* Corresponding author.

E-mail addresses: liuch@ecn.purdue.edu (C.R. Liu), yuebin@ecn.purdue.edu (Y.B. Guo)

International Journal of Mechanical Sciences 42 (2000) 1069}1086
Finite element analysis of the e!ect of sequential cuts and
tool}chip friction on residual stresses in a machined layer
C.R. Liu, Y.B. Guo*
School of Industrial Engineering, Purdue University, West Lafayette, IN 47907, USA
Received 2 October 1998; received in revised form 7 April 1999
Abstract
A thermo-elastic}viscoplastic model using explicit "nite element code Abaqus was developed to investi-
gate the e!ect of sequential cuts and tool}chip friction on residual stresses in a machined layer. Chip
formation, cutting forces and temperature were also examined in the sequential cuts. The a!ected layer from
the "rst cut slightly changes the chip thickness, cutting forces, residual strain and temperature of the
machined layer, but signi"cantly a!ects the residual stress distribution produced by the second cut. Residual
stress is sensitive to friction condition of the tool}chip interface. Simulation results o!er an insight into
residual stresses induced in sequential cuts. Based on simulation results, characteristics of residual stress
distribution can be controlled by optimizing the second cut. 2000 Elsevier Science Ltd. All rights
reserved.
Keywords: Residual stress; Surface integrity; Finite element analysis; Metal cutting
1. Introduction
The quality of a mechanical component such as its geometrical stability and fatigue life are
signi"cantly a!ected by the surface integrity of the sublayer generated by the machining process.
Surface integrity generally can be determined in terms of mechanical, metallurgical, chemical, and
topological states. Residual stress is a major part of the mechanical state of a machined layer. It is
generally believed that residual stresses result from plastic deformation, thermal stress, and phase
transformation of the machined layer. Residual stresses in high performance alloys and steels are of
0020-7403/00/$- see front matter 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 0 - 7 4 0 3 ( 9 9 ) 0 0 0 4 2 - 9
Nomenclature
o total elastic stress
o #ow stress
o
"
static yield stress
H strength coe$cient
n strain hardening exponent
DCJ elasticity matrix
cCJ total elastic strain
cNJ e!ective plastic strain
c

NJ e!ective plastic strain rate

cNJ
"
critical e!ective plastic strain
cNJ
D
failure e!ective plastic strain
temperature
c
G
material constants
D material parameter
p material parameter
V temperature increment
f

work-to-heat conversion factor

f
`
conversion e$ciency factor
, density
C
N
speci"c heat
t friction shear stress
j friction coe$cient
F
A
cutting force
F
R
thrust force
: tool rake angle
c coe$cient
i increment number
U displacement vector
At time increment
M diagonal lumped mass matrix
F applied load vector
I internal force vector
S

