Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com
Procedia
Engineering
ProcediaProcedia
Engineering 00 (2012)
Engineering 000–000
19 (2011) 125 – 131
www.elsevier.com/locate/procedia

1st CIRP Conference on Surface Integrity (CSI)

3D heat transfer analysis for a hybrid approach

to predict residual stresses after ball-end milling
N. Guillemota,b*, M. Winterb, A. Souto-Lebela,b,
C. Lartiguea, R. Billardonb
a
LURPA, ENS Cachan, EA 1385, Université Paris-sud 11, 61 Av du Pdt Wilson, 94230 Cachan, France
b
LMT-Cachan, ENS Cachan, CNRS (UMR 8535), UPMC-Paris 6, 61 Av du Pdt Wilson, 94230 Cachan, France

Abstract

A hybrid approach is developed to predict the residual stresses field induced by ball-end milling. This hybrid
approach consists, first in the determination of a thermal mechanical loading equivalent to the metal cutting process
and, second, to a 3D Finite Element analysis simulating the effect of this thermal mechanical loading on to the final
workpiece. This work is focused on a 3D heat transfer analysis developed to predict the thermal loading along the
flutes of the tool, considering elementary cutting forces derived from measurements made during orthogonal cutting
tests and milling tests. The predicted temperature fields are compared with Infra-Red camera and thermocouples
measurements. Residual stresses profiles derived from the complete hybrid approach are favourably compared with
the results of experimental measurements made on flat specimens made of a 25MnCrSiVB6 bainitic steel.

© 2012 Published by Elsevier Ltd. Selection and peer-review under responsibility of Prof. E. Brinksmeier

Keywords: Ball-end tool, Oblique cutting, FEM simulation, Temperature

1. Introduction

The prediction of the surface quality of safety parts is a permanent concern in aeronautics, automobile
industry, mechanical industry, ... Surface quality of a workpiece may refer to various properties including
mechanical resistance to fatigue loading. Indeed, the endurance limit of a metallic component may be
highly affected by its surface integrity as induced by the machining process and the values of the cutting
parameters used to machine it. In this context, surface integrity refers to microgeometry, residual stresses
and hardening. Hence, the development of High Speed Machining fosters researches the aim of which is
to predict surface integrity with the final objective of taking account of this surface integrity when
assessing the fatigue life of machined mechanical parts. Residual stresses play a key role on the
endurance limit of high strength steel such as the bainitic steels considered herein [1].

* Corresponding author. Tel.: +33 1 47 40 29 96;

E-mail address: nicolas.guillemot@lurpa.ens-cachan.fr

1877-7058 © 2011 Published by Elsevier Ltd.

doi:10.1016/j.proeng.2011.11.090
126 N. Guillemot
N. Guillemot et Procedia
et al./ al. / Procedia Engineering
Engineering 19 (2011)
00 (2012) 125 – 131
000–000 2

Three main classes of approaches are used to predict the residual stresses induced by the thermo-
mechanical loading associated to the machining process [2]. Closed form solutions – using Oxley
partition energy model [3] – have been proposed while restricting to the heat flux in the Primary Shear
Zone (ZCP) the loading applied to the surface [4]. Computation time is then very short (say a few
seconds), but the loading considered is poorly representative of the real cutting phenomenon. Various
forms of the finite element method can be used to simulate numerically the behaviour of both the
workpiece and the tool during the complete material removal process [5]. These numerical analyses
intend to be the most representative of the real phenomena. However, – apart from the fact that their
computer cost is hugetheir accuracy relies on sophisticated modelling of so many physical phenomena
that their application is presently essentially restricted to orthogonal cutting. The hybrid approach tends to
be a compromise between accuracy of the predictions, complexity of the models implemented and cost of
the experiments and numerical simulations. The last step of this approach consists in a finite element
analysis simulating the effect of a thermal mechanical loading applied directly on to the final machined
surface without modelling the chip removal per se [6-8]. The first steps of this approach consist in the
evaluation – from experimental measurements – of the thermal mechanical loading representative of the
cutting process.
Successive paths =
N successive loadings

