maintenance of a γ′-phase with 12,5 %
(EI437BUVD) up to 38 % (EP109VD) values Pz and
Py increase in 1,5 - 2 times. It should be taken into
account by development of designs of the broaches,
adaptations and at a choice of traction efforts of
machine tools.

4 RESEARCH OF QUALITY OF THE

PROCESSED SURFACE
Comparative research of influence the traditional
broaching on low speeds of cutting (V=2 m/min) by
the fast-cutting broaches R18 and the high-speed
broaching (V=26 m/min) by the hard-alloy broaches
VK8 on formation of key parameters of quality of a
superficial layer of details from heat resisting steels
and alloys, has yielded the following results.
4.1 A roughness
The size of a roughness is usually a determining
technological criterion of process of broaching. Wear
of a roughness in a range of adverse speeds is
connected to amplification of process of an
outgrowth-formation. Such law is especially
characteristic at the broaching of the heat resisting
steels EP517H, EI736H, EI961H, EP609H and
others. The broaching of these steels in a zone of
Fig. 4: Influence of speed of cutting V on change of size of
roughness Ra of broached surfaces from the steel EI961H and
the titanic alloys VT9, VT18U (a) and the heat resisting alloys
EP109VD, EP742ID, EI437BUVD, EI787VD (b).
Besides at the broaching of steel EI961H by the fast-
cutting broaches (R18) the smaller size of a
roughness is received, than at use of the hard-alloy
broaches (VK8). It speaks a various degree of
adhesive interaction of steel R18 and alloy VK8 with
heat resisting steels, various heat conductivity R18
and VK8 and, hence, in various temperature of
contact at the same modes of cutting. Besides at the
broaches from R18 after sharpening value of radiuses
of a rounding off of cutting edges ρ=8-15 µm, and at
the broaches from VK8 ρ=20-30 µm. And the more
the radius of a rounding off, the is more outgrowth
and degree of an outgrowth-formation with
corresponding growth of size of a roughness of a
stretched surface.
For the corrosion-proof heat resisting steels,
application of hard-alloy broaches VK8 is expedient
at speeds of cutting more than 24 m/min since at
smaller speeds because of an outgrowth-formation the
required roughness (for gas-turbine engine parts is
required Ra≤1,25 µm) (Figure 7). For all other groups
of the researched materials the application of the
hard-alloy broaches provides a required roughness in
all a range of researched speeds of cutting (Figure 7).
4.2 Fatigue durability
Application of the high-speed broaching promotes
increase of fatigue durability of gas-turbine engine
parts in comparison with the traditional broaching on
low speeds of cutting. At the broaching of the
grooves in rings from the steel EI962H and the titanic
alloy OT4 the fatigue durability raises on 8-10 %, and
at the broaching of the grooves in disks of
compressors from the titanic alloys VT3-1 and
VT18U the increase in fatigue durability makes a tail
more than 60 % in comparison with the broaching on
low serial speeds of cutting. At the broaching of the
heat resisting deformable and foundry alloys increase
of fatigue durability with increase in speed of cutting
occurs to a lesser degree.
5 CONCLUSIONS
Development and investigation of a high speed
broaching method has allowed to formulate a
generality for the process of broaching in the never
before experienced range of speeds.
The results obtained allow providing high quality
and fatiguing durability for work pieces made of
hard machining materials broaching at 25 – 30
m/min, long life of hard alloyed broaches exceeding
10 times that of high speed steels instruments. On
the basis of the results obtained there were
developed recommendations allowing to introduce
the process of high speed broaching into a batch
production of gas turbine engines for power and gas
–pumping installations and also for manufacturing
of new modern aviation engine PS – 90A for IL 96 –
300 and TU – 204 and etc.
Researches and investigations of high speed
broaching process applying to more difficult
surfaces are still proceeding.
REFERENCES
1. Broach for Rolls-Royce. Machine-Tool Review, 1975, 63,
№368, 149.
2. Broaching turbine discs. Iron Age Metalworking
International, 1972, 11, № 3, 27.
3. Forst high-speed broaching. Machinery and Production
Engineering, 1968, 113, № 2920, 931-932
4. High-speed broach has a 12 sec. Cycle. Metalworking
Production, 1973, 117, № 9, 77.
5. Hoffman Kurt. Le brochage rapide et sa signification.
Machine Modern, 1971, 65, № 753, 7-9.
6. Makarov V.F., Chigodaev N.E., Tokarev D.I
Optimization of Cutting Process at the High-Speed
Broaching of Gas-Turbine Engine Parts. Proceedings of
the 9th CIRP Internahional Works on Modeling of
Machining Operations, May 11-12, 2006, Bled, Slovenia,
327-334.
Numerical prediction of punch wear in the context of blanking process of
copper alloy thin sheet
H. Makich1, J. Chambert2, P. Picart2, G. Monteil1, X. Roizard1
1
LMS – ENSMM, 26 Chemin de l’Epitaphe 25000 Besançon, France
URL: www.lms.ens2m.fr e-mail:hamid.makich@ens2m.fr;guy.monteil@ens2m.fr; xavier.roizard@ens2m.fr

2
Institut FEMTO-ST – DMARC, 24 Chemin de l’Epitaphe 25000 Besançon, France
URL: www.femto-st.fr e-mail: jchamber@univ-fcomte.fr; philippe.picart@univ-fcomte.fr

ABSTRACT: Through the study presented here, we want to identify how the simulation of the blanking
process could reduce defects of production, improve punches life and assist in the progression and control of
the process. This was carried out with using software LS-DYNA. Based on the results obtained by simulation
of blanking operation in two dimensions, we can predict the evolution of the gradual wear of active parts of
punch according to the number of press strokes; by developing a calculation algorithm based on the Archard
law of wear.

Key words: Blanking, Numerical Simulation, Finite Element Method, Punch Wear.

the contact area. The abrasive wear occurs most

1 INTRODUCTION
The numerical simulation is a domain in full
alteration which gradually stands out as a separate
productive tool. The simulation of blanking process,
subject of this paper, is a typical example of this
surge in numerical calculation technology in the
industry in order to reduce costs and improve the
process. The implementation of a numerical
modeling of blanking process seems important to
control aspects of this operation. The objective of
this work is to develop a finite element approach for
predicting the evolution of the punch wear according
to the number of press strokes. It is about
implementing a numerical modeling for blanking
process of copper alloy thin sheet in 2D under
software LS-DYNA and correctly predicting the
evolution of the geometry of the punch.
2 MODELING OF WEAR IN BLANKING
The wear of the sharp elements in a blanking tool
determines the productivity of the companies and the
quality of the products. The wear corresponds to the
progressive loss of material of the body active
surface, as a result of the relative movement of
another body on this surface. For a cutting punch,
this means a change of the geometry of the sharp
part. The blanking tools often show abrasive wear in
generally in the tools when the surface of the sheet
contains hard particles, such as carbides or oxides.
When the tools wear out, there is a district of their
cutting edges leading to an increase of party edges
rays. The main consequences for blanking are:
Increased penetration to break, appearance of shape
defects and quotation variations, increased roll-over
depth, increased burr, degradation of the mechanical
characteristics of the sheet material and increased
effort to blanking
2.1 The Archard wear model
Possession of wear laws helps to predict the
behaviour and life of a system. Several equations are
based on the hypothesis that conventional material
property will be important in the process of wear.
An example of this type is presented by Archard [1,
2]. He established an experimental equation to
estimate the volume of used material in the case of a
sliding contact at a constant speed and very different
hardness of materials. He used a cylindrical wiper on
a disk. From this, he measured different dry-rubbing
materials. He studied rough wiper’s wear against a
rough plan, and adopted the following hypothesis:
the amplitude of roughness does not vary during the
process and the area of contact remains the same.
The Archard equation is: V=K×F×x
Were V = wear volume, K = coefficient of wear • The wear of tools is taken into account using the
(mm²/ N), F = the load applied to the contact and x = model of Archard.
sliding distance. The coefficient of wear (K) itself The mechanical properties of the sheet [5] are
varies with the other tribological parameters such as summarized in the following table.
temperature, environment, nature of friction body,
sliding velocity, normal charge… Its value is also Table1. Mechanical properties of sheet metal
Size Value
linked to the nature of the wear.
Young modulus (E) 99500 MPa
Poisson ratio (ν) 0.31
Yield strength (σ) 199 MPa
3 NUMERICAL SIMULATION OF BLANKING
The geometric parameters of the sheet and tools
3.1 Model for simulation used in the simulation are represented in this table.
The model proposed in this section concerns the Table2. Geometric characteristics of tools and sheet metal
axisymmetric blanking from sheet thickness Size Value
0.254mm. It is an explicit formulation of 2D process Sheet thickness 0.254mm
of blanking, done on the code of finite element Sheet width 15mm
Diameter edge of punch 0.05mm
calculation ANSYS / LS-DYNA [4]. The explicit
Diameter edge of die 0.05mm
algorithms are the most used thanks to non-linearity Clearance punch - matrix 20µm
encountered in the field of cutting, which might Angle end of the punch 0°
make diverge implicit algorithms [3]. The aim is to Punch diameter 3.7mm
simulate in a realistic way the operation of blanking.
The operation studied in this paper is performed 3.3 Mesh definition
with a cylindrical punch. Given that the cut pieces
are fairly simple geometries, the shapes are One of the most important steps for simulation by
symmetrical revolution around an axis, loading and the finite element method is the meshing phase.
boundary conditions are also of revolution around
this axis, it is possible to deal with the problem by (a)
adopting a hypothesis of axisymmetric (Fig. 1).