normal residual stress in cutting direction

considerable industrial importance because they can a!ect failure by fatigue, creep or cracking. As
#exibility and e$ciency are becoming more and more important, super"nish hard turning was
proposed to replace abrasion-based super"nishing (or AB super"nishing) in terms of residual stress
and surface "nish requirements [1]. A stumbling block to industrial application of hard turning
1070 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086
technology can be attributed partly to uncertainty about the residual stress induced in hard
machining [1,2].
The bulk of the existing work on the surface integrity of the machined layer has been limited to
experimental studies. Liu and Barash [3,4] found three quantitative measures to de"ne the
mechanical state of a machined layer. They also found that the e!ect of linear thermal expansion
in the machined layer on residual stress distribution is negligible. In case of #ank wear, the
residual stress pattern is lightly or severely modi"ed by temperature generated depending
upon the cutting speed. Liu and Barash [5] showed that mechanical deformation of the
workpiece surface is the one of the causes of producing both tensile and compressive residual
stresses in machining. Kono et al. [6] presented "ndings that residual stress increases with the
cutting speed, but does not monotonically change with the depth of cut in hard turning. These
results contradict those from machining soft steel [7]. Schreiber and Schlicht [8] con"rmed that
the mechanical properties of the workpiece material to be machined have a great in#uence
on the magnitude and distribution of the residual stresses. Matsumoto et al. [9] and Wu and
Matsumoto [10] concluded that material hardness has a signi"cant e!ect on the pattern of residual
stress that remains in the machined part. Konig et al. [2] indicated that residual tensile stress
is attributable to wear land friction, and the development of residual tensile stress depends
primarily on the extent of wear land in turning 16 MnCr5 case hardened steel. Tonsho! et al. [11]
found that the residual stresses were in#uenced by both the cutting speed and the tool wear because
increasing the cutting speed will accelerate the tool wear. Liu and Mittal [1] proposed single step
super"nishing hard turning for producing super surface integrity in terms of residual stress and
surface "nish.
Development of an analytical model of residual stress in metal cutting has been slow due to the
inherent complexity of machining processes. Okushima and Kakino [12] considered the ploughing
forces and the temperature distribution as the main causes of residual stresses. However, the
calculated results did not agree well with the experimental results. Wu and Matsumoto [10]
proposed a closed form solution to study the e!ect of workpiece hardness on the pattern of residual
stresses in orthogonal cutting. Their results showed an agreement with the experimental data
trends.
The "nite element method has been widely used since 1980s for performing stress analysis in
metal cutting. Several "nite element models have been developed to predict residual stress using
cutting forces and temperature from experiments and/or analytical solution as applied loads
without chip formation by Lajczok [13], Natarajan and Jeelani [14], and Liu et al. [15,16].
Strenkowski and Carroll [17] used an e!ective plastic strain criterion for chip separation. The
e!ective plastic strain criterion is the main factor to control the residual strain and residual stresses
in the machined layer. Lin et al. [18,19] applied a strain energy density criterion to predict residual
stress in machining. The FEM model by Shin [20] showed that the e!ects of thermal stresses and
the cutting and clamping forces on the "nal distribution of residual stresses were important to the
accurate prediction of residual stresses. The FEM results of Obikawa et al. [21] showed that
residual stress induced in the continuous chip formation is greater than that in the discontinuous
chip formation. However, several important factors (such as e!ects of temperature, strain rate, and
force unloading) that in#uence residual stress were neglected. Residual stresses in drilling 304
stainless steel were demonstrated by "nite element analysis by Guo and Dornfeld [22], but the
residual stress was not a major concern of this research.
C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086 1071
A literature review of residual stress in machining has shown that few practical models to predict
the e!ect of sequential cuts on residual stress distribution in the machined layer have been reported.
However, the a!ected layer by previous cutting is usually cut again in metal cutting process. It has
been proposed in the metal cutting literature that the machining process is not unique and the
residual stress will be a!ected by the previous cut because the newly generated surface layer has
residual stress and strain, and it is also strain hardened. Further, a methodology to control the
residual stress and its sensitivity to friction condition of the tool}chip interface has not been
addressed. Theoretical analysis of residual stress formation is a challenging problem. In general,
closed form analytical solutions for general elastic}plastic problems with large deformation, high
strain and strain rate, and high-temperature coupling are very di$cult to derive. Thus, "nite
element method is a feasible approach for modeling residual stresses in metal cutting. To solve
these problems, an explicit dynamic FEM software package, ABAQUS/Explicit [23], was used
in this study. The method has some important advantages for modeling metal cutting processes:
(1) It is computationally e$cient for the analysis of large models with relatively short dynamic
response times and for the analysis of extremely discontinuous events or processes. (2) It can be
used to perform analysis with complicated contact conditions. Contact algorithms are robust and
straightforward than their implicit counterparts. Explicit are therefore attractive for problems such
as metal cutting which involve complicated contact situations. (3) It uses a consistent, large-
deformation theory models can undergo large rotation and large deformation which occurs in
metal cutting. (4) It can be used to perform an adiabatic stress analysis if inelastic dissipation is
expected to generate heat in the material. Adiabatic analysis is typically used to simulate high-
speed manufacturing processes involving large amount of inelastic strain, where the heating of the
material properties caused by its deformation is an important e!ect because of temperature-
dependent material properties. (5) It allows for either automatic or "xed time increment. These
features of explicit dynamics procedure make it ideally suitable for analyzing the residual stress in
metal cutting.
2. Finite element modeling and assumptions
The two-dimensional "nite element model, Fig. 1, was developed under the plane strain
assumption that the width of the cut is at least "ve times greater than the uncut chip thickness. The
length of edge BC is 20 times longer than that of the uncut chip thickness for assuring steady-state
cutting. The workpiece is assumed to be semi-in"nite in size compared to the cutting tool. This is
accomplished by using semi-in"nite elements at the bottom of the workpiece. The element type
used in the other part of the workpiece is a four-node bilinear and reduced integration with
hourglass control to deal with the large deformation caused by chip formation process. Note that
some elements in the mesh are longer in the horizontal direction. This is done in anticipation of the
intense compression and shear straining to be expected as the tool moves from right to left. By
selecting this shape, numerical problem results from severe element distortion may be avoided. In
this model, edge AB is constrained against displacement, and other edges are not constrained to
allow deformation. For simplicity, a perfectly rigid cutting tool is assumed in view of the
signi"cantly high elastic moduli of most tool materials. This is also an acceptable approximation
since the elastic de#ection of the cutting tool is insigni"cant relative to the large plastic deformation
1072 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086
Table 1
Cutting tool geometry and cutting conditions
Rake angle 153
Clearance angle 53
Cutting speed 3 m/s
Uncut chip thickness 0.25 mm
Initial temperature 253C
of the workpiece. A sharp cutting tool was modeled using rigid elements. The cutting tool is free to
move in 1 direction. The boundary condition is enforced to constrain the motion of the cutting tool
in two direction. The perfect cutting tool has friction capability to model the sticking and sliding
interaction of the tool}chip which allows formation of the second shear zone as the tool advances
through the workpiece.
A critical problem of simulating the e!ect of cutting sequence on residual stress is how to
keep the physical state from the "rst cut and use it as initial condition of the second cut.
In this study the physical state (all residual stress and strain components) of the machined
surface after unloading in the "rst cut was used as initial conditions for the second cut. Other
cutting conditions of the second cut are the same as those of the "rst cut. In order to predict
residual stress, three unloading steps were implemented at the end of each cut in this study: (1)
releasing the cutting forces, (2) releasing the clamping forces, and (3) cooling of the workpiece to
room temperature.
The geometry of the cutting tool modeled and cutting conditions are given in Table 1. Residual
stresses from phase transformation and cutting tool wear were neglected. The workpiece is
assumed to be residual stresses free before the "rst cut. In addition, continuous chip formation was
assumed in this study.
3. Governing equations and numerical solution
In this analysis, the workpiece material is 304 stainless steel. The thermo-elastic}plastic material
properties of 304SS in Table 2 were used in this model [24,25]. The constitutive equations of the
workpiece material are described below.
3.1. Flow stress
In order to quantify the e!ect of plastic #ow characteristics of the workpiece material on the
cutting process, a thermo-elastic}plastic model with isotropic hardening and strain rate depend-
ence was considered in the analysis. In the linear elasticity regime, the total stress is de"ned from
the total elastic strain as [23]
o"DCJcCJ, (1)
C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086 1073
Table 2
Material properties of annealed 304 stainless steel
Density 7800 kg/m` Young's modulus (Gpa, 3C) (193, 25)
Strain-rate dependent D"1500, p"6 (179, 260)
Inelastic heat fraction 0.9 (160, 480)
(140, 700)
Speci"c heat (450, 25)
(J/kg, 3C) (500, 100) Yield and fracture strength, (230, 0, 25)
(525, 450) fracture strain and temperature (554, 1.6, 25)
(550, 850) (Mpa, 3C) (154, 0, 426)
(430, 1.19, 426)
Poisson's ratio (, 3C) (0.3, 25) (146, 0, 537)
(0.3, 260) (390, 1.35, 537)
(0.28, 480) (96, 0, 648)
(0.28, 700) (289, 0.77, 648)
where o is the total elastic stress, DCJ is the elasticity matrix, and cCJ is the total elastic strain. DCJ is
de"ned by giving Young's modulus and Poisson's ratio, Table 2. These parameters are given as
functions of temperatures.
In the plastic regime, #ow stress may be expressed in the form
o "f (cNJ, c