Z N
n N p’ q’
N Z Z
ae
Y
ae Y Initial loadings
Vf
βf
dZ ae
at the interface
tool-chip-part
ae ae
Y
hc θ4 Y
ap
Z 2DEZ
X Resulting loadings
Y X
X
applied on the final
2DEZ workpiece
Final workpiece
Fig. 1. Principle of a hybrid modelling for milling process [8]

To apply this hybrid approach – developed for turning by Valiorgue et al. [6] – to ball-end milling
requires specific developments (see Fig.1). In particular, the evaluation of the thermal mechanical loading
representative of the three-dimensional, oblique and discontinuous metal cutting operation associated to
ball-end milling, is a challenging problem. A first approach to tackle this problem consisted in making an
inverse identification of the heat flux from Infra-Red camera measurements [1,8]. Considering the limited
reliability of these measurements, the heat flux was assumed as being uniform in the cutting area, viz. all
along the cutting edge and during the rotation of the tool. The aim of this paper is to improve this step of
the hybrid approach by deriving a non-uniform heat flux from cutting forces measurements. This hybrid
approach is developed in the context of ball-end finishing milling of a 25CrMoSiVB6 bainitic steel.

2. Hybrid approach to model ball-end milling

The main steps of the approach presented herein are the following. First, the cutting edge is discretized
in Elementary Cutting Tools (ECT), cutting forces on each ECT are derived from milling tests,
orthogonal cutting tests, and oblique cutting modelling [9]. Then, different hypotheses are made and a
heat transfer analysis is carried out to derive from the loading at the interfaces chip-tool-workpiece the
thermal mechanical loading on to the final surface. Finally, the effect of successive tool paths is simulated
by applying the equivalent thermal mechanical loadings on to the final surface.

2.1. Machining conditions and global forces measurement

The cutting parameters, viz. axial depth of cut, ap, cutting speed, Vc, feed per tooth, fz, cusp height, hc
(controlled through radial depth of cut, ae) and lead angle of the tool, βf are defined on Fig. 1. The value
of the axial depth of cut, ap = 0.5 mm, considered herein is representative of finishing milling conditions.
 N.N.
Guillemot et et
Guillemot al.al./
/ Procedia
ProcediaEngineering
Engineering1900(2011)
(2012)125 – 131
000–000 127
3

The values of others parameters, viz. Vc = 300 m/min, fz = 0.2mm/tooth, hc = 5 μm and βf = -3°, were
selected because they lead large compressive residual stresses beneath the machined surface [1]. Using
these values of the cutting parameters with a tool, 10 mm in diameter with a helix angle of 30°, cutting
tests were performed on a 5-axis MIKRON UCP710 milling centre. The results of cutting forces
measurements carried out with a triaxial KISTLER 9443 force dynamometer are plotted on Fig. 2(a).
The entry of a tooth into the material corresponds to angle θ1 whereas the extremum values of the
cutting forces correspond to angle θ2 (Fig. 2 (a)). These two angles –which appear as being different by
approximately 180°- define a plane –close to the YZ plane- referred to as the 2D-Elementary Zone
(2DEZ) in the sequel. This plane corresponds to the plane where the uncut chip is the thicker. In the
following, it is assumed that the maximum value of the chip thickness corresponds to the maximum value
of the strains, and hence, to the maximum value of the residual stresses induced by the cutting process.
180°
200
θ1 θ2 Fx

100
Fy
Rake Z
Fz
(N)

face
coupe(N)

0   s
deforces

0 90 180 270 360 450 Chip

-100
dFa Element
of the
dFc
Cutting

cutting
s s
Efforts

-200
dFt
edge
f
-300 dFr Relief 
Entrée
Entry of de
thela1st tooth Entrée
Entry of thede la
second db face

-400
1ère dent
in the material tooth2ème
in thedent
material tn Primary
shear zone
Rotation
Angleangle of the tool
de rotation (°) (°)
Fig. 2. (a) Measurement of the cutting force components [8] ; (b) Parameters and force components definition in oblique cutting ;
(c) Orientation of the Elementary Cutting Tool vs. lead angle

