See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/239773788

Cutting Forces Evaluation in Milling using an Analytical /Finite Element

Mixed Approach

Article · September 2008

CITATION READS

1 58

4 authors, including:

Edouard Riviere Enrico Filippi

Université de Mons Université de Mons
79 PUBLICATIONS   497 CITATIONS    65 PUBLICATIONS   476 CITATIONS   

SEE PROFILE SEE PROFILE

Laurent D'Alvise
NOMASYS Consulting
25 PUBLICATIONS   192 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fab-IoT-Lab View project

Robotic machining View project

All content following this page was uploaded by Edouard Riviere on 06 January 2014.

The user has requested enhancement of the downloaded file.

 Cutting Forces Evaluation in Milling using an Analytical /Finite Element
Mixed Approach
E. Rivière-Lorphèvre1*, J. Barboza2, E. Filippi1, L. D’Alvise2
1
Faculté Polytechnique de Mons, Service de Génie Mécanique, rue du Joncquois 53, B-7000 Mons
(Belgium)
2
CENAERO, Virtual Manufacturing Group, Bâtiment Eole 1er étage Rue des Frères Wright 29 B-6041
Gosselies (Belgium)
*
[email protected]

Abstract
Simulation of the machining process is a widespread technique to predict the optimal
parameters for machining operations. The simulation of the cutting forces can be
achieved by using two different techniques. The first is the use of mechanistic models
based on linear analytical expression of the cutting forces and is the most common one.
These models are able to compute the efforts acting on complex tools but need experi-
mental tests to determine the specific parameters.
The second is based on the finite element method. Finite element computations of the
machining are able to predict cutting forces, chip formation, heat generation from the
knowledge of the properties of the materials.

The aim of this paper is to present a method to link cutting forces computed by finite
element method with a local geometry to the global efforts on a complex milling tool.The
finite element simulation is performed with the finite element software Morfeo (Manufac-
turing ORiented Finite Element tOol). Morfeo is Cenaero's in-house code devoted to re-
spond to the industry specific needs. For this paper, an updated Lagrangian description
with an automatic and adaptive remeshing was used. Simulation results include predic-
tion of cutting forces, which require an accurate computation of the complex phenomena
on the contact interface. At the contact interface nonmatching meshes are considered
and a frictional contact inducing heat generation is used. In addition, the heat transfer
coefficient depends on the contact pressure.
The efforts computed by Morfeo can then be used as inputs for the dynamic simulation
of a complex milling process including self excited vibrations.

1 INTRODUCTION literature [5]: analytical models and models

1.1 Aims of the simulation of the machining based on numerical simulation (e.g. finite ele-
processes ment methods). Both models have their own
advantages and drawbacks.
The optimization of production techniques is a
Analytical methods compute the elementary
challenging issue for many manufacturing
effort with simple functions of macroscopic va-
companies. The specific drawbacks of machin-
riables (chip thickness axial and radial, depth of
ing processes (material consumption, low
cut for example). The simplest of those models
productivity,...) put a particular pressure on
is the linear one where the cutting force is
them to remain profitable in comparison with
simply the product of the undeformed chip sec-
other production techniques.
tion by a constant parameter (often called spe-
Process optimization by mean of simulation is
cific pressure).
thus frequently used for various aims: reduction
The analytical models are very simple to use
of the cycle time ([1], [2]), prediction and sup-
and doesn’t require intensive computation but
pression of chatter vibrations ([3],[4]), minimiza-
the coefficients of the models have no clear
tion of tool wear, improvement of part quality,...
correlation with intrinsic properties of the mate-
All these simulation methods have in common
rials (Young’s modulus, hardness or yield
the need of a model to compute the cutting
strength for example) and must be identified
forces produced during the machining process.
from machining test with the same cutting
1.2 Cutting forces models tool/workpiece material couple[6]. In addition,
In the field of cutting forces simulation, two these models only give global information and
main types of models are frequently used in the

Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)
the user has no access to local values of field rimental measurements to obtain the parame-
variable. ters of the cutting force model.
The use of a numerical method to model at a 2 FINITE ELEMENT SIMULATION WITH
microscopic level is a fairly active topic in the MORFEO
scientific community. This type of simulation is In order to fit to the industry specific needs, the
able to model at a detailed level many field va- virtual manufacturing group of Cenaero devel-
riables (stress, strain or temperature e.g.) to get ops an in-house Finite Element software, Mor-
a full analysis of the situation. The complexity feo (Manufacturing Oriented Finite Element
of the modeled phenomena (high strain, high tOol).. The main objective is to develop a nu-
strain rates, friction, chip separation,...) often merical tool able to simulate the welding and
limit the simulation to simple orthogonal cutting machining processes for large size components
tests [7]. The complex geometry of the milling and within a reasonable computation time.
operations would lead to oversized models with To achieve the accurate and fast simulation of
a computing time often prohibitive for practical the manufacturing processes, the following fea-
use in the shop floor [8]. tures were implemented in Morfeo:
1.3 Analytical/Finite element mixed ap- - Displacement formulation of discretiza-
proach tion for the FE structural analysis.
In order to take advantage of both methods, we - Velocity/pressure formulation of discre-
propose to adopt a mixed approach analyti- tization for the FE flow analysis.
cal/finite element to simulate the cutting forces - Staggered thermo-mechanical coupling.
for practical milling situation. This method is - Eulerian thermo-fluid coupling.
divided in four steps (see Figure 1): - Thermo-mechanical-fluid coupling using
- The cutting forces for a given pair of the MpCCI (Mesh-based parallel Code
materials (tool and workpiece) are com- Coupling Interface) protocol: local/global
puted by a finite element method in or- approach.
thogonal cutting. - Updated Lagrangian analysis for the
- The results of these simulations are description of the continuous medium.
used to extract the parameters of the or- - Elasto-plasticity (displacement formula-
thogonal cutting force model tion) and viscoplasticity (fluid formula-
- A classical orthogonal to oblique trans- tion).
formation is performed to obtain the cut- - Non linear thermal analysis: thermo-
ting forces parameters for the geometry physical properties function of tempera-
to be simulated. ture.
- Finally those parameters are used to - Multibodies contact treatment: Mas-
simulate the cutting forces for a com- ter/Slave detection algorithm and non
plex milling operation. interpenetration procedure managed by
the penalty method. Adaptive remesh-
ing 2D and 3D.
- Connection with a new generation opti-
mization code (Minamo, Cenaero) for
inverse analysis and process optimiza-
tion.
Morfeo main originality lies in its capability to
handle both structural and fluid flow problems
in the same FE analysis. Moreover, a lo-
cal/global approach is implemented: this means
that at the local scale, corresponding to a re-
stricted zone of the whole part, a multi-physics
simulation is computed with a refined mesh.
Figure 1: diagram of the analytical/FEM mixed The results of this analysis are then transferred
approach to the global scale computation, corresponding
This method allows the simulation of the cutting to the full-size part. This ensures accuracy, ra-
forces on a variety of cases for which a com- pidity and applicability of Morfeo in industrial
plete finite element would be difficult or too time environments.
consuming and does not need to perform expe-

Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)
3 SIMULATED TESTCASE αr: rake angle=12°
The cutting forces produced during milling of a δ: clearance angle = 5°
titanium part will be used as an example along h: chip thickness = 0,05;0,15 and 0,25 mm
this paper. The data of both models (analytical V: cutting speed = 30m/min
and finite element) are summarized in the fol- The machining of a rectangular block with a
lowing paragraphs. triangular tool (see Figure 2) is simulated using
3.1 Milling testcase the same technological parameters (cutting
speed, rake angle, etc.) as for the milling test-
The milling testcase is an example extracted case. Various chip thicknesses are simulated in
from [9]. Cutting forces during half immersion order to obtain several simulation points for the
downmilling of Ti6Al4V titanium alloy with a cy- extraction of the coefficients.
lindrical endmill are computed and compared 4 FEM SIMULATION OF ORTHOGONAL
with the simulation using a mechanistic model. CUTTING
The cutter has a diameter of 18,1 mm, a lead
angle of 30° and a rake angle of 12°. In FE modeling, there are mainly two types of
The operation is performed at 30 m/min cutting formulation that are used to describe a conti-
speed and with a 0,05 mm/tooth feed. nuous medium: Eulerian and Lagrangian. In a
The mechanistic model used for the simulation Lagrangian formulation, the mesh of the conti-
is one of the most commonly used in the litera- nuous medium deforms with the material (i.e.
ture for which the efforts along cutting, feed and the mesh velocity is equal to the material veloc-
axial directions are computed using six para- ity) whereas in an Eulerian formulation it is
meters: fixed in space (the mesh velocity is equal to
Fc = Fcc + Fce = b ⋅ h ⋅ K cc + b ⋅ K ce (1) zero). A third type of formulation, which is a
combination of the two above types (the mesh
F f = F fc + F fe = b ⋅ h ⋅ K fc + b ⋅ K fe (2) velocity is arbitrary), may be used: Arbitrary
Fa = Fac + Fae = b ⋅ h ⋅ K ac + b ⋅ K ae (3) Lagrangian Eulerian (ALE). In the technical lite-
where b is the width of cut, h is the chip thick- rature of the FE modeling of orthogonal cutting,
ness, K.c the coefficients linked to the shearing these three formulations are used. In major
of the chip and K.e are the coefficients linked to models a Lagrangian formulation type is used.
the edge forces. The values of those coeffi- This formulation requires a criterion for separa-
cients are given in Table 1. tion of the undeformed chip form the workpiece.
Coefficient Value Along this paper we use an Updated Lagran-
gian formulation, which is to take as a refer-
Kcc 1478 MPa
ence configuration at time t, the previous confi-
Kfc 247 MPa guration when using a Lagrangian formulation.
Kac 577 MPa In addition to the Updated Lagrangian formula-
Kce 24 N/mm tion, an explicit time integration technique is
Kfe 43 N/mm used with automatic remeshing to simulate chip
Kae 0 N/mm formation and separation [7]. The adaptive re-
Table 1: Coefficient of the cutting forces model meshing allows alleviating mesh element dis-
3.2 Geometry and machining data of the tortions.
orthogonal cut test In following sections we describe the equations
The orthogonal cutting simulation is performed used to model the material flow and the contact
to get the input values for the orthogonal to ob- interface between the workpiece and the tool.
lique transformation. 4.1 Constitutive models
To model the thermo-mechanical problem, in
Morfeo a staggered strategy is used. The latter
consists of solving the thermal problem (or me-
chanical problem) and the results are used as
input parameters in the mechanical (or thermal)
problem.
Mechanical model
To formulate the mechanical equations govern-
ing the continuous medium, the Newton's law
must be satisfied for each material element, it
Figure 2: Geometry of the orthogonal cutting
implies to take the inertial and static effects into
simulation
Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)
account and the gravity as an external body v = ∑ Vnϕ n
force. For the testcase of this paper, the gravity n
(9)
forces have a negligible impact, thus only static p = ∑ Pnφn
equilibrium is considered. By considering the n
density as independent of time, the equilibrium In order to have a consistent approximation, the
can be reduced to: shape functions ϕ n and φn must verify compa-
divσ = 0
(4) tibility conditions. For this, the mini-element is
div(v ) = 0 used. The velocity is decomposed additively in
The second equation of 4 expresses the in- linear term v l and an additional “bubble” term
compressibility of the medium.
v b , the pressure is simply discretized linearly:
In addition to equations 4, an additive decom-
position of the stress tensor is performed to ob- v = v l + v b = ∑ Vnlϕ nl + ∑ Velϕ el
tain a hydrostatic pressure p part (which does n e
(10)
not depend on the plastic deformation) and the p = ∑ Pnφn
deviatoric part s : n

σ 11 + σ 22 + σ 33 The detailed expression of the “bubble” term

σ = s − pI , p=− (5) may be found in [10].
3 By using the discretized expression of the ve-
Let Ω be the domain occupied by the work- locity and the pressure solving the problem de-
piece at any time t . Denoting ∂Ω v the boun- fined by the equations 7 and 8 become a nonli-
dary part of Ω where the velocity is imposed, near problem where the unknowns are Vn
∂Ω Τ the constrained boundary part and ∂Ω c and Pn :
the contact interface where sliding and friction R(Vn , Pn ) = 0 (11)
may occur. The equilibrium of the domain Ω is
In Morfeo, the nonlinear problems are generally
given by the principle of virtual power:
solved by using a Newton-Raphson like method
∫ σ : ε& dV − ∫ Τ.v dS
∗ ∗
Ω ∂Ω v
based on the generalized mimimal residual
(6) (GMRES), with an ILUT preconditioner to solve
− ∫ Τ.v dS − ∫ Τ.v dS = 0
∗ ∗
the related linear systems.
( )
∂Ω Τ ∂Ω c
At each time t a nodal solution v t , p t of equ-
for any variation v ∗ of v and ε& * is the strain
ations 7 and 8 are computed, and the Lagran-
rate of the variation. With the definition of the
gian configuration is updated by following:
deviatoric stress tensor s in equation 5, the
equation 6 can be transformed into: x t + ∆t = x t + ∆tv t (12)