NJ, , c1, c2,

2
), (2)
where cNJ, c

NJ are e!ective plastic strain and e!ective plastic strain rate, respectively, is the
temperature, and c1 and c2 are material constants. Flow stress with strain hardening may be
expressed as
o "HcNJ
"
for cNJ)c
A
, (3)
where o is the #ow stress, H is the strength coe$cient, n is the strain hardening exponent, cNJ is the
e!ective plastic strain, and c
A
is the critical e!ective strain. In this study, the material strain
hardening was modeled by de"ning #ow stresses as a function of plastic strain and temperature,
Table 2. The values of H and n were determined from these input values.
3.2. Strain rate
The work material deforms extensively along the shear plane as it #ows through this zone.
Considering high strain rate in metal cutting is essential because of its e!ects on the strength of the
material. For many alloy steels this e!ect becomes very important in metal cutting. The strain rate
is represented by over stress power law
c

NJ"D

o
o
"
!1

N
for o *o
"
, (4)
where c

NJ is the equivalent plastic strain rate, o is the e!ective yield stress at a nonzero strain rate,
o
"
is the static yield stress (zero strain rate), and D, p are material parameters that may be functions
1074 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086
of temperature. These parameters are indicative of the strain rate sensitivity and can be determined
from experiments as 1500 and 6, respectively [25].
3.3. Heat generation
Heat generation during metal cutting is important in tool wear and plays an important role in
surface integrity and chip formation. The majority of the heat generated in the metal cutting
process comes from plastic deformation and chip tool interface friction. High-temperatures and
large-temperature gradient in metal cutting causes strength reduction of the workpiece material.
The ductile failure model used in this study also indicates this phenomenon.
As a "rst approximation, the heat generation incorporated in this study assumes adiabatic
condition. The temperature increment associated with the heat generation may be expressed by
V"
f