(b)

Fig. 1. Geometry of the simulation model

3.2 Getting data for the simulation

During the modeling and in order to be as close as

possible to the operating conditions, we take into
account the following aspects [1]:
• The punch and die are supposed rigid
• The interface elements are defined by the type of
sliding contact
• The material is described by a elastoplastic
behaviour law
• The process is considered as quasi-static Fig. 2. (a and b) Decomposition of the mesh into three zones.
• The inertial effects are neglected
The sensitivity of the numerical mesh is well known
in calculating. In large deformations, this sensitivity
is even more important because of the large loads
applied. In blanking, the material undergoes great
rate of strain located in a small strip. A correct
numerical description of mechanical phenomena
occurring in this area requires the choice of small
size elements. In order to set the mesh size, we can
separate the sheet into three distinct areas (fig. 2).
Zones 1 and 3 may be meshed grossly because the
deformations are limited to it. The size of the mesh
is of the order of 0.09mm. While the meshing of the
zone 2 must be clearly refined, it is of the order of
0.003mm. For the choice of the element type, an
initial meshing, consisted of elements quadrangles
axisymmetric in 4 knots, was used.
3.4 Boundary conditions
Regarding the boundary conditions, the die and the
blank-holder are blocked in rotation and translation.
For the punch, only the vertical translation is
allowed.
4 IMPLEMENTATION OF THE WEAR
CRITERION
4.1 Sliding distance
At the moment (t): the node M1 is considered belong
to the sheet metal and to the punch.
At the moment (t+dt): geometric point M1 becomes:
- M2, which belongs to the sheet metal.
- and the H point, which belongs to the punch.


On the other hand, ( U ) represents punch penetration
between the two increments of time (t) and (t+dt).
The hypothesis put forward in this context is to
consider the distance between the two points (H) and
(M2) as the sliding distance with a given length of
metal faced to a punch node. Thus, for all the nodes
in contact between the sheet metal and the punch,
we have the following data (fig.3):




 X1 



 X2 
  0 






OM1 =   OM2 =   U =   with M 1 H = U
Y1  Y2  Y 
And: (X1, Y1, X2, Y2 and Y) Known data.
 OH = OM 1 + M 1H = OM 1 + U





















Hence: 


  X 1 
 So : OH = 
  Y1 + Y 





  X 2 -X1 
Finally: HM 2 =  
 Y2 -(Y1 + Y) 
So the sliding distance (x) is given by:
x = HM 2 = ( X 2 -X1 ) + ( Y2 -(Y1 + Y) )





 2 2

The method that is used is the determination of wear

with the Archard law. First, this is realized with the 4.2 Normal effort
sliding distance (x) calculation of each node, this
calculation is realized between two increments of The determination of the normal effort in each node
time from the results of node displacement which in the contact between the punch and the sheet was
come from the simulation. The second step is the geometrically studied, as well as the sliding distance.
determination of the normal effort in each node in Thus, we proceeded in the following manner (fig.4):
the contact between the punch and sheet metal. The
calculation of node sliding between two increments M3
is done in the following way (fig.3.):
M2


n2
M1


n1

N

Fig. 4. Normal effort at the nodes

In figure 4 (M1M2) and (M2M3) represent elements





Fig. 3. Sliding between two increments between two successive nodes. ( n1 ) and ( n 2 ) are
two normal unit vectors of (M1M2) and (M2M3).

x  x 
M 1 M 2 =  1  ; M 2 M 3 =  2  Punch Punch
 y1   y2 
 x1   n1x   0 
So: M 1 M 2 . n1 = 0 ⇒  .  =   Sheet Sheet
 y1   n1 y   0  metal metal

 x1 . n1x + y1 . n1 y = 0  n1x 
 n =  
Hence:  ⇒ 1 n 
Fig. 5. Distribution of Von Mises equivalent stress.
 n1 = n1x + n1 y = 1
2 2
 1y  (a) 0.07mm penetration and (b) 0.217mm penetration
The calculation of nodes sliding between two
 x   n2 x   0 
And: M 2 M 3 . n 2 = 0 ⇒  2 .  =  

increments shows that the rate of displacement of
 y 2   n2 y   0  nodes in the shear zone is more important than other
nodes (Fig. 6).
 x2 . n2 x + y 2 . n2 y = 0

 n2 x 
Hence:  ⇒ 2 
n = 
 0.03
 n 2 = n 2 x + n 2 y = 1
2 2 n
 2y 
0.025






  N x = n1x + n 2x 

Sliding distance (mm)

0.02
Finally: N = n1 + n 2 N =  
 N y = n1y + n 2y  0.015

N  n1 


0.01

n Unit vector of N1 ; so: n =
 = 







N  n2  0.005

  Fx 
F is the force measured to each node as: F=  
 Fy 
 
So, the efforts at each point are calculated by using

the following equation: Fn =F.n
4.3 Remark
The next work aim an experimental study to identify
the coeficient (K), and integrated by following an
incremental form of the Archard law, to establish a
relationship giving the volume of punch wear at all
increments of penetration.
5 RESULTS AND CONCLUSIONS
Preliminary results show the mesh distortion in the
band shear and a loss of contact after a few depths of
penetration. Figure 5 shows the distribution of Von
Mises equivalent stress for two different
penetrations, we can see there is a loss of contact
between the punch and sheet metal, for a penetration
of 0.217mm, representing about 85% of thickness of
the sheet metal.
0
0 5 10 15 20 25
Nodes of sheet metal in contact zone
Fig. 6. Sliding distance of nodes in the contact
Currently we are working on improving the contact
to get closer to the behaviour during blanking. In
another, we plan experience to evaluate the
empirical parameter (K) for materials used.
REFERENCES
1. R. Hambli, Blanking tool wear modeling using the finite
element method, International Journal of Machine Tools
& Manufacture, 41 (2001) 1815-1829.
2. J. F. Archard, Contact and Rubbing of Flat Surfaces,
Journal of Applied Physics, Vol. 24, pp. 981-988, 1953.
3. A. Slimani, Simulation numérique du découpage de tôles
minces – Prédiction de l’usure du poinçon, Internal
Report, University of Franche-Comté, France, 2007.
4. B.N. Maker, X. Zho, Input parameters for metal forming
simulation using LS-DYNA. Livermore Software
Technology Corporation, 2000.
5. F. Gréban, Découpabilité du cuivre et des alliages
cuivreux, Ph.D. Thesis, University of Franche-Comté,
France, 2006
Identification of Wear Characteristics in Tool Wear Model of Cutting
Process
T. Matsumura1, T. Shirakashi2, E. Usui1
1
Department of Mechanical Engineering, Tokyo Denki University
– 2-2 Kanda Nishiki-cho, Chiyoda-ku, Tokyo, 101-8457, JAPAN
URL: www.skynet.m.dendai.ac.jp e-mail:tmatsumu@cck.dendai.ac.jp;
2
Department of Precision Machinery Engineering, Tokyo Denki University
– 2-2 Kanda Nishiki-cho, Chiyoda-ku, Tokyo, 101-8457, JAPAN
e-mail: shirakashi@cck.dendai.ac.jp

ABSTRACT: The tool wear progress is simulated with the cutting force and the cutting temperature. The
cutting force is predicted with piling up the orthogonal cuttings in the force model based on the minimum
cutting energy. The cutting temperature is predicted in the finite volume method with assuming the stress
distribution. The abrasive wear model is applied to the flank wear prediction. The wear characteristic
constants in the wear model are identified to minimize the model error between the prediction and the
measured wear rates. Because the initial flank wear generally occurs in the flank wear progress, the prediction
starts at the initial wear offset. The initial wear offsets are also identified to minimize the error between the
predicted and the measured flank wear lands. Some of flank wear progresses are measured to identify the
wear characteristics constants and the initial wear offsets in the turning operation. The tool wear prediction is
performed to verify the identification of the wear characteristics.