∫ s : ε& dV − ∫ p.div(v )dV − ∫ Τ.v dS

t
∗ * ∗ where x is a nodal position of a vertex of the
Ω Ω ∂Ω v
(7) mesh of Ω .
− ∫ Τ.v dS − ∫ Τ.v dS = 0
∗ ∗
Thermal model
∂Ω Τ ∂Ω c
For the test-case of this paper, we focus on the
In analogy, the incompressibility in equation 4
heat diffusion equation with temperature de-
is imposed in the weak sense as following:
pendent parameters for transient simulations.
− p * .div (v )dV = 0
∫ (8) We assume there is no convective transport
Ω
and phase transformations are taken into ac-
for any variation p ∗ of p . The velocity and count through a variable heat capacity. The
pressure become the unknowns of the problem. heat equation governing the diffusion is ex-
In this testcase, by neglecting the atmospheric pressed by:
pressure ∂Ω Τ is considered as a free surface ∂ρΗ
− div(k∇T ) = Q(T ) (13)
that implies Τ = 0 on ∂ΩT . ∂t
The velocity v and pressure p can be discre- where T is the temperature, ρ is the density
tized by using the standard finite-element ap- and Η is the specific enthalpy, and k is the
proximation in terms of the nodal velocity vec-
material heat conductivity. The last term Q is
tors Vn and nodal pressures Pn :
the source of heat generation. If there is no
phase change and assume a constant pres-
sure, the specific enthalpy (notion of energy per
unit volume) is given by:
Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)
 By using a standard FEM the approximation is
Η (T ) = ∫ c p dT
T
(14)
0 written in terms of nodal values Tm :
where c p is the heat capacity.
T = ∑ Tmψ m (21)
The boundary conditions equations can be one m
of the following: substitution of equations 19 and 21 into the
- a prescribed temperature on a part of weak form of the heat conduction equation 17
the boundary; results in:
[Μ ] (ρΗ )θ − (ρΗ )n + [Κ (Tθ )]Tθ
- an imposed heat flux;
- a convection boundary condition; = b(Tθ ) (22)
∆t
- a radiation boundary condition.
or
The convection and radiation boundary condi-
tions introduce nonlinearity. The thermal ex- [C (Tθ )] Tθ − Tn + [Κ (Tθ )]Tθ = b(Tθ ) (23)
change between the workpiece and the tool is ∆t
described in section 4.2. where:
To these boundary conditions, a heat source - [Μ ] = ∫ ψψ i j dΩ is the mass matrix
Ω
induced by the plastic deformation of material is
added: which is independent of the tempera-
Q = α TK (σ : ε& ) (15) ture,

where ε& is the plastic strain rate, α TK is the

- [Κ (T )] = k (T )∇ψ i ∇ψ j dΩ the rigidity
∫Ω

Taylor-Quinney coefficient corresponding to the or diffusion matrix and can also depend
plastic fraction of the work converted into heat. of temperature through the thermal
The weak form of these equations is obtained conductivity k (T ) ,
by multiplying by a test function T * and inte- - b(T ) = ∫ hψ iT0 d∂Ω − ∫ hψ iψ j d∂Ω is
grating over the volume Ω : ∂Ω ∂Ω

∂ρΗ the term corresponding to the boundary

∫ T* dΩ − ∫ T *div(k∇T )dΩ conditions and may also depend of the
Ω ∂t Ω
(16) temperature (i.e. in case of radiation),
= ∫ T QdΩ
Ω
*
- [C (T )] = ρc pψ iψ j dΩ is the equivalent
∫
Ω
which, using Green’s formula and the diver- specific heat matrix, and it is dependent
gence theorem, becomes: of the temperature through the latent
∂ρΗ heat and the heat capacity parameter.
∫T
*
dΩ − ∫ ∇T * .(k∇T )dΩ These equations are nonlinear if material prop-
Ω ∂t Ω
(17)
erties are thermally dependant or introduce
= ∫ T QdΩ + ∫ T Qimp d∂Ω
* *
nonlinear boundary conditions.
Ω ∂Ω