*
f
`
*
o
*
ccNJ
,C
N
, (5)
where V is the temperature increment. f

is the work-to-heat conversion factor, f

`
is the
conversion e$ciency factor, both f

and f
`
were taken as 0.9 as most of the workpiece deformation
is converted to thermal energy. o is the e!ective stress and ccNJ is the e!ective plastic strain
increment, and , and C
N
are the material density and speci"c heat, respectively.
3.4. Chip separation criterion
In the "nite element analysis of metal cutting, two factors are important in choosing a chip
separation criterion, i.e., the type of criterion and the magnitude of criterion. The type of criterion
should be based on the physics of chip separation. The magnitude depends on properties of the
work material under cutting conditions. A variety of chip separation criteria, either geometrical or
physical variables, and associated magnitudes have been proposed to simulate metal cutting
process. The physics of chip separation has not been clearly characterized because elements cannot
be made small enough to simulate a dislocation. Further, a reliable procedure for determining the
magnitude of the criterion is not available. Therefore, it is hard to determine which of these
separation criteria is better. However, the magnitude of the chip separation criteria a!ects the chip
separation process, the stress distributions in the machined layer, and the distributions of e!ective
plastic strain both in the chip and the machined surface [26]. E!ective plastic strain criterion [17]
and strain energy density criterion [19] have been used to investigate residual stress distribution.
Geometrical criterion has also been used to study residual stress [20,21]. The disadvantage of the
geometrical criterion is that it has no physical meaning. A material ductile failure model based on
e!ective plastic strain was used in this study. When the e!ective strain of the material reaches cNJ
"
,
the critical e!ective strain, damage in the material initiates. When the e!ective strain reaches cNJ
D
, the
failure e!ective strain, due to further deformation, the material fails and loses its ability to resist any
further load. The metal plasticity model in this paper uses the Mises stress potential for isotropic
behavior. However, the potential depends only on the deviatoric stress and is independent of
hydrostatic stress instead of normal stress on the shear plane. The proposed fracture criterion does
not take care of the microcracks in the free surface. Plastic deformation of the material assumes
C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086 1075
no inhomogeneity due to microcrack formation. The ductile failure criterion, denoted as FC, is
de"ned as
FC"(cNJ
"
, cNJ
D
). (6)
This criterion applies to all elements that are potentially part of the chip which contact the
cutting tool edge. The determination of values for cNJ
"
and cNJ
D
in steady-state cutting process has
been derived based on cutting mechanics [22]. In this study, cNJ
D
and cNJ
"
were 0.85 and 0.8,
respectively, from the experimental data.
3.5. Friction condition
The study of tool}chip interaction in "nite element simulation concerns force transmission at the
interface, and thus determines surface integrity of a machined layer. The key issue is the variation of
frictional and normal stress on the rake face. Two distinct regions of sliding and sticking were
observed on the rake face. Experimental results [27,28] exhibited a constant coe$cient of friction
in the region for frictional shear stress lower than certain value and an approximate constant
frictional stress (shear #ow stress) thereafter. Based on experimental stress distribution and direct
observation of the rake face, a constant coe$cient of friction is proposed in the sliding region and
for dry friction conditions and an equivalent shear stress limit, t

, in the sticking region for ideal

seizure conditions as an approximation. Whether a contact point is sliding or sticking depends on
the stress state at that point. Regardless of the magnitude of the contact pressure stress, sliding will
occur if the magnitude of the equivalent shear stress reaches t

. Mathematically, this can be

stated as
t
APGR
"jo when t(t

, (7)
t"t

when t*t

. (8)
The Coulomb friction model de"nes the critical shear stress, t
APGR
, at which sliding of the surfaces
starts as a fraction of the contact pressure, o, between the surfaces (t
APGR
"jo). The shear stress limit
is typically introduced in cases when the contact pressure stress may become very large (as can
happen in some manufacturing processes such as metal cutting) causing the Coulomb theory to
provide a critical shear stress at the interface that exceeds the yield stress in the material beneath
the contact surface. An upper bound estimate for t

is o
W
/(3, which o
W
is the Mises yield stress of
the material adjacent to the surface.
An ideal way to obtain the coe$cient of friction in the sliding region is based on the data of the
stress distribution on the rake and #ank wear faces from machining test. However, on-line
measurement of stress distribution in the faces in machining is a changeling. Alternatively, the
mean coe$cient of friction in the sliding zone will be calculated from measuring cutting and feed
forces. In case of sharp tool, the mean coe$cient of friction between tool}chip in orthogonal
cutting may be calculated by
j"c
F
R
#F
A
tan:
F
A
!F
R
tan:
, (9)
1076 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086
where F
A
and F
R
are measured cutting and thrust forces, : is rake angle, and c is a coe$cient. The
value of c depends on the di!erence between measured and simulated forces. If the di!erence is less
than 5%, the friction coe$cient can be considered acceptable because the variation in measuring
cutting and thrust forces can be as high as 10}15%. Otherwise, a modi"ed coe$cient value will be
used and the di!erence will be checked again. This process will continue until the acceptable
calculated forces are obtained. The advantage of this proposed approach is that the data of
measured cutting and thrust forces are relatively easy to obtain.
3.6. Equations of motion
The explicit dynamics analysis procedure is based on implementation of an explicit integration
rule together with the use of diagonal or lumped element mass matrices. The equations of motion
for the body are integrated using the explicit central di!erence integration rule [23]
u 'G>`'"u 'G`'#
At'G>'#At'G'
2
uK 'G', (10)
u'G>'"u'G'#At'G>'u 'G>`', (11)
where the subscript (i) refers to the increment number, ; represents the displacement vector and At
represents the time increment. The central di!erence integration operator is explicit in that the
kinematic state may be advanced using known values of u 'G`' and uK 'G' from the previous increment.
The explicit integration rule provides high computation e$ciency due to the use of diagonal element
mass matrices because the accelerations at the beginning of the increment may be computed by
uK 'G'"M