Key words: Turning, Cutting force, Cutting temperature, Tool wear, Abrasive wear

machine shops. It is the major issue that the wear

1 INTRODUCTION
Recent machine shops have required the simulation
of the cutting process to review and improve the
cutting operation. Tool life is one of the important
factors in estimation of the machining performance
as well as the machining accuracy and the surface
finish. Airplane and automobile industries have also
required the prediction of the tool life in cutting of
especially difficult-cut-materials such as Inconel and
Fiber reinforce plastics. Many researches have been
made on the tool wear model to optimize the cutting
conditions so far. Tool wear originally depends on
the stress and the temperature in the interface
between the cutting tool and the workpiece. Usui et
al. proposed a wear model considering their effects
on the wear rate [1]. The model characterizes the
wear rate with the sensitive parameters of the stress
and the temperature on the tool face. Therefore, the
measurements of the stress and the temperature have
been required to apply the model to the wear
prediction. However, the experiments with the
dedicated set up cannot be performed easily in the
characteristic constants should be analysed in the
wear progress from the practical point of view.
The paper presents an inverse analysis for the
identification of the wear characteristic constants.
The constants can be identified in the cutting tests
without the specialized equipments. An example of
the identification is shown to verify the presented
approach. The cutting processes are then simulated
in the wear model based on the identified constants.
2 PREDICTION OF FLANK WEAR PROGRESS
Because the prediction is described in detail in
Reference [2][3], this paper briefly summarizes the
simulation. The cutting force is predicted in the
model based on the minimum cutting energy. Figure
1 shows the force model in the three-dimensional
cutting process. The process is interpreted as a piling
up of the orthogonal cuttings in the plane containing
the cutting velocity and the chip flow velocity. The
shear angle φ, the friction angle β, the shear stress on
the shear plane τs, and the tool-chip contact length lc
 End cutting edge Workpiece
Orthogonal cutting plane
Feed component
Front cutting edge
Chip αe
αs φ
Workpiece ηc Radial τt Chip
Chip flow component
f direction ηc
τs αe
Principal
Tool
d αb component
Cutting velocity Side cutting edge Chip flow
τf=σf direction
Fig. 1. Three-dimensional cutting model in turning Flank wear
αb,αs, and ηc are back rake angle, side rake angle, and chip land
Main cutting edge
flow angle; and f and d are feed rate and depth of cut.
Fig. 2. Stress distribution on tool face
φ, shear angle; τt, friction stress on rake face; τs, shear
in the orthogonal cutting can be associated with the stress on shear plane; and τf and σf are friction stress and
cutting velocity V, the uncut chip thickness t1, and normal stress on flank wear land.
the rake angle α in the following equation:
⎧φ = f (V , t1 , α ) dW ⎛ λ⎞
⎪ β = g (V , t , α ) = C exp⎜ − ⎟ (2)
⎪ 1 σ t dL ⎝ θ⎠
⎨ (1)
⎪ τ = h (V , t1 ,α ) where dW/dL is the wear volume in unit area per the
⎪⎩l c = J (V , t1 , α ) sliding distance. σt and θ are the normal stress and
the temperature respectively. C and λ are the wear
The orthogonal cutting data are prepared in the characteristic constants given in the combination of
combination of workpiece material and cutting tool. the workpiece and the tool. Then the flank wear rate
With assuming the chip flow direction, the cutting (dVB/dt) can be given by the following equation:
energy, which is the sum of the shear energy on the
shear plane and the friction energy on the rake face, dV B ⎛ λ ⎞⎛ 1 ⎞
can be calculated with making the orthogonal cutting = Cσ f exp⎜ − ⎟・⎜
⎜ − tan α ⎟⎟V (3)
dt ⎜ θ ⎟ ⎝ tan γ ⎠
model. The chip flow direction, which is specified ⎝ f ⎠
by the angle ηc, is determined to minimize the where σf and θf are the normal stress and the
cutting energy. Then, the principal, the feed and the temperature on the flank wear land; and V, α and γ
radial components of the cutting force can be are the cutting speed, the rake angle, and the relief
predicted in the chip flow model determined. angle. The temperature distribution can be analysed
The shear stress τs on the shear plane and the friction with assuming the friction stress τf on the flank wear
stress τt on the rake face can be distributed as shown land. The wear rate is then calculated in Eq. (3).
in Figure 2 [4]. The distributions of the normal stress Because the wear rate is generally constant over the
σf and the friction stress τf are assumed on the flank flank wear land, σf and τf are modified so that the
wear land, where the friction stress is equal to the wear rate is the same over the flank wear land.
normal one [1]. Finally, the stress distribution, the wear rate, and the
The cutting temperature can be calculated in the temperature distribution can be determined. Then the
finite volume method [5]. All mechanical works, flank wear VB at the cutting time T can be predicted
which are the plastic work on the shear plane and the in the following equation [2]:
friction works on the rake face and the flank wear
T ⎛ dV ⎞
land, are converted into the heat generation based on VB (T ) = VB 0 + ∫ ⎜ B ⎟dt (4)
the stress distributions. 0
⎝ dt ⎠
The wear rate can be given by the following abrasive
wear model, which is derived from Rabinowiczs’ where VB0 is the initial wear offset, which is the
analysis [6]: width of flank wear land at the time T=0 in the
calculation.
3 IDENITFICATION OF WEAR Table 1. Cutting conditions in wear tests
CHARACTERISTIC Workpiece 0.45% Carbon steel
Tool material Carbide P20
Tool geometry (-5, -5, 5, 5, 15, 15, 0.8)
3.1 Wear rate
Depth of cut 1.0mm
Cutting speed 100m/min, 200m/min
The abrasive wear model was verified in Reference Feed rate 0.1mm/rev, 0.2mm/rev
[1] as well as the adhesion model [4]. The accurate Lubrication dry
prediction can be performed in the flank wear -36
progress as long as the appropriate the wear
Measured
characteristic constants are applied to the model.

log(dw/σ・dL) m2N-1
-38 Identified
The wear characteristic constants are identified to
minimize the model error between the predicted and -40
the measured wear rates, which is expressed as:
2 -42
m n ⎡⎛ dVB ⎞ * ⎛ dVB ⎞ ⎤
E1 = ∑∑ ⎢⎜ ⎟ −⎜ ⎟ ⎥ (5)
⎢⎝ dt ⎠ i , j ⎝ dt ⎠ i , j ⎥⎦
i =1 j =1 ⎣ -44
0.0011 0.00112 0.00114 0.00116 0.00118 0.0012
1/θ K-1
where (dVB/dt)*and (dVB/dt) are the predicted and
the measured wear rates. m and n are the number of Fig. 3 Wear characteristic
the sets of the wear progress and that of the wear
rates on the wear progress curves respectively The the model error between the predicted and the
flank wear progress is generally measured with the measured flank wear lands expressed as follows:
initial wear at higher rates than the steady wear ones.
[ ]
m n 2

E 2 = ∑∑ (V ) − (VB )i , j
*
Because the wear model in Eq. (2) or Eq. (3) can be B i, j (8)
applied to the steady wear progresses, the initial i =1 j =1

wear progresses are eliminated from the measured
data. The wear progresses are also observed with
scattering due to the measurement error. The
measured wear progresses with the cutting time t are
approximated by the following equation:
VB = ut v (6)
where u and v are constants, which can be estimated
in the method of least squares. The wear rate can be
given as follows:
dVB
= uvt ( v −1) (7)
dt
The wear rates on the approximated curve at the
observed time are applied to the measured data in Eq.
(5). Because the stress and the temperature
distribution on the flank wear land in the Eq. (3)
change with the wear characteristic constants, the
model error is minimized in the steepest decent
method.
3.2 Initial wear offset
There is no model for the prediction of the initial
wear because the initial wear offsets depend on
uncertain factors such as the machine vibration. This
study determines the initial wear offsets to minimize
where (VB)*and (VB) are the predicted and the
measured wear lands. The flank wear lands at the
observed times on the approximated curves by Eq.
(6) are applied to the measured data in Eq. (8). The
model error in Eq. (8) is also minimized in the
steepest decent method. Being acquired in tested
cutting conditions, the initial wear offsets can be
reduced to the data table. The neural network learns
the table of the initial wear offsets in back
propagation method [7].
4 CASE STUDY
The wear tests are performed in the cutting
conditions as shown in Table 1. The wear
characteristic constants are identified as follows:
C = 8.515412 × 10 −12 [m 2 / N ]⎫
⎬ (12)
λ = 12783.33 [K ] ⎭
Figure 3 shows the identified wear characteristics at
the tool tip in Eq. (2) with the measured data.
The identified initial wear offsets are shown in Table
2. The reference data in Table 2 is used for the
training of the neural network, which consists of 10
units associated with the cutting conditions and the
tool geometry in the input layer, 5 units in the
Table 2. Initial wear offset
Cutting speed Feed rate Initial wear offset
100m/min 0.1mm/rev 0.0951mm Chip Tool Chip
100m/min 0.2mm/rev 0.0829mm Tool
200m/min 0.1mm/rev 0.1068mm
200m/min 0.2mm/rev 0.1032mm

900K
700K
0.2

700K
500K
500K
Flank wear land mm

0.15

900K
0.1
Workpiece
100m/min, simulated Workpiece
0.05 100m/min, measured
200m/min, simulated (a) Cutting speed, 100m/min (b) Cutting speed, 200m/min
200m/min, measured Fig. 5. Cutting temperature
0 Cutting conditions: depth of cut, 1mm; feed rate, 0.1mm/rev.
0 5 10 15 20 25 30 35 40
Time min 600 1200
Fig. 4. Wear prediction based on the identified constants Temperature
Cutting conditions: depth of cut, 1mm; feed rate, 0.1mm/rev. 500 1050
Normal stress MPa

Temperature K
400 900
hidden layer and a unit associated with the initial 300 Normal stress 750
wear offset.
200 600
Figure 4 compares the predicted wear progresses
with the measured ones at a feed rate of 0.1mm/rev. 100 100m/min 100m/min 450
200m/min 200m/min
The wear characteristics are verified with the 0 300
presented identification in agreement of the 0 0.05 0.1 0.15 0.2
Distance mm
prediction with the measurement. Figure 5 shows the
Fig. 6. Normal stress and temperature distributions on the
distributions of the cutting temperature. Figure 6 flank wear land
shows the normal stress and the temperature Cutting conditions are the same as those in Fig. 5.
distributions on the flank wear land.

REFERENCES
5 CONCLUSIONS
The flank wear progresses are predicted in the
abrasive wear model, which gives the wear rate with
the stress and the temperature on the flank wear land.
The wear characteristic constants are identified to
minimize the model error between the predicted and
the measured wear rates. The flank wear prediction
starts at the initial wear offset at cutting time t=0.
The initial wear offsets are also identified to
minimize the error between the predicted and the
measured flank wear lands. A neural network is
trained by the identified offsets. The flank wear
progresses can be predicted accurately in the
abrasive wear model with the identified wear
characteristic constants and the neural network
model trained by the initial wear offsets identified.
1. E. Usui, S. Shirakashi and T. Kitagawa, Analytical
Prediction of Tool Wear, In: Wear, 100 (1984) 129-151.
2. T. Matsumura, T. Obikawa, T. Shirakashi and E. Usui,
Automonus Turning Operation Planning with Predicting
Tool Wear and Surface Roughness, Transactions of the
North American Manufacturing Institution of SME, 21
(1993) 359-366.
3. T. Matsumura, T. Obikawa, T. Shirakashi and E. Usui,
On the Development of Cutting Process Simulator for
Turning Operation, Proceedings of the 6th International
ESAFORM Conference on Material Forming, (2003)
519-522.
4. E. Usui, T. Shirakashi and T. Kitagawa, Analytical
Prediction of Three dimensional Cutting Process (Part 3),
Journal of Engineering for Industry, Transactions of
ASME, 100 (1978) 236-243.
5. Suhas V. Patankar, Numerical Heat Transfer and Fluid
Flow, Hemisphere Publishing Corporation (1980).
6. E. Rabinowicz, L. A. Dunn and P. G. Russell, A study of
abrasive wear under three-body conditions, Wear, 4
(1961) 345.
7. D. E. Rumelhart, J. L. McClelland and the PDP Research
Group, Parallel Distributed Processing, MIT Press,
Cambridge, MA, (1986), p. 219-228.
 < 7 =,
( ) .
# " "A7 : "B
!!!" # " " $ %"% & "
'

!!!" # " " $ & "

(
) # % * + %
!!!" " " $ ,-# -& "
.
) # % * + %
!!!" " " $ & " "

+ 2 + >
, ? =@ $ $ 278
" $ +
$ % "
# $ " $
$ #

8 $ > $ " "+ "%

2
# # 0
# # 0/ '3*

$ $
'3"4"5* "
! " # 2
# #0
$ $ " 6% /$ $
# $
$ 278
! " "
" $
#
%& ' "(")* + , %
! $ '9*
$ " # + : $
# ,
)- $ ';*
" #
( < < %= =2 +
'.*
! " $
# " " , /$
" # $ / : # /
$ # " $ 0 $
0 # '4* $ 0
$ / $ $ "
$ 0 $ : > $ "
, $ " $ + #
/ 1 $ # $ : $ /
 3C5 $
$ 2 $ /
/ $ ?A@" / ?@ ) # ? @
# , $ 0 / $
$ # $ 0 "
< $
$ / #

# %
% < $ B

? @ 4- (.-

% ? @ -( -.