2.2. Evaluation of the mechanical loading

Computation of the elementary forces on the cutting edge elements
According to Armarego [9] and Lee and Altintas [10], discretization of the cutting edge of the ball-end
tool and modelling of oblique cutting make it possible to define elementary cutting forces on each ECT
along the flute. The radial, tangential and axial components of these elementary forces – defined in a
spherical frame- are respectively denoted by dFr, dFt and dFa (see Fig.2 (b)). Assuming that the chip flow
angle is small – when the depth of cut, ap, is small – and such that η ≈ λs [11], the expressions of these
components are the following
 s sin(  n   n )
 K rc  sin( ) cos( ) c
 dFr  K rc  f z sin   dZ  K re .db  n s

 
  s cos( n   n )  tan 2  s  sin  n
dF  K  f sin   dZ  K .db with K  (1)
 t tc z te  tc
dF  K  f sin   dZ  K .db  sin( n ) c
   s cos( n   n )  sin(  n ) tan  s
a ac z ae

 K ac 
 sin( n ) c
and with c  cos 2 ( n   n   n )  tan 2  s  sin 2  n
where θ denotes the spindle angle, db the width of the chip and dZ the height of the ETC. The sets of
parameters (Krc, Ktc, Kac) and (Kre, Kte, Kae) respectively denote the sets of cutting force coefficients and
edge force coefficients. Cutting force coefficients depend on normal rake angles γn, shear angle Φn,
normal friction angle βn, as well as shear stress τs, and acute angle of the cutting edge λs.
Acute angle, λs. is a function of the geometry of the ball-end tool and its inclination [12] so that
128 N. Guillemot
N. Guillemot et Procedia
et al./ al. / Procedia Engineering
Engineering 19 (2011)
00 (2012) 125 – 131
000–000 4

 s ( Z )  tan 1 ( R ( Z ) / R0 ) tan  s 0 sin    f  (2)

where R0, R(Z) and λs0 respectively denote the tool radius, the effective radius and the helix angle, and
where angle χ describes the orientation of the ECT with Z-axis [8], see Fig. 2 (c).
Shear angle and edge force coefficients calculation
Shear angle, Φn, was derived from orthogonal cutting, i.e. turning, tests, carried out for different values
of effective speed Veff (Z)=R(Z)/R0 Vc, but for a constant value of the feed equal to 0.14 mm, which
corresponds to the angle denoted by θ=θ2-θ1 on Fig. 2 (a) such that the cutting forces are maximum.
Shear stress, τs, and normal friction angle, βn, were also derived from the measurement of forces Ft and Fr
during these tests such that
Ft  Fr cos n   n   n sin  n
2 2

 s (Z )  with  n  Fr / Ft (3)
w f
where w and f respectively denote the width of cut and the feed. The results are reported in Tab. 1.
Table 1. Results of orthogonal cutting data for f = 0.14 mm/tr

R(Z) (mm) Veff (m/min) Φn (°) τs (MPa) βn (°)

Close to the end of the tool 0.09 6 23 501 36
Towards the cylindrical area 1.91 115 28,5 442 38
After transformation from the spherical frame attached to the tool to the Cartesian coordinate system
attached to the workpiece (see Fig. 1), the result of the summation of all elementary cutting forces, (dFr,
dFt, dFa), can be compared to the maximum values of the cutting forces components, (Fx, Fy, Fz),
measured during the milling test (Fig. 2(a)). The values of the edge force coefficients identified via this
procedure were Kre= 79 N/mm and Kte = 95 N/mm, Kae being negligible – as already pointed out in [10].

Distribution of mechanical loads in the shearing zones

Assuming that the friction coefficient on both rake and relief faces verifies 0.2 < μSSZ = μTSZ < 0.5 [7],
once elementary forces, (dFr, dFt, dFa), on every ECT are determined, their partition between the TSZ
and the SSZ is made according to the force decomposition illustrated by Fig. 3(a). Following this
procedure [8], elementary cutting forces in the Tertiary Shear Zone (TSZ) take the following expression
dF  cos  n  sin  n   dFr  SSZ sin  n  cos  n 
dFr TSZ  c SSZ (4)
 TSZ  SSZ cos  n  sin  n    SSZ sin  n  cos  n
dFt TSZ  dFr TSZ  TSZ  cos  s and dFa TSZ  dFr TSZ  TSZ  sin  s

whereas elementary cutting forces in the Secondary Shear Zone (SSZ) take the following expression
 dFc  SSZ  dFr  SSZ  TSZ
dFr SSZ  (5)
 TSZ  SSZ cos  n  sin  n    SSZ sin  n  cos  n
dFt SSZ  dFr SSZ /  SSZ  cos  s and dFc TSZ  dFr SSZ /  SSZ  sin  s