(F'G'!I'G'), (12)
where Mis the diagonal lumped mass matrix, F is the applied load vector, and I is the internal force
vector. The explicit procedure integrates through time by using many small time increments for
assuring convergence.
4. Simulation results and discussion
4.1. Chip formation
Fig. 1 shows chip formation process in the sequential cuts. It is recognized that the steady-state
cutting is obtained in the point of chip formation, cutting forces, and distributions of stress/strain.
Chip morphology of the "rst cut is almost identical to that of the second cut except that thickness
of the second chip is about 3% thinner, but the thickness di!erence is so small that it can be
ignored. From calculation, the shear angle is 14.53 and 15.13 for the "rst and the second cut,
respectively. The model suggests that the "rst cut produced work hardening in the workpiece that
the second cut `seesa. The e!ect of increased hardness on the onset shear plate is to increase the
shear angle, so the model is reasonable. The increase in shear angle results in a thinner chip (large
shear angle for the same uncut chip thickness) and large forces and less surface damage. This shows
evidence that material properties have been modi"ed by the "rst cut.
C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086 1077
Fig. 1. Chip formation in subsequent cuts.
Fig. 2. Equivalent plastic strain (Peeq) distribution in vertical direction.
Fig. 2 shows the equivalent plastic strain distribution in the two direction. The equivalent plastic
strain on the machined surface goes down from 0.14 to 0.12. Thus the degree of strain hardening
near the machined surface from the "rst cut is larger than that from the second cut. The depth of
a!ected layer may be expressed by the distance o from the machined surface. o