% # $ /$ ? @ -. -9

+ "
# $ /$ : 2 " $ /" # #
, # $ $ ,
" , $ # (
# # $
$ $ > # ( ,
=, A ? @
# ? F @ ? F @ ? @ ? @
!= + ? @ 0 0 0 5- )
( 0 0 0 ((4 9
& ! /$ " ) 0 0 G )-
. 0 0 G 34 .
" " 3 0 G 0 ;- .
?+ @ # > 4 0 G 0 (3 -
5 0 G G 5)
! − 9 0 G G (; 5
,
+= ?(@ ; G 0 0 4
- G 0 0 .;
G 0 G .)
# # ( G 0 G 9
) G G 0 ).
-9 /$ $
. G G 0 59
# # 2 " # 3 G G G -5
$ '4* 2 # 4 G G G (9
/$ ("

# # (
) =2 D % =E7= = 2 #
'9*
, $ " #
>
,
= =

, ? =@
∑ −∑ $
# ,
, " #
# $ $
+ $ 0 # ("
' " $ # # ,
# ' -* (
, " $ $
() # $ $ % (0 " # #
#
 "
$ $ , , # ) + A+ %
# # 2 2 % 7
2! 2! A A

45(45 ;; 4 ( )) (4 .9 I- ---
+0A )-;;) 4- )-;;) 4- 5) ( I- ---
0 ). - 34 ). - 34 9 -4 - - ;3
0 35(3 4 35(3 4 )5 . - ---(
+ (5(. 9. (5(. 9. 4 .. - -) ;
+ )- 9 9 )- 9 9 )- 53 - ---.
);3 -( );3 -( ) )- - -(;
)9 - (( ; .() )4
< / ); -4 ); -4 ). - - ;
%
7 (45 4 9 ))) 9;
=
5 -59 (( 3

% ( =

$ #
$ # #

# # $ )
,
" $ +"
+
#
+ : ?+ A+@
$ # ) + A+ # "
#H 0 /

' *
% . ) #
J-4

. =D =22 + +<K2 2

# $
# #
' (*
$ >

= β - + β 0 + β ( 0( + β ) 0) + β . 0 0( + β 3 0 0) ?)@

& #
% ) # # 0 " 0( " 0) 1
β " β( "
%0 (4 .9
-- β-
%0A !
+ 70A - -3 $ >

= .; (. 0 .. - + 0 . 4- 0 ) )3 + ) -3+ + (9 3)+
# $ +"
$ . #
# # ,
% 3
 # # # ! 1"
, $ M O "
.." ?(--.@ (5 C (9(
9 D : : " 2 8
# " L = 2 2
2 % %
1 " 2 O M " 5(" ?(--4@
53.054-
; M < "M % "+ < "
L+ : $ H
1" % M + " 45" ?(---@
(- C(-)
- = P/" L
D% 7 #
2 + 1" M +
" 7 ?(--3@
= # : /" L+
$
#
+ 5-530 43 1" M + "
% 3 # # 7 ?(--3@
( B 8 M 2 " L+
3 < 2 2 + (-)
# 1" M " (-"
$ / $ ? ;93@ (930 (;-

" $
,
+ +

,
# $ +
# " $

# $
$

=%= = =2

< : " L+
$ /
1" M
7 " (." ?(--(@ ;;C -.
( = & + =# $ " L7
% B 2 1" M
+ " )" ? ;;5@ 0 4
) K K " < " N K K < " L #
#
1" M 7
" 5(" ?(--4@ ..3C.3-
. < 7: # / 2 /" L <
% 7
1" M + " )" ? ;;5@ ())0
()4
3 8 "M 7 D $ " L+
$
1" M + " (9" ?(--4@ 59 C
5;
4 8 H 2 , " L+ #
1" M +
" (4" ?(--3@ 5(4C5)(
5 M + + " L7
Distortions prediction during multi-pass machining simulations
by using the level-set method

O. Pierard1 , J. Barboza1 , M. Duflot2 , L. D’Alvise1 , A. Perez-Duarte3

1
Cenaero, Virtual Manufacturing group
Rue des Frères Wright, 29 - B-6041 Gosselies - BELGIUM
URL: www.cenaero.be
e-mail: olivier.pierard@cenaero.be, josue.barboza@cenaero.be, laurent.dalvise@cenaero.be
2
Cenaero, Multi-scale Material Modeling group
Rue des Frères Wright, 29 - B-6041 Gosselies - BELGIUM
URL: www.cenaero.be e-mail: marc.duflot@cenaero.be
3
SNECMA, Centre de Compétence Industrielle en Usinage
Rue Henri Auguste Desbruères BP81, 91003 Evry cedex, FRANCE
URL: www.snecma.com e-mail: alexis.perez-duarte@snecma.fr
ABSTRACT: In this presentation, an innovative approach for the representation of machining cutting paths
based on the level-set method is proposed. During most machining processes, the highest level of distortions
usually comes from relaxation of residual stresses so that a simulation at the global scale (i.e.: without taking
into account chip forming and tool/workpiece interactions) is sufficient for the prediction of shape error. A
classical finite elements approach requires that the workpiece mesh matches the different cutting paths. Fur-
thermore, given the fact that cutting paths are given in the initial configuration while the workpiece deforms
during the process, numerous remeshing steps are needed. On the contrary, making use of the level-set (signed
distance map) enables to define independently cutting paths and workpiece mesh; only the level-sets need to be
recomputed between two operations.

KEYWORDS: Machining, distortions, residual stresses, level-set

1 INTRODUCTION 2 REPRESENTATION OF CUTTING PATHS

Modeling the machining process through classical fi-

One of the major difficulties of the machining process
is to control the final shape which can deviate from
the nominal one. Major causes of distortions come
from relaxation of residual stresses, cutting forces, ef-
forts due to clamping systems and vibrations of the
tool and/or workpiece [3]. In this contribution, effect
of residual stresses is examined, which is for many
workpieces the most critical effect. More precisely,
residual stresses are generated during previous oper-
ations such as forging and heat treatments. However,
these are redistributed during the machining operation
and lead to undesired distortions which can be out of
tolerance bounds. Numerical prediction of this effect
is thus very important but complex due to the numer-
ous parameters involved (material properties, number
of cutting paths, paths geometry,...).
nite element simulations using conforming meshes
faces a major limitation: its flexibility. Indeed, each
path requires a remesh as it needs to match the final
boundary. Moreover, cutting paths cannot be defined
within the workpiece mesh once before the simula-
tion. In order to circumvent these limitations, an in-
novative approach based on the level-set has been de-
veloped within our software MORFEO (for Manufac-
turing ORiented Finite Element tOol). The level-set
is a signed distance function defined at the nodes of
the finite element mesh. It is commonly used to repre-
sent material interfaces, holes, cracks, etc (see [1] and
[2]). In this application, it is used to represent machin-
ing paths and proper integration algorithms are used
according to the level-set sign. This approach pro-
vides us the required flexibility to simulate multi-pass

1
machining process by avoiding expensive remeshing. side, as illustrated on Figure 2.
Furthermore, level-sets do not deform with the mesh
but are recomputed on an updated configuration af-
ter each pass so that distortions occurring during the
process are taken into account. These are computed
by considering a new free boundary conditions along
the cutting path represented by the level-set. A new
equilibrium is found and gives the corresponding dis-
placement field.

3 INDUSTRIAL APPLICATION

This approach has been initially developed for two-

dimensional axisymmetric industrial disks as illus-
trated on Figure 1 before being extended to fully
three-dimensional problems. In this case, the algo-
rithm to compute the level-set is modified (compu-
tation of a point-triangle distance instead of point-
segment).
Typically, the input of this problem consists of
the residual stress map obtained from the simulation
of forge and heat treatment operations. Intermedi-
ate geometries between the various passes are also
given so that position of the cutting paths can be com-
puted. The considered material can be considered as
linear elastic given the level of residual stresses and
the global unloading during the process.
Figure 2: Various cutting paths considered for this operation
The validation of the method has been conducted
in close collaboration with SNECMA. Typically, ex-
perimental measurements at different points of the
section on several specimens in order to avoid the
dispersion effect are confronted to numerical results.
This comparison is done at the end of each operation
(i.e. machining of both sides) and before and after
removal of the clamping system. Some typical exam-
ples of distortions so obtained are illustrated on Fig-
ure 3 and 4. On Figure 3, distortions induced by the
first machining path (which gives the highest level of
distortions) are presented. At the end of the first op-
eration (machining of the upper side), clamping sys-
tems are removed and gives the displacement field il-
lustrated on Figure 4.