Elementary forces in the PSZ, decomposed as illustrated in Fig. 3(a), take the following expression
dFr PSZ  dFc SSZ sin( n   n )  dFr SSZ cos( n   n ) ; dFc PSZ  dFc SSZ cos( n   n )  dFr SSZ sin( n   n )
(6)
dFt PSZ  dFc PSZ . cos  s ; dFa PSZ  dFc PSZ  sin  s

3. Heat transfer analysis

The SSZ being located at the interface between the chip and the tool, the workpiece can be affected
only by the thermo-mechanical loadings in the PSZ and the TSZ. Recent studies [7,13] show that the
partition of the heat flow in the TSZ towards the tool or the workpiece depends on the friction coefficient.
However, as a first approximation, it is assumed that most of this heat is evacuated by the tool [6].
 N.N.
Guillemot et et
Guillemot al.al./
/ Procedia
ProcediaEngineering
Engineering1900(2011)
(2012)125 – 131
000–000 129
5

Considering that the efforts in the PSZ mainly affect the uncut chip which is then removed, their effect on
the workpiece is neglected [2,6]. Furthermore, it is assumed that mechanical power in the PSZ is entirely
transformed into heat, so that the heat source associated to each ECT can be computed as
Veff cos  n  Veff (7)
Pth PSZ  Pm PSZ  Vn sin n  dFr PSZ  V p  dFc PSZ with V p  and Vn 
cos s cos n   n  coss 
where Vn and Vp denote the cutting speed components in the PSZ, see Fig. 3 (a), to be associated to
forces dFr PSZ and dFc PSZ. It is recalled that component Vp, by definition perpendicular to the cutting edge,
is approximately parallel to effective cutting speed Veff [13].
A‐A

dFc SSZ dFc Tool dFc SSZ

γn dFc PSZ γn Tool
A
dFr
n dFr SSZ dFr PSZ dFr SSZ
dFc TSZ n
dFr TSZ

Vn λ Veff
s
Vchip γn Tool
Vp
Vr γn
A
Vn n

Fig. 3. (a) Forces assessment applied on the tool, the chip and speeds components ; (b) heat sources generated during machining [14]

Assuming that 90% of the heat in the PSZ is evacuated in the chip [14-15], the heat flux density
towards the workpiece takes the following form
 ZCP part (9)
 PSZ chip  0.9 Pth PSZ and  PSZ  workpiece  0.1 Pth PSZ ;  PSZ  workpiece 
t n / sin n  db
The evolution of this heat flux density along the cutting edge is plotted in Fig. 4 (a) for two different
values of the friction coefficient, viz. μ=µSSZ=µTSZ equal 0.2 or 0.5.
250
Thermal flux density (W/mm²)

Feed direction for βf = -3
200

150 Tool

100
fμ==0,2
0,2 ap
fμ==0,5
0,5
50

T°C max = 207°C
0
0 0,1 0,2 0,3 0,4 0,5
Coordinates Z (mm) 2 mm

Fig. 4. (a) Heat flux density along the cutting edge; (b) 3D temperature field simulation for μ=0.2; (c) Infrared camera measurement

3D finite element heat transfer analyses were carried out to predict the temperature field induced in a
workpiece by this localized heat flux. The parallelepipedic part is meshed with 4-node tetrahedral
elements, initial temperature is set to 293K, convective-radiative exchange is assumed between the air
and the upper free face of the part, whereas the latter is large enough to assume that there is no conduction
on its other 5 faces The result of the analysis carried out with μ=µSSZ=µTSZ =0.2 is illustrated by Fig. 4(b).
The maximum value of the temperature predicted on the surface is in good agreement with Infra-Red
camera measurements illustrated by Fig 4(c). Temperatures are evaluated with a CEDIP Jade III type
Infra-Red camera at a 1000 Hz frequency on a 64×128 pixels window and a range of 150-300°C.
130 N. Guillemot
N. Guillemot et Procedia
et al./ al. / Procedia Engineering
Engineering 19 (2011)
00 (2012) 125 – 131
000–000 6