of 375 m in the
"rst machined layer is larger than o
`
of 276 m in the second machined layer. Therefore, the
a!ected layer from the "rst cut is deep.
The in#uence of the strain hardened machined layer from the "rst cut is expected. It is partly
responsible for the di!erences in shear angle and hence shear plane length, residual stress and strain
distribution in the two cuts. Liu and Brash [3] found that there is a close correlation between the
shear plane length and residual stress of the machined surface. The relationship of the a!ected layer
and residual stress is not clear, However, it can be inferred that they are closely related if the second
cut is within a certain depth of the a!ected layer with residual stress/strain and strain hardening
from the "rst cut.
4.2. Cutting forces
The calculated cutting F
A
and thrust force F
R
are shown in Figs. 3 and 4. The di!erence between
the calculated average forces and the measured forces are smaller than 5%. The measured forces to
1078 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086
Fig. 3. Cutting force F
A
.
Fig. 4. Thrust force F
R
.
determine the friction coe$cient of tool}chip interface are not shown. No appreciable di!erences of
calculated cutting and thrust forces exist between the two cuts in terms of magnitude and pattern.
Steady-state cutting is assumed to be reached when the cutting forces approach steady-state values.
Both cutting and thrust forces vary within certain amplitude due to node separation in cutting
process. This variation is inherent in the "nite element method. Compared to the average cutting
forces, the average thrust forces are small because of the high rake angle of 153 used. In addition to
cutting conditions, the values of cutting and thrust forces may also be a!ected by the plane strain
assumption, which restricts material #ow in the width direction of the chip. However, a negligible
error exists because the ratio of the uncut chip thickness to the undeformed chip width is
""""
in
this study.
Cutting and thrust forces are generally una!ected by the separation criterion because they are
calculated by averaging nodal forces over the entire rake face. The e!ect of the tool tip node is
signi"cantly lessened.
4.3. Residual stresses distribution
Residual stresses induced in the sequential cuts were investigated. Figs. 5 and 6 show the change
of the distribution of residual normal stress S11 in the cutting direction of the two cuts. High tensile
stresses are generated near the machined surface after the "rst cut, while the tensile residual stress is
considerably changed after the second cut, and a compressive residual stress region near the
machined surface is observed in Fig. 7. It is clear that the level of residual stress S11 decreases
C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086 1079
Fig. 5. Normal stress S11 (Pa) in horizontal direction in the "rst cut.
Fig. 6. Normal stress S11 (Pa) in horizontal direction in the second cut.
1080 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086
Fig. 7. Normal stress S11 distribution in vertical direction.
Fig. 8. Von Mises stress (Pa) in horizontal direction in "rst cut.
greatly after the second cut. Simulation result shows that the values of residual von Mises stress,
Figs. 8 and 9, also decrease, but not as much as that of the residual normal stress S11.
Fig. 7 shows the distribution of S11 in the machined layer versus the depth from the machined
surface in the two cuts. The high tensile stress layer left after the "rst cut is changed by the identical
C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086 1081
Fig. 9. Von Mises stress (Pa) in horizontal direction in the second cut.
second cut. The pattern of the stress distribution is modi"ed. A compressive region exists just under
the "nished surface of the second cut, and the level of the compressive stress increases in the inner
layer. Compressive residual stress S11 goes deep into the machined layer in the "rst cut. Thus, it
indicates that residual stress S11 control in the machined layer is possible by optimizing the second
cut. Experimental result [29] shows that residual stress S11 is tensile usually on the machined
surface in the "rst cut in machining aluminum. That S11 may become compressive depends on the
uncut chip thickness of the second cut. This provides indirect experimental evidence that the
residual stress pattern follows the simulated pattern.
Compared to the cutting forces, all components of residual stress may be changed by the second
cut. The magnitude of the residual stress usually increases with the value of the separation criterion
because the separation criterion is applied only to the nodes closest to the tool tip. Therefore,
caution must be exercised in selecting an appropriate value of the criterion to solve the residual
stress problem. By constraining the nodes to remain "xed to the workpiece for a longer time,
cutting forces are transmitted to the subsurface of the machined surface, resulting in the observed
residual stress.
4.4. Residual stress sensitivity to friction condition of tool}chip interface
The average friction coe$cient of tool}chip interface determined from the measured forces was
used to study the e!ect of sequential cuts on residual stresses in the machined layer. However, the
1082 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086
Fig. 10. Residual stress sensitivity to friction condition of tool}chip interface: A * 0.3; B * 0.5; C * 0.7.
friction coe$cient of tool}chip interface is a!ected by both the tool rake angle and the coolant
used, besides the in#uence of workpiece material. For the same tool rake angle and workpiece
material, the friction coe$cient may vary greatly, and thus the friction force applied to the
workpiece. So residual stresses are expected to be a!ected by the friction force on the tool}chip
interface. Considering the rake angle of 153 used in this study, the variation range of the friction
coe$cient is around 0.3}0.7 using di!erent coolant in machining steels [30]. In order to investigate
residual stress sensitivity to the friction condition, three levels of friction coe$cient were selected to
be 0.3, 0.5, and 0.7. For improving simulation accuracy, the "nite element mesh was re"ned and the
uncut chip thickness was assumed to be 0.1 mm. Other cutting conditions were same as those in
Table 1. Fig. 10 shows residual stress S

distribution under the three friction conditions. Residual

stress S

is tensile on the machined surface under low friction coe$cient 0.3. As the friction
coe$cient value increases from 0.3 to 0.5, then 0.7, residual stress S