∂T For θ = 0 , the temporal scheme is subject to a

where Qimp = k brings together the different stability condition: the length of diffusion asso-
∂n ciated to the time step ∆t must be smaller than
terms of boundary conditions and Q the de- the smallest mesh size. Apart this issue, the
formation power heat source. explicit scheme corresponding to equation 22
For transient computation, a first order implicit poses no particular problem. With the method
time integration scheme is used to give the so- of the equivalent specific heat 23 , the matrix
lution at time n + 1 . The variation rate of the [C (T )] is calculated according to Tn and is
nodal variables is considered constant during a
therefore known. By simply reversing the latter,
time step ∆t . Consider a particular time tθ in temperature variations at the nodes
the time interval t n and t n+1 : ∆T = Tn+1 − Tn can be calculated using:
tθ = (1 − θ )t n + θt n+1 ,0 ≤ θ ≤ 1 (18) ∆T = ∆t ([C ]n ) {− [K ]n Tn + bn }
−1
(24)
The solution T (or ρΗ ) can be expressed as: For 0 < θ < 0.5 the corresponding temporal
Tθ = (1 − θ )Tn + θTn+1 (19) schemes are conditionally stable and require
and the variation rate is given by: the inverse of the matrix [Κ ] at each time step.
∂Tθ 1 For θ = 0.5 (used in this testcase and which
= (Tθ − Tn ) (20) corresponds to the Crank-Nicolson mixed
∂t ∆t

Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)
scheme is stable and offers a higher precision Thermal exchange
than the implicit Euler scheme ( θ = 1 ). At the tool/workpiece contact interface, a con-
4.2 Contact interface treatment duction and a heat dissipation term due to fric-
Mechanical treatment tion are added. In this document we assume
that the heat dissipation is shared between the
To ensure the non-interpenetration between the tool and the workpiece according to:
workpiece and the tool, the master/slave tech-
∂T E
nique is used [11]. In this test-case the tool is −k = h( pc )(T − Ttool ) + (τ .∆vt ) (27)
assumed to be the master and the workpiece to ∂n E + Etool
be slave and the contact interface lies on the where E = ρc p k is the material effusivity,
slave. With this technique the nodes on the
slave contact interface are not allowed to be and h( pc ) = apc + b the heat transfer coefficient
inside the master. At the contact interface ∂Ω c which is assumed to be a linear function of the
we assume that: contact pressure pc .
Τ1 = σ 1n = Τ,Τ2 = σ 2 (− n ) = −Τ1 , Laws behavior and parameters
Τn = (Τ.n )n = σ n n
4.3
(25)
τ = Τ − (Τ.n )n As indicated above, a viscoplastic Northon fric-
∆vt = (v − vtool ) − [(v − vtool ).n]n tion model is used at the contact interface (eq-
uation 28), and the Northon Hoff law is used to
where σ 1 is the stress on the slave, σ 2 the model the Ti6Al4V flow (equation 29).
stress on the tool, n the outward normal of the The Norton friction law is expressed as:
contact interface, τ is the shear stress, v the p −1
τ = −α f .K . ∆ν t ∆ν t (28)
velocity field on the workpiece, vtool the tool ve-
where α f = 0.3 is the viscoplastic friction coef-
locity. The contact pressure is defined
as: pc = −σ n ficient, K = 125 Mpa is the viscoplastic consis-
Written in terms of velocity, the contact condi- tency of the material, p = 0.1 is the sensitivity
tions are expressed in Signorini form: coefficient to the sliding velocity.
σ n ≤ 0, The Norton Hoff law is expressed as:
∆vt .n ≤ 0, (26) β
s = 2 K exp( )( 3ε&eq ) m−1 ε& (29)
(∆vt .n )σ n = 0 T + 273
The equations 26 express the fact that the where ε&eq is the equivalent strain rate,
normal stress σ n must be compressive, inter- m = m1 (T + 273) + m0 ( m1 = 0.0 , m1 = 0.1 ) the
penetratrion is not allowed (second equation), sensitivitycoefficient to the strain rate,
and the last equation describes either there is
noncontact ( σ n = 0 ) or there is contact without
β = 200 K the thermal activation coefficient.
−1