Figure 1: Final machined workpiece containing a section before

any operation

For this simulation, three cutting paths of decreas-

ing thickness are considered on the upper side of the Figure 3: Axial distortions induced by relaxation of residual
disk, then clamping system is removed and finally, stresses after one machining pass (clamping system retained)
one machining operation is performed on the lower

2

Figure 4: Axial distortions induced by relaxation of residual
stresses after one machining pass (after removal of clamping
system)
During the machining of the upper part of the disk,
fixations are retained the central part of the disk tends
to go up, which affects subsequent passes. Once the
clamping system is removed, an higher level of dis-
placements occur over the whole workpiece.
4 CONCLUSIONS
During this work, the level-set method has been suc-
cessfully adapted to multi-pass machining simula-
tions. For this, the iso-0 level-set represents the cur-
ring cutting path and the configuration is updated at
3
the end of each pass so that distortions occurring dur-
ing the process are taken into account.
Among the numerous prospects of this work, an
automated optimization loop has been developed in
order to study the influence of the following design
parameters: final workpiece position within the raw
material, number and geometrical position of tool
paths and the effect of the position of the clamping
systems.
Finally, a coupling with a local-scale model is cur-
rently under development so that cutting forces can
be taken into account. The cutting forces are obtained
using a semi-analytical procedure: numerical simula-
tions are performed at the local scale for a nominal
tool shape while an analytical model is used to ac-
count for shape variations and/or different tool orien-
tations.
REFERENCES
[1] T. Belytschko, N. Moës, S. Usui, and C. Parimi. Arbitrary
discontinuities in finite elements. Int. J. Num. Meth. Engrg.,
51:943–960, 2001.
[2] Y. Guétari, S. Le Corre, and N. Moës. Etude des possi-
bilités de la méthode x-fem pour la simulation numérique
de la coupe. Mécanique et Industries, 6:315–319, 2005.
[3] D. Hornbach and Prevéy. Development of machining pro-
cedures to minimize distortion. In Proceedings of the 17th
Heat Treating Society Conference and Exposition, pages
13–18. ASM, 1998.
Modelling of the residual stresses induced by belt finishing on a
AISI52100 hardened steel
J. Rech1, G.Kermouche1, C. Claudin1, A.Khellouki1, W.Grzesik2
1
Ecole Nationale d’Ingénieurs de Saint-Etienne, LTDS, 58 Rue Jean Parot – 42000 Saint-Etienne - France
e-mail: joel.rech@enise.fr
2
Opole University of Technology, Department of Manufacturing Engineering and Production Automation,
Poland
e-mail: w.grzesik@po.opole.pl

ABSTRACT: New technological process consisting of hard turning followed by belt finishing, in place of the
widely used method in industry, e.g., grinding, has lately been launched in the automotive industry. This is
because, many transmissions parts, such as synchronizing gears, crankshafts and camshafts require superior
surface integrity along with appropriate fatigue performance. This paper provides a modelling of part residual
stresses produced after belt finishing in order to provide some explanations about the experimental results
previously obtained. Indeed, it has been shown that the belt finishing process improves very significantly the
surface integrity by the induction of strong compressive residual stresses in the external layer. The model has
confirmed this trend. It has also revealed that, among the process parameters of the belt finishing technique,
the lubrication appears as a key parameter to get compressive stresses, whereas the elementary force on each
abrasive grains influence the depth of the affected layer.

Key words: Hard turning, Belt finishing, Superfinishing, Residual stresses, Finite Element Model.

<< 0.1 µm and much more suitable flat BAC shape.

1 INTRODUCTION
Hard turning of hardened steel with c-BN cutting
tools can be essentially performed as precision
operation when the surface roughness Ra parameter
is less than 0.3 µm [1]. Precision finishing of
hardened steel components offers manufacturers an
attractive alternative to traditional grinding. In
particular, it can often cut manufacturing costs,
decrease production time, and improve overall
product quality [1]. Despite many advantages of
hard turning technology, the following limitations
can be distinguished, namely [1-4]: high sensitivity
to tool wear leading to unacceptable surface
roughness, white layers and tensile residual stresses.
Additionally this technique necessitates using
limited values of cutting speeds if tensile stresses
have to be avoided in the external layer, which limits
its productivity.
Superfinishing processes, such as belt finishing or
honing or roller burnishing, are proposed in addition
to hard turning to improve the surface texture [2,4-
6], allowing very good surface roughness with Ra
Moreover the surface finish becomes much more
stable and independent from the deviation induced
by the wear of the c-BN inserts in HT, which is a
strategic improvement for large scale productions in
fully automatic production systems.
However, few works have been made to investigate
the effect of these superfinishing processes on the
residual stresses obtained. As far as belt finishing is
concerned, this process induces strong compressive
residual stresses in a narrow layer ~ 5-10 µm of a
AISI52100 hardened steel in any direction tangential
to the surface since [6]. So, the aim of this paper
deals with the modelling of the modifications due to
the belt finishing operation on the residual stress
state previously obtained by hard turning.
2 DESCRIPTION OF BELT FINISHING
It is necessary to provide some details about the
kinematic of this superfinishing technique in order to
understand the principle of the model. As shown in
Fig. 1, the belt finishing technique consists in
applying an abrasive belt between the surface and a
soft roller. Abrasive belts have a single layer of
Al2O3 abrasive grains stocked on an elastic paper
strip reinforced with fibres. A force is applied by a
pneumatic jack on the polymeric roller in order to
obtain a sufficient pressure in the contact between
abrasive grains and the surface. At the microscopic
scale, the contact is localized within the peaks of
abrasive grains (Fig. 1). The pressures developed in
the local contact zones are high enough to perform
material removal and to induce local plastic
deformations. The rotation of the workpiece is
necessary to obtain a homogeneous surface but the
material removal process is mainly due to the axial
oscillation of the belt+roller system. The belt
finishing process necessitates a plain oil lubrication.
The belt finishing duration is very limited: few
seconds. Two steps have to be considered. As
described by [6], during the first step, the belt
eliminates very quickly the peaks of the surface
texture until abrasive grains reach the lower part of
the surface texture. After this period, the shape of
the surface texture remains constant and is no more
improved. The second step consists in a rubbing
phase of abrasive grains. The workmaterial is
ploughed on each side of grains.
Belt feed
Pneumatic jack
Workpiece
rotation
Oscillation of
the belt+roller
Abrasive belt
system
(worn)
Lubrication Abrasive belt Abrasive grains
Belt structure
Workpiece
Polymer roller
Fig. 1. Principle of belt finishing process.
3 NUMERICAL APPROACH
The objective of this paper is to understand the
mechanisms leading to the evolution of the residual
stresses towards compression in a very thin external
layer.
3.1 Context of the model
The stress state induced by belt finishing originates
during the second step of the process consisting in a
rubbing phase. In this period, a large amount of the
abrasive grains are much rounder. As a consequence,
the main work of grains consists in scratching and
rubbing the surface. It has been assumed that, from a
statistical point of view, all abrasive grains have a
similar shape and are indented into the workmaterial
with a similar force. So, we have only examined the
action of a single grain of the belt on the workpiece
during a scratching phase (Fig. 2). Of course, each
area of a workpiece is scratched by a large quantity
of different grains having different trajectories. So
the action of a single grain provides only a trend.
However, multiple scratching does modify the trend.
3.2 Assumptions related to the local physics of the
belt finishing process
The relative motion between abrasive grains and the
workmaterial is small enough to assume that the
Abrasive belt
(fresh) local temperature reached during this technique
should be low and smaller than the critical
temperature inducing microstructure
transformations. This assumption is in accordance
with the experimental observations made during the
investigations [6]. Consequently, temperature effects
are not considered in the present model.
The modelling of scratching of a single grain on a
AISI52100 (Fig. 2) can be performed using a finite
element modelling based on previous developments
by [7-8]. The indentation force P applied by a single
grain on the surface can be determined by the
macroscopic force F (~ 340N), the grain size (30µm
diam.) and n: number of grains in the contact. Grains