4. Residual stresses profile and distribution simulation

Figure 5 illustrates the predicting capability of the hybrid analysis. Most technical details concerning
the thermo-mechanical finite element analysis and the residual stresses measurements are presented in
[8]. Thermo-mechanical loads tend to create plastic strain in the material. Then, residual stresses are
obtained after stress unloading, i.e. equal to zero. The numerical simulation coupling thermal and
mechanical effects allow to obtain results illustrated by Fig. 5.
The residual stresses profiles predicted beneath the machined surface and along the cusps with the
thermal mechanical loading above-discussed are in good agreement with X-ray diffraction measurements.
200 0
0 0,2 0,4 0,6 0,8 1 1,2
σxx (MPa)

100
-100 Residual stresses distribution

Residual stresses (MPa)

Average residual stresses
0
0 50 100 150 200 250 300 350 400 -200 Modelling
Residual stresses

-100
-300
-200

Simulation -400
-300
Experiment
-500
-400

-500 -600
Depth beneath the machined surface (µm) X-axis coordinate (mm)
Fig. 5. (a) Average analysis in the depth ; (b) Local analysis at 50μm beneath the machined surface

5. Conclusion
A simplified approach based on oblique cutting analysis is proposed to model the heat flux towards the
final workpiece induced by finishing ball-end milling. Thanks to milling cutting forces measurement and
orthogonal cutting tests, this model is able to take into account the local rake angle, the helix angle as well
as the lead angle of the tool. Despite a strong hypothesis on loadings in the primary and tertiary shear
zones, the 3D temperature field calculation is close to experimental evidences. This heat transfer analysis
constitutes the first step of a hybrid approach which makes it possible to predict a residual stresses field
also in good agreement with experimental measurements made on a 25MnCrSiVB6 bainitic steel.

Acknowledgements
The authors are grateful to T. Braham who carried out orthogonal cutting tests, and to ASCOMETAL
CREAS for supplying the material, viz. a 25MnCrSiVB6 steel marketed as METASCO MC®.

References
[1] Guillemot N. Thèse de doctorat de l’ENS Cachan, 2010.
[2] Rech J, Hamdi H, Valette S. In: J. Paulo Davim editor, Chapter 3, Springer, 2008.
[3] Oxley PLB. Mechanics of Machining: an analytical approach to assess machinability, Chichester: Ellis Horwood Ltd., 1989.
[4] D Ulutan, B Erdem Alaca, I Lazoglu. Mater. Process. Tech. 2007;183:77-87.
[5] Outeiro JC, Umbrello D, M’Saoubi R. Inter. J. Mach. Tools Manufact. 2006;46:1786-1794.
[6] Valiorgue F, Rech J, Hamdi H, Gilles P, Bergheau JM. J. Mater. Process. Tech. 2007;191:270-273.
[7] Mondelin A, Valiorgue F, Rech J, Coret M, Feulvarch E. Adv. Mat. Res. 2011;223:411-420.
[8] Guillemot N, Beaubier B, Braham T, Lartigue C, Billardon R. Adv. Mat. Res. 2011;223:391-400.
[9] Armarego EJA. Whitfield AC. Ann. CIRP 1985;34(2):65-69.
[10] Lee P, Altintas Y. Inter. J. Mach. Tools Manufact. 1996;9:1059-1072.
[11] Stabler GV. Advances in Machine Tool Design and Research, 1st ed. New York: Macmillan, 1964.
[12] Fontaine M, Moufki A, Devillez A, Dudzinski D. J. Mater. Process. Tech. 2007 ;189:73-84.
[13] Bonnet C, Poulachon G, Rech J, Girard Y. 12th CIRP Conference on Modeling of Machining Operations (2010), p.237-242.
 N.
N.Guillemot
Guillemotetetal.
al.// Procedia
Procedia Engineering
Engineering 19
00 (2011)
(2012) 125 – 131
000–000 131
7

[14] Battaglia JL, Elmoussami H, Puigsegur L. C.R. Mécanique 2002;303:857-864.

[15] Schmidt AO, Roubik RJ. Distribution of heat generated in drilling, ASME 1949; 71:245-252.