The thermal parameters are defined as:

interpenetration ( ∆vt .n = 0 ). parameters tool workpiece
To these contact equations a friction law is ρ 11400 Mpa 7690 Mpa
added. The friction law describes the relation cp 376 J/kg/C° 677 J/kg/C°
that allowed computing the shear stress as
function of several contact parameters as ther- k 35 W/(m C°) 26 W/(m C°)
mo-mechanical variables, normal stress and Table 2: Thermal parameters
roughness of the contact interface. It is very 5 USE OF FEM SIMULATION RESULTS TO
difficult to enter into a single mathematical for- SIMULATE COMPLEX TOOL
mula the influence of all these parameters. In In order to compute the global efforts on a
practice, we try to express the friction law complex tool, the mechanistic parameters iden-
based on the most important variables and de- tification proposed by Budak[12] is performed,
pending on the context. The other variables are and then an orthogonal to oblique transforma-
included in the value of the friction coefficients. tion is used to recover the cutting forces para-
In the testcase Norton friction law is used (equ- meters for the simulation of a general milling
ation 28). To impose the contact conditions 26 example
the penalty method is used as described in 5.1 Analytical law parameters extraction
[11].
Linear regression
Equation 1 and 2 show that if the values of Fc
and Ff are a linear function of chip thickness.
Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)
 cos( β − α )
K cc = τ . (30)
sin(ϕ ). cos(ϕ + β − α )
sin( β − α )
K fc = τ . (31)
sin(ϕ ). cos(ϕ + β − α )

Dividing equation 30 and 31 lead to a value of

the inclination angle of
F fc
β = α + Tan −1 ( ) (32)
Ftc
The shearing angle can be deduced from Mer-
chant minimum energy principle and has a val-
ue of:
Figure 3: linear regression of the efforts π
ϕ= − (β − α ) (33)
The directing coefficient is the K.c coefficient 4
(specific pressure) multiplied by the width of cut This angle can also be deduced from the
and the intercept point is the K.e coefficient measure of the ratio between undeformed and
(edge force coefficient) multiplied by the width deformed chip thickness during the simulation.
of cut (see Figure 3). τ can be deduced from the total force due to
shearing:
Chip thickness Fc Ff
0,05 mm 56,67 N 56,32 N Fc = Ftc2 + F fc2
0,15 mm 133,94 N 80,16 N (34)
0,25 mm 174,02 N 78,65 N and from the area along which shearing acts:
Table 3: Mean cutting forces obtained by Mor- b.h
feo As = (35)
Table 3 shows the values of the cutting forces
sin(ϕ )
obtained by Morfeo. Linear regression is per- The shearing stress is the ratio of those two
formed to obtain the following cutting forces values:
parameters: Fs
τ= (36)
- Kcc = 586,7 MPa As
- Kfc = 111,67 MPa
We finally obtain the following values for those
- Kce = 33,53 N/mm
parameters:
- Kfe = 54,95 N/mm
- β=22,77°
If we compare those values with those of [9]
we can see that we tend to underestimate that - φ=34,22°
component of the effort linked to shearing and - τ=308,18 MPa
to overestimate the component linked to the Those values would be the input data for the
friction. simulation of the milling testcase.
Orthogonal/oblique transformation 5.2 Cutting forces simulation

The orthogonal to oblique transformation allows The values of τ, β, Kte and Kfe obtained by the
the extraction of general parameter to simulate orthogonal cut simulations can be used as input
oblique cutting cases with potentially different parameters for the simulation of the cutting
configurations (rake angle for example). Budak forces on a complex tool.
[12] proposes to decompose the cutting coeffi- The classical simulation algorithm is used. The
cients linked to the shearing of the chip using tool is divided in elementary slices of dz height
the primary shearing angle φ, the inclination of on which the elementary efforts are computed.
the cutting force on the rake face β and the Those elementary efforts are analytically inte-
grated along the cutter to obtain the total effort.
shear stress τ. After mathematical develop-
Figure 4 compares the cutting forces during
ment, equations can be obtained:
one revolution of the tool between the refer-
ence case [9] and the use of the analytical/FEM
mixed approach. We can see that the cutting
forces are slightly underestimated but that the
trend is correct.

Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)
 7 REFERENCES
[1] Tandon, V., El-Mounayri, H., Kishawy,
H., 2002, NC end milling optimization
using evolutionary computation. Int. J.
Mach. Tools & Manuf. 42:595-605.
[2] Uhlmann, E., Mattes, A., 2007, Optimiz-
ing the feed velocity of NC-tool path,
procceding of the 10th CIRP internation-
al workshop on modeling of machining
operation, 245-251.
[3] Altintas, Y., Budak, E., 1995, Analytical
Prediction of Stability Lobes in Milling,
Annals of the CIRP,44:357-362.
[4] Insperger, T., Mann, B.P., Stepan, G.,
Bayly, P.V., 2003, Stability of Up-Milling
Figure 4: comparison of the cutting forces ob-
and Down-Milling, Part 1: Alternative
tained by the mixed method (plain) with refer-
Analytical Methods Int. J. Mach. Tools &
ence
Manuf., 43:25-34.
[5] van Luttervelt, C.A. and al, 1998,
Present Situation and Future Trends in
6 SUMMARY – CONCLUSION Modelling of Machining Operations.
This paper develops a mixed approach analyti- Progress Report of the CIRP Working
cal/finite element simulation to compute the cut- Group 'Modelling of Machining Opera-
ting forces for a complex milling case. This ap- tions', Annals of the CIRP, 47/2:587-
proach benefits from the advantages of both 626,
methods (no experimental tests for FEM, fast [6] Rivière-Lorphèvre, E., Filippi, E., De-
computation for analytical model). The cutting hombreux P., 2007, Inverse method for
forces for an elementary orthogonal cut are cutting forces parameters evaluation.
computed using the finite element simulation Engineering Mechanics, 14/5:345–357.
software Morfeo that has been developed by [7] Sekhon, G.S, and Chenot, J.-L, “Some
Cenaero for industrial analysis. These elemen- Simulation Experiments in Orthogonal
tary cutting forces are then used to identify the Cutting”, Numerical Methods In Indus-
parameters of an analytical cutting forces trial Forging Processes, 901-906, 1992.
model which is used for the simulation of the [8] Filice, L., Micari, F., Rizzuti, S., Umbrel-
efforts on a complex geometry. lo, D., 2007, Dependance of Machining
This paper shows that the global method is Simulation Effectiveness on Material
able to model the cutting forces for complex and Friction Modelling. That's Why In-
geometry of the cutter with a computing time dustrial Application of Machining Simu-
smaller than for a full FEM simulation. It also lation is still so far, Proceedings of the
shows that Morfeo is able to model all the phe- 10th CIRP International Workshop on
nomena linked to machining. Modelling of Machining Operations,
One perspective of this work is the improve- 181-187.
ment of the simulation with Morfeo and further [9] Altintas, Y., 2000, Manufacturing auto-
validations. It will be necessary to implement mation, Metal cutting, Machine tool vi-
more representative material law (such as bration and CNC design. Cambridge
Johnson Cook’s law). The parallelization ca- university press.
pacities of Morfeo can be tested on a more [10] E. Perchat, « MINI-Élément et
complex testcase (in oblique cutting for exam- factorisations incomplètes pour la paral-
ple where 3D time consuming simulation is re- lélisation d’un solveur de Stokes 2D.
quired). Application au forgeage ». Thèse de
Finally, the coupling between Morfeo and the doctorat, École Normale Supérieure des
analytical cutting force model will be imple- Mines de Paris, Sophia Antipolis, juillet
mented to allow simulation of global effects of 2000.
machining in Morfeo (interaction between cut- [11] Barboza J., Fourment L. and
ting tool and structural deformations or compu- Chenot J.-L., Contact algorithm for 3D
tation of residual stresses after machining). multi-bodies problems: application to
forming of multi-material parts and tool
Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)
 deflection. In WCCM V – Fifth Congress
on Computational Mechanics. 2002.
Wien, Austria.
[12] Budak, E., Altintas, Y., Armare-
go, E.J.A.,1996, Prediction of milling
force coefficients from orthogonal cut-
ting data, Transactions of ASME
118:216–224.
[13] Lesuer, D.R., Kay, G.J., LeB-
lanc, M.M., 1999, Modeling Large-
Strain, High-Rate Deformation in Met-
als. Third Biennial Tri-Laboratory Engi-
neering Conference Modeling and Si-
mulation.

Proceeding of the 11th CIRP Conference on modeling of machining operations (16-18 sept 2008)

View publication stats