Polymer roller
are considered as rigid spheres with no space
between them (Fig. 2). Then, Hertz theory of contact
(rigid sphere over a linear elastic half space [9])
makes it possible to obtain a first approximation of
the average penetration depth h of one grain during
sliding. In this work, P~0.6 N and h~1 µm.
3.3 Finite element model
The finite element software used is SYSTUS. The
developed model is based on the previously
published work of [8]. Loading is achieved by
monitoring the displacement of the grain, which is
first pushed down vertically into the workpiece and
then pushed horizontally for the study of scratching.
The finite element domain is a right-angled
parallelepiped. The x-axis and z-axis are
respectively the scratching and indentation axes. The
plane y = 0 is a symmetry plane (Fig. 2). The mesh
is constituted of 8-node-brick elements with a
selective reduced integration scheme to ensure
plastic incompressibility. In this study, the material
behavior is modelled using large deformation
(Updated Lagrangian Formulation) and elastoplastic
theory. Because of their high hardness, hardened
steels have a high yield stress but cannot ensure high
plastic deformation. The AISI52100 hardened steel
mechanical properties are chosen in relation to the
study of [11]. Due to the low ratio h/R -penetration
depth over abrasive grain radius- the element
distortion is limited and it does not require the use of
remeshing procedure [8,12].
Fig. 2. Modelling of the action of a single abrasive grain on the
workpiece.
The nature of the contact grain / workpiece being
unknown (frictionless or adhesive contact), the two
following cases are investigated:
• frictionless contact: The macroscopic friction
depends only on the plastic shearing in the
bulk of the material.
• local friction modelled using a Coulomb
coefficient : µ = 0.2.
3.4 Numerical results
The extraction of the residual stress profiles induced
by the scratching is illustrated in Fig. 3 and the
profiles obtained are plotted in Fig. 3 and 4.
In the first case without any local friction (lubricated
contact), scratching induces plastic yielding and as a
consequence high compressive stresses (around
2000 MPa) in the external layer (around 10 µm). As
a consequence, tensile residual stresses appear in the
sublayer in order to respect the mechanical
equilibrium.
In presence of local friction, a much more important
plastic yielding occurs leading to a peak of tensile
stresses (around 500 MPa) in the very thin external
layer (around 200 nm). The peak of compressive
stresses is lower (around 1500 MPa) in presence of
local friction but the trend is similar for both curves.
Moreover the use of plein oil as lubricant during the
belt finishing operation makes it possible to assume
that the friction can be neglected. As a consequence,
the avoidance of friction phenomena between
abrasive grains and the workmaterial seems to be
more relevant.
Slid Zz
ing
di rect y
Z io n x
Y X
Residual stress, σ11 (MPa)
1000
0
0 0,01 0,02 0,03
-1000
-2000
-3000
Distance from the surface (mm)
Fig. 3. Extraction of the σ11 residual stress profile after sliding
in the steady state of the scratch.
In both cases, the residual stress field is affected
over a depth of 10 µm which is in agreement with
experimental observations showing an affected zone
around 6 µm since [6]. Even if the absolute
difference is not so much important, the problem is
related to the inaccuracy of the penetration depth
computation due to the different assumptions we
used previously. Fig. 9 shows residual stress profiles
obtained numerically without any local friction and
with different penetration depth: 0.25 µm to 2 µm. It
is shown that the size of the affected zone highly
depends on the penetration depth of abrasive grains
(e.g. on the local force applied on the grain). The
higher the penetration is, the deeper the affected
layer is. Nevertheless the shape of the curves
remains similar. In practice, penetration depth
depends on the macroscopic force applied by the
system (the pneumatic jack), on the rigidity of the
polymeric roller, on the grain size and on the density
of grains.
Residual stress, σ11 (MPa)
1000
500 0 0,01
0 load applied on the grain
-500
-1000
-1500
No local friction : µ=0
-2000
With local friction : µ=0,2
-2500
Distance from the surface (mm)
Fig. 4. Distribution of the residual stress σ11 on the sub layer
for a penetration depth of 1µm (FE results).
Residual stress, σ11 (MPa)
0 0,01
1000
500
0 finishing process on the surface texture obtained after
-500
-1000
-1500
-2000
-2500
Distance from the surface (mm)
Fig. 5. Distribution of the residual stress σ11 on the sublayer
for different penetration depth h and without any local friction
(FE results)
These analyses show that belt finishing affects an
external layer mainly by strain hardening and
induces a compressive field along some
micrometers; its thickness depending on the average
penetration of the abrasive grains in the contact
between the belt and the workpiece (i.e. on the
macroscopic force applied by the system, on the
rigidity of the roller, on the grain size and on the
density of grains).
4 CONCLUSIONS
The objective of the paper was to investigate the
modifications of the surface integrity induced by a
belt finishing operation on hardened bearing steels.
In parallel to the well-known great surface texture,
previous work have shown experimentally that belt
finishing process generates compressive residual
stresses on surfaces. This paper has shown by means
of a finite element analysis, that the effects of belt
finishing on the residual stress distribution can be
explained by the action of abrasive grains rubbing
the surface at the end of a belt finishing operation. In
standard conditions, compressive stresses are
induced in the skin.
The compressive layer affected by belt finishing is
strongly related to the lubrication. A lack of
lubrication could induce tensile stresses.
0,02 0,03
The depth of this affected layer depends on the local
REFERENCES
1. F. Klocke, E. Brinksmeier, K. Weinert, Capability
profile of hard cutting and grinding processes, CIRP
Annals 54(2) (2005), 557-580.
2. H.K. Tönshoff, C. Arendt, R. Ben Amor, Cutting of
hardened steel, CIRP Annals 49(1-2) (2000), 431-434
and 547-566.
3. J. Rech, A. Moisan, Surface integrity in finish hard
0,02 0,03 turning of case-hardened steels, International Journal of
Machine Tools and Manufacture, 43 (2003), 543-550.
4. A. Khellouki, J. Rech, H. Zahouani, Influence of the belt
h = 0,25 µm
hard turning, Journal of Engineering Manufacture,
h = 0,50 µm
h = 0,75 µm (2007), 221, Iss. 7 (2007), 1129-1137..
h = 1,00 µm
h = 1,50 µm 5. A. Coulon, Effect of the honing drum upon the
h = 2,00 µm inducement of compressive residual stresses, Journal of
Mechanical Working Technology, 8 (1983), 161-169.
6. J.Rech, G.Kermouche, W.Grzesik, C.Garcia-Rosales,
A.Khellouki, V.Garcia-Navas. Characterization and
modelling of the residual stresses induced by belt
finishing on a AISI52100 hardened steel, Journal for
Materials Processing, Accepted for publication, (2008).
7. M. Barge, G. Kermouche, P. Gilles, J.M. Bergheau,
Experimental and numerical study of the ploughing part
of abrasive wear, Wear, 255 (2003), 30-37.
8. G. Kermouche, J.M. Bergheau, Simulation numérique de
la rayure des matériaux, European Journal of
Computational Mechanics, 15 (2006), 221-232.
9. K.L. Johnsson, Contact mechanics, Cambridge
University Press (1985).
10. J.M. Challen, P.L.B. Oxley, An explanation of the
different regimes of friction and wear using asperity
deformation models, wear, 53 (1979), 229-243.
11. G.Poulachon, Aspects phénoménologiques, mécaniques
et métallurgiques en tournage c-BN des aciers durcis, Ph.
D. Thesis, ENSAM Cluny, France, (1999).
12. G. Kermouche, J.L. Loubet, J.M. Bergheau, A 3D finite
element pocedure for modelling scratch test, Proceedings
of the 10th International Conference Metrology and
Properties of Engineering Surfaces, Saint-Etienne, ISBN
2-86272-389-4, (2005), 311-320.
Prediction of Tool Wear Progress in Machining of Carbon Steel using
different Tool Wear Mechanisms
D. Umbrello1, L. Filice1, F. Micari2, T. Matsumura3, T. Shirakashi3
1
Dep. of Mechanical Engineering, University of Calabria – 87036 Rende (Italy)
URL: www.unical.it e-mail: d.umbrello@unical.it; l.filice@unical.it
2
Dep. of Manufacturing and Management Engineering, University of Palermo – 90128 Palermo (Italy)
URL: www.unipa.it e-mail: micari@ dtpm.unipa.it;
3
Dep. of Mechanical Engineering, Tokyo Denki University 2-2 Kanda Nishiki-cho, Chiyoda-ku, Tokyo, 101-
8457, (Japan)
URL: www.skynet.m.dendai.ac.jp e-mail:tmatsumu@cck.dendai.ac.jp; shirakashi@cck.dendai.ac.jp
ABSTRACT: In this paper the prediction of tool wear on carbide uncoated tools was taken into account. In
particular, two different tool wear models based on the diffusion mechanism and on the abrasion mechanism
were considered. The calibration of the utilized models was done using the results obtained by experimental
analysis performed on an orthogonally machined AISI 1020 tube. Once the calibration was executed,
numerical simulations, for both the utilized tool wear models, were simultaneously performed with the aim to
test the capability of the proposed numerical procedure. The comparison between the two tool wear
mechanisms for predicting the flank tool wear is discussed in the paper.

Key words: Tool Wear, AISI 1020, Orthogonal Cutting, FEM.

The present paper was developed according to the

1 INTRODUCTION
Machining processes are undergone to relevant
improvements in the last years due to the
development of high speed machines that allow to
carry out some works on hard metals with high
productivity and, sometimes, avoiding other
expensive steps of manufacturing. On the other
hand, cutting speed increasing induces a significant
worsening of all the wear related aspects. The high
speed, in fact, generates a strong heat amount on the
tool that may rapidly go out of service. Furthermore,
due to the world wide diffuse attention to the
environmental safeguard, the use of lubricant and
coolant is fully dissuaded. For the above discussed
reasons, the research focused on materials and
geometry has become a strategic key of success for
all the industries that produce metal cutting tools.
From a pure scientific point of view all the
phenomena related to wear in cutting are well
known and they are accurately discussed in technical
literature [1, 2]. The new challenge is the
implementation of effective models in a finite
element environment in order to perform a powerful
methodology for tool designers and developers.
above considerations. In fact, the aim was to verify
the capability of two different tool wear models. The
former is based on the diffusion mechanism while
the latter takes into account the abrasion mechanism.
The calibration of the two models was executed
using the experimental flank wear data revealed by
an optical microscope on the utilized uncoated ISO
P25 tool. The experiments were conducted on an
orthogonally machined AISI 1020 tube. Once the
calibration was executed, the empirical tool wear
models were implemented in the numerical codes in
order to validate their capabilities. Finally, the
comparison between the two approaches for
predicting the flank tool wear is reported.
2 EXPERIMENTAL TESTS
A plan of experiments was designed in order to
calibrate, by an inverse approach, the wear models
based on diffusion and abrasion mechanisms.
The workpiece was a 2.5 mm thick AISI 1020 tube,
with an initial diameter equal to 115 mm. Uncoated
carbide tools, ISO P25, characterised by a relief
angle α=6° were used all over the tests. The
experiments were stopped after a cutting time of 4/5 3 THE PROPOSED WEAR MODELS
minutes. For this value the wear rate can be
considered stationary and, therefore, not dependent Wear due to abrasion and diffusion mechanisms
on the cutting time. Furthermore, in order to exalt appears to play the major role in the continuous dry
wear phenomenon and temperature increasing no cutting of steel with tungsten carbide tools. Wear is
lubricant was used during the tests. Flank tool wear proportional to the cutting distance and closely
was measured using an optical microscope (200X). related to the shape, hardness and distribution of the
Table 1 reports the average flank wear data after the abrasive particles. On the contrary, the diffusion
mentioned cutting time varying cutting speed, feed mechanism is a physico-chemical contribution
rate and rake angle, γ. associated with the temperature.
Thus, in order to investigate which mechanism is
Table 1. Experimental flank wear land after 5 minutes cutting predominant when AISI 1020 work material is
time and predicted average temperature on flank land by FEM. machined with uncoated WC tools, the two
Cutting speed Feed rate Rake Flank wear Predicted TAVE
[m/min] [mm/rev] angle [mm] [K]
mechanisms were calibrated through the same
120 0.050 0° 0.400 808 experimental data and, then, separately applied
120 0.125 0° 0.441 943 during validation.
120 0.200 0° 0.510 1088
180 0.050 0° 0.429 868 3.1 Diffusion wear model
180 0.125 0° 0.461 1023
180 0.200 0° 0.536 1148 The diffusion wear model can be derived from
240 0.050 0° 0.438 923
240 0.125 0° 0.482 1073
Takeyama and Murata [4]:
∂w ⎛ E ⎞
240 0.200 0° 0.615 1213 = D ⋅ exp⎜⎜ − ⎟⎟ (1)
155 (4min) 0.120 -7° 0.305 867 ∂t ⎝ RT ⎠
235 (4min) 0.075 7° 0.219 1007 being D a material constant, E the activation energy
(75,35 kJ/mol), R (8,314 kJ/mol K) the gas constant
Table 1 also reports the predicted average
and T the local temperature, measured in K. The
temperature by FE numerical simulations along the
predicted wear rate δw/δt, defined as the lost volume
flank land where wear occurs. These numerical
for unit of surface and time, is strongly dependent on
results are needed to calibrate the wear models by an
the predicted temperature; on the other hand, the
inverse approach. It is important to highlight that the
calibration of parameter D becomes a strategic task.
simulative capability on temperature prediction by
Simple geometrical considerations permit to
FEM was already verified in a previous work [3]
calculate flank wear land VB through the following
conducted by some of the present authors.
equation:
In turn, the predictive capability of the wear models
∂w 2⋅t
will be tested taking into account a new set of VB = VB 0 + ⋅ for γ > 0°
experimental results. The latter were conducted ∂t (1 − tgα ⋅ tgγ ) ⋅ tgα
∂w 2 ⋅ t ⋅ cos(α − γ )
varying of both the cutting parameters and cutting VB = VB 0 + ⋅ for γ < 0° (2)
time in order to verify the generalization capability ∂t cos γ ⋅ sin α
of the derived laws (Table 2). being t the cutting time (in minutes), α the relief
angle, γ the rake angle and VB0 the value of the
Table 2. Testing experimental flank data and predicted average sudden flank wear which occurs in the former few
temperature on flank land by FEM.
Process Cutting Flank wear Predicted
cutting instants. In fact, it is well known from the
parameters time [min] [mm] TAVE [K] experimental evidences that the tool undergoes to a
VC=155m/min 1 0.124 sudden wear when machining begins.
1 f=0.100 mm/rev 2.5 0.251 939 An inverse procedure was utilized to estimate the
γ=0° 5 0.453 unknown parameter D in the assumed wear model,
VC=235m/min 1 0.130 based on the experimental data reported in Table 1
2 f=0.075 mm/rev 2.5 0.255 977
γ=0° 5 0.462 and the predicted cutting temperature on the flank
VC=200m/min 1 0.141 land. In particular, the best interpolation of D value
3 f=0.180 mm/rev 2.5 0305 1093 was obtained by a third order polynomial law
γ=0° 5 0.598 function of the temperature, as illustrated in the
following equation:
D(T ) = 4.0814 ⋅10 − 11 ⋅T 3 −1.1279 ⋅10 − 7 ⋅T 2 + (3) temperature according to the FE code database. As
+ 1.0672 ⋅10 − 4 ⋅T − 3.0387 ⋅10 − 2
far as friction is regarded, a simple constant shear
where T is the local temperature, measured in K, model was implemented and the friction factor, m,
calculated by the 2D numerical simulations. was set equal to 0.82.
Furthermore, the diffusion wear rate equation (3),
3.2 Abrasion wear model [5] including the D(T) third order function was directly
implemented in the FE code by a proper user
The abrasion wear model can be derived from the subroutine. In this way, the code was able to take
Rabinowicz’s [6] equation: into account the flank wear evolution through a tool
σt geometry updating procedure [7]. Table 3 shows the
dW = K dL (4) measured flank wear and its prediction when only
H
where dW, dL and σt are the wear volume, the diffusion wear model is taken into account.
sliding distance and the normal stress respectively. K Table 3. Comparison between experimental flank wear and
is a constant expressed as: predicted ones due to diffusion wear model.
⎛ A ⎞ Cutting time EXP Flank NUM Flank E%
K = A1 exp⎜ − 2 ⎟ (5) [min] wear [mm] wear [mm]
⎝ T ⎠ 1 0.124 0.126 1.6%
where A1 and A2 are constants. T is the local 1 2.5 0.251 0.280 11.6%
temperature. H is the hardness of the material which 5 0.453 0.451 -0.4%
can be associated with the following equation: 1 0.130 0.129 -0.8%
2 2.5 0.255 0.252 -1.2%
⎛B ⎞
H = B1 exp⎜ 2 ⎟ (6) 5 0.462 0.461 -0.2%
⎝T ⎠ 1 0.141 0.139 -1.4%
where B1 and B2 are constants. The abrasion wear 3 2.5 0305 0.290 -4.9%
5 0.598 0.575 -3.8%
model can be expressed as the following equation by
substituting Eq.(5) and (6) into Eq. (4): 4.2 Prediction of flank tool wear by abrasion model
dW ⎛ λ ⎞
= C r exp⎜ − r ⎟ (7)
σ t dL ⎝ T ⎠ The flank wear rate (dVB/dt) can be given by the
The flank wear progresses can be predicted by Eq. following wear characteristic equation:
(7). The constants Cr and λr are identified in the dV B ⎛ λ ⎞⎛ 1 ⎞
= C r σ f exp⎜ − r ⎟・⎜⎜ − tan α ⎟⎟V (9)
wear progresses in the cutting conditions as shown dt ⎜ θ ⎟ ⎝ tan γ ⎠
⎝ f ⎠
in Table 1. The wear characteristic constants are
identified as follows: where σf and θf are the normal stress and the
temperature on the flank wear land; and V, α and γ
C r = 1.065331×10 -12 ⎫
⎬ (8) are the cutting speed, the rake angle, and the relief
λr = 8653.147 ⎭ angle. The temperature distribution can be given in
finite volume analysis [8] with assuming σf and τf on
the flank wear land. Then the wear rate is calculated
4 VALIDATION in Eq. (9). Since the wear rate is generally constant
over the flank wear land, σf and τf are modified so
4.1 Prediction of flank tool wear by diffusion model that the wear rate is the same over the flank wear
The machining process was modelled by means of land. Consequently, the wear rate and the stress and
the SFTC-Deform-2D finite element code using a the temperature distributions can be determined.
coupled thermo-viscoplastic Lagrangian model with Then the flank wear VB at the cutting time t can be
isotropic strain hardening. A plane-strain coupled predicted by the following equation [9]:
t ⎛ dV ⎞
thermo-mechanical analysis was carried out. The VB (t ) = VB 0 + ∫ ⎜ B ⎟dt (10)
workpiece was initially meshed by means of 5000 0
⎝ dt ⎠
iso-parametric quadrilateral elements while the tool, where VB0 is the initial wear offset, which is the
modelled as rigid, was meshed into 1000 elements. width of flank wear land at the time t=0 in the
The Oxley’s law was implemented to describe calculation. The initial wear offsets are determined
material flow as a function of strain, strain rate and to minimize the error between the simulated and the
measured the flank wear lands in the conditions of in this paper with the aim to highlight their
experiments. The initial wear offset, which depends difference in simulative prediction. In particular,
on the machining vibration on the machine tool, can both the diffusion and the abrasion wear model
be estimated by a neural network. were, firstly, calibrated by inverse approach and,
Table 4 shows the simulated and the measured flank then, separately tested to verify their effectiveness.
wear progress, where the wear characteristic Nevertheless the models are conceptually very
constants shown in Eq. (8) are used in the simulation. different, both of them supply a good wear progress
prediction.
Table 4. Comparison between experimental flank wear and This is not very easy to justify from a pure
predicted ones due to abrasion wear model.
Cutting time EXP Flank NUM Flank E%
theoretical point of view since it’s well known that
[min] wear [mm] wear [mm] abrasive phenomenon becomes relevant at low
1 0.124 0.120 -3.2% cutting speed, when temperature decreases.
1 2.5 0.251 0.252 0.4% Thus, in the opinion of the authors, the effort of
5 0.453 0.505 11.4% researchers in the next future have to be focused on
1 0.130 0.129 -0.8% the development of a criterion closer to the process
2 2.5 0.255 0.257 0.8%
5 0.462 0.480 3.9% physics. This approach is strongly required since the
1 0.141 0.153 8.5% only way to develop a model able to be generally
3 2.5 0305 0.295 -2.3% applicable is to model what really occurs.
5 0.598 0.521 -12.9% The development of a procedure which partially or
totally activates, through control variables (i.e.
temperature, etc..) all the known mechanisms
5 RESULTS DISCUSSION probably could constitute a good trade-off, both
from a purely scientific and pragmatic point of view.
Analysing the results reported in Tables 3 and 4 it
can be outlined that, in general, the two proposed
models adequately furnished a good prediction of REFERENCES
the flank tool wear. This is mainly due to the 1. M.C. Shaw, Metal Cutting Principles (2nd Ed.), Oxford
empirical calibration which permit to obtained Science Publications (2005).
effective tool wear models. 2. V.P. Astakhov, The assessment of cutting tool wear,
In fact, paying particular attention at the Eq. 1 and 4, International Journal of Machine Tools & Manufacture,
it is easy to verify that they have the “same 44 (2004) 637-647.
3. L. Filice, F. Micari, L. Settineri, D. Umbrello, On the
structure”. If the material constant D in Eq. (3) is finite element simulation of thermal phenomena in
assumed to work as the hardness or the contact machining processes, Advanced Methods in Material
stress, depending on the temperature, Eq. (1). can be Forming, (2007) 263-278.
reduced to Eq. (7). D(T) and –(E/R) indicate the 4. H. Takeyama, T. Murata, Basic investigations on tool
stress and the temperature sensitivity. Therefore, wear, Trans. of the ASME - J. Eng. Ind., 85 (1963) 33-
38.
both of equations well follow wear progress though 5. E. Usui, T. Shirakahi and T. Kitagawa, Analytical
the equations are derived from the different model. Prediction of Cutting Tool Wear, Wear, 100 (1984) 129-
What is more, results in Table 3 are more accurate 151.
than that of Table 4 because the stress sensitivity is 6. E. Rabinowicz, L. A. Dunn and P. G. Russell, A study of
expressed precisely as the high order function in Eq. abrasive wear under three-body conditions, Wear, 4
(1961) 345.
(2). Eq. (7) is defined the wear rate explicitly as the 7. L. Filice, D. Umbrello, F. Micari, L. Settineri, Wear
function including the stress, where the stress modelling in mild steel orthogonal cutting when using
sensitivity is estimated to be linear. The error is uncoated carbide tools, Wear, 262/5-6 (2007) 545-554.
caused by the non-linear effect of the stress on the 8. Suhas V. Patankar, Numerical Heat Transfer and Fluid
wear rate. Flow, Hemisphere Publishing Corporation (1980).
9. T. Matsumura, T. Obikawa, T. Shirakashi and E. Usui,
Automonus Turning Operation Planning with Predicting
Tool Wear and Surface Roughness, In: Transactions of
6 CONCLUSIONS the North American Manufacturing Institution of SME,
21 (1993) 359-366.
Two different tool wear mechanisms were analysed
Laser machining in lamellar cast iron
C. Vincent, G. Monteil, T. Barrière, J.G. Gelin
Institut FEMTO-ST, LMARC, 24 chemin de l’épitaphe, 25000 Besançon, France.
URL: www.femto-st.fr e-mail: christophe.vincent@ens2m.fr; guy.monteil@ens2m;fr;
thierry.barriere@ens2m.fr; jean-claude.gelin@ens2m.fr

ABSTRACT: Today, the laser machining is commonly used in micromachining. As a matter of fact
laser allows surfaces to be textured with small dimensions in a wide range of applications.

This work presents different laser surface textures machined in a heterogeneous material. The
structure can be lines, dimples or prisms. The ratio between the width and the depth of the textured
forms has been set to 10, with a depth about 10 µm. As the textured material was lamellar cast iron
which is a heterogeneous material, the laser machining parameters need to be set precisely in
order to obtain the desired dimensions of the grooves with a sufficient accuracy. In addition to the
power adjustment of the laser, the machining strategy has to be determined experimentally with the
aim to reach a high level of surface roughness quality. This goal can be achieved by tilting the
galvanometric scanner allowing the laser beam to be tilted up to 20° [1].
In addition, to improve the quality of the laser machining, a laser polishing can be made thanks to
an offset between the focal point of the laser beam and the surface [2]. Consequently, the metal is
just melted and the Ra roughness parameter can be decreased.

At least in order to characterize the laser patterned surface, an InfiniteFocus® topomicroscope has
been used. A microscope is combined with a camera allowing 3D images to be acquired. By means
of an area color focus sensor and thanks to a motorized table, the camera shoots a set of images of
the sample at different focusing depths. For each image, the software compares the contrast of
each point with its neighbors and builds a 3D image by means of these points. Such data sets allow
optical image, topographic image, profile, surface, volume to be obtained.
Key words: Laser surface texturing, InfiniteFocus®, heterogeneous material, laser polishing.

parameters have to be adjusted.

1 INTRODUCTION

Laser machining is an alternative technology to the 2 LASER SURFACE TEXTURING

traditional mechanical machining. The material is
ablated and not deformed. This technique allows the
machining of various material and geometries thanks
to the knowledge of the interactions between the
material and the laser beam. The field of
applications is very wide and in this particular study
laser surface texturing has been chosen with the aim
to improve the tribological properties of a lubricated
contact [3-6].
This paper presents the laser machining of a strongly
heterogeneous material: lamellar cast iron. The
machining of textured surfaces by a laser shows the
skills of this technique. Nevertheless, to read the
desired dimensions and surface quality, various
2.1 Laser machine
The laser machine is an industrial equipment: a
DML 40 SI of Gildemeister (Germany). The laser is
a Nd:YAG (wavelength = 1064 nm) with a mean
power of 100 Watts. The diameter of the laser beam
is 40 µm. The pulse duration is about of some µs
and the frequency of the impact about some kHz.
The laser machine is equipped with a galvanometric
scanner which allows the laser beam to be tilted up
to 20°.
The process of laser machining is comparable to the
commonly machining with different steps:
• CAD
• CAM
• Laser machining
• Finition
The final step of finition must be realized to
eliminate the burrs which if formed during the laser
machining. In fact, the particles of metal are
pulverized and aspired by an extraction system but
some one cools at the edge of the surface. So a fine
final grinding is necessary.
2.2 Adjustments
Different adjustments are possible during the CAM
about the driving of the laser beam or before the
laser machining with the intrinsic laser parameters.
2.2.a CAM parameters
The laser process to machine a surface is composed
of the border cutting and the hatching area. For
instance, the following displacement between two
border cutting and track displacement between two
laser hatching can be adjusted. The influence of
these parameters has been studied by the laser
machining of a series of dimples (Ø50 µm and depth
= 50 µm).
2.2.b Intrinsic laser parameters
Four parameters can be adjusted:
• The intensity of the current of the lamp: I
• The opening time of the quality switch
• The closing frequency of the quality switch: r
• The speed ablation: V
The three last parameters can adjust the recovery
of two impacts of the laser beam [7] by the
equation (1):
⎛ V ⎞
U D = ⎜1 − ⎟100% (1)
⎝ r×d ⎠
where UD = rate of overlap, V = ablation speed,
r = frequency of repetition and d = diameter of laser
beam.
The first parameter, the intensity of the current of
the lamp, is very important to machine not only with
a good quality but with the right dimensions for the
depth too. During the CAM, the thickness is
determined, typically between 2 and 5 µm. If the
programming is realized with a final depth of
100 µm with a thickness of 5 µm, the number of
machining passes will be of 20. If the parameter I is
not adjusted to 5 but 4 µm, the number of machining
passes does not change and the final depth will be of
80 µm.
2.3 Laser machining
The laser machined final textures consist on grooves
and dimples in a heterogeneous material: lamellar
cast iron. The laser surface texturing is a well known
application to reduce lubricated friction. So a
material with intrinsic good friction properties has
been selected.
The laser surface texturing is commonly composed
of dimples. Here, it is principally composed of
grooves with different cross section: rectangular,
semi circular, trapezoid and triangular. The ratio
between the depth and the width is the same for each
groove and equal to 10.
The galvanometric scanner has been used sometimes
and a comparison has been made.
3 CONTROL OF THE LASER SURFACE
TEXTURING
3.1 Topographic microscopy
This optical technique is based upon an optical
microscope coupled with a digital camera. This
apparatus uses a contrast color sensor The whole
heights extend of the measured area is divided in a
set of focus planes. The resulting set of images gives
rise by a software reconstruction to a 3D map of
focus points leading to an image of the topography
of the area under control. The selection of the
focussed pixels is made by a neighbouring
comparison of contrast. Joined area can also be
measured leading to a greater accuracy in case of
large fields. The resolution in heights rises up to 20
nm with an objective x100.
3.2 3D Acquisitions data
The acquisitions of the different laser surface
texturing are optical (Fig 1) and topographic images
(Fig 2) in 3D and the associated profiles (Fig 3).
 Fig. 1. Optical image of a trapezoid groove.

Fig. 4. Optical image of dimples with the influence of the

Fig. 2. Topographic image of a trapezoid groove.
parameter of the number of border cutting.
The galvanometric scanner allows the stairs on the
flanks of the grooves to be avoided (Fig 5). In
additions, to improve the quality of the laser
machining, a laser polishing can be operated. This
technique is explained in the next paragraph.

20

0
0 0.2 0.4 0.6 0.8 1 1.2
-20
depth (µm)

-40

-60

Fig. 3. Associated profile of a trapezoid groove.
4 DISCUSSIONS
The adjustment of the intensity of the current of the
lamp is very difficult. Only a mean depth can be
realized which is approximately the desired one
(about 2 µm). Concerning the lateral dimensions, the
precision is quite good, about 5%.
The CAM parameters have an influence about the
respect of the geometry of the textured surfaces. The
desired dimples can be for example changed in
polygons (Fig 4).
-80
perpendicular strategy
-100
tilted strategy
-120
length (mm)
Fig. 5. Comparison between the two machining strategies for a
triangular groove.
5 LASER POLISHING
5.1 Principle
To machine correctly with a laser, the surface of the
sample and the focus of the laser beam must be the
same. If an offset is integrated between the two
plans, the energy will be less important. So instead
of ablating the material, it will be only melted.
For lamellar cast iron, the laser machined surfaces
exhibit scoria and pockets coming probably from the
higher rate of ablation of the graphite in the cast iron
(Fig 6). These pockets can be fill up by the laser
polishing and consequently, giving a better
machining quality (Fig 7).
 Table1. Measure of Ra for different offset
Offset (mm) Ra (µm)
-3 2.95
-2 2.75
-1 2.25
0 2.98
+1 2.46
+2 2.72
+3 3.62
10 µm

Fig. 6. SEM image of the bottom of a groove without laser 6 CONCLUSIONS

polishing.
In order to achieve the laser surface texturing of a
heterogeneous material like lamellar cast iron some
difficulties have to be overcome. Firstly, there is a
need to adjust the energy parameters of the laser
which are conditioning the groove depth, the quality
and the accuracy of the groove profiles. For
example, concerning the cross section geometry of
10 µm the grooves, the use of the galvanometric scanner for
the deepest grooves is recommended. The machining
Fig. 7. SEM image of the bottom of a groove with a laser time is much longer but the quality is much higher
polishing (offset = -1 mm). by eliminating the stair effect. To optimize the
aspect of the machining, a laser polishing can be
5.2 Results operated.

The range of the offset is -3 mm up to 3 mm by step

REFERENCES
of 1 mm. To control the roughness in the bottom of a
rectangular groove, a roughness profilometer has
been used. The criteria Ra has been selected to
compare the laser polishing (Table1). This criteria
has been measured for rectangular grooves without
laser polishing and it was about 2.74 µm.
1. C. Vincent, G. Monteil, T. Barrière and J.C. Gelin, Laser
surface texturing of lamellar cast iron, In: Proc.
ESAFORM 2006, edited by N. Juster and A.
Rosochowski, Glasgow (2006), 567-570.
2. T. Dobrev et al., Laser polishing, In: Proc. 4M 2006,
edited by W. Menz et al., Grenoble (2006) 273-276.
3. X. Wang and al., Tribology International vol 34, 703-
711 (2001).
4. A. Erdemir., Tribology International vol 38, 249-256
(2005).
5. U. Pettersson, "Surfaces designed for high and low
friction", Ph.D. Thesis, Uppsala University, 2005.
6. . A. Kovalchenko and al., Tribology International vol
38, 219-225 (2005)
7. A. Kaldos and al., Journal of Materials Processing
Technology 155-156, 1815-1820 (2004).