changes from tensile to com-

pressive on the machined surface. The residual stress distribution pattern also changes correspondingly.
4.5. Cutting temperature
Fig. 11 shows the temperature distribution of the chip}tool interface in the two cuts. They show
that there is a friction heat source on the tool}chip interface as well as the heat source of plastic
deformation. The highest chip}tool interface temperature resulting in crate wear is not located at
the tip, but at a point along the rake face of the tool. The maximum temperature at the chip}tool
interface is around 10503C. The computed temperature may be slightly higher than that from the
experiment due to the assumption of the adiabatic condition and no heat conduction occurs
between the chip and the cutting tool. The temperature di!erence of the chip}tool interface at the
C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086 1083
Fig. 11. Chip temperature distribution in the two cuts.
Fig. 12. Temperature distribution in the machined layer in vertical direction.
two cuts is negligible. However, Fig. 12 shows that the temperature of the machined surface after
the second cut, 365.43C, is slightly lower than that of the "rst cut, which is 398.73C. Temperature
gradients in the machined layer are high at both cuts. They a!ect only a very thin region below the
machined surface. The temperature on the machined surface plays an important role in determin-
ing the residual stress distribution as temperature may a!ect the residual stresses distribution by
inducing phase transformation and thermal expansion. It may also a!ect other aspects of surface
integrity. From calculations, the maximum temperature on the machined surface is 398.73C, which
is lower than the phase transformation temperature for the material used in this study. Therefore,
residual stress distribution might not be a!ected by phase transformation. This justi"es the modeling
assumption.
1084 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086
5. Conclusion
A thermo-elastic}plastic coupling FEM model was proposed to predict the e!ect of sequential
cuts on the residual stress in the machined layer. The mechanical state from the "rst cut is used as
the initial condition for the second cut.
Tensile residual stress on the machined surface in the "rst cut may be changed to compressive by
optimizing the second cut. The levels of von Mises stress and hydrostatic stress are also reduced but
not as much as that of normal stress in cutting direction. The region of the a!ected layer in the
second cut becomes thinner. Changes of shear angle and chip thickness have all in#uence on
residual stress/strain of the second cut. Residual stress on the machined surface is sensitive to the
friction condition of the tool}chip interface.
Cutting forces, chip geometry, and temperature of the machined layer are only slightly a!ected
by the sequential cuts. Temperature on the machined surface is below the phase transformation
temperature.
Acknowledgements
This research has been supported by NSF under grant DMI-96102022, cost shared by USAF
Manufacturing Technology program, Caterpillar Inc., Torrington Co., and Ford Motor Co.
References
[1] Liu CR, Mittal S. Single-step super"nish hard machining: feasibility and feasible cutting conditions. Journal of
Robotics and Computer-Integrated Manufacturing 1996;12(1):15}27.
[2] Konig W, Berktold A, Koch KF. Turning versus grinding * a comparison of surface integrity aspects and
attainable accuracy. Annals of the CIRP 1993;42(1):39}43.
[3] Liu CR, Barash MM. The mechanical state of the sublayer of a surface generated by chip-removal process, Part 1:
cutting with a sharp tool. Transactions of the ASME, Journal of Engineering for Industry 1976a;Novem-
ber:1192}201.
[4] Liu CR, Barash MM. The mechanical state of the sublayer of a surface generated by chip-removal process, Part 2:
cutting with a tool with #ank wear. Transactions of the ASME, Journal of Engineering for Industry 1976b;
November:1202}8.
[5] Liu CR, Barash MM. Variables governing patterns of mechanical residual stress in a machined surface. Transac-
tions of the ASME, Journal of Engineering for Industry 1982;104(3):257}64.
[6] Kono Y, Hara A, Yazu S, Uchida T, Mori Y. Cutting performance of sintered CBN tools, Cutting tool materials.
Proceedings of the International Conference, American Society for Metals, Ft. Mitchell, Kentucky, September
15}17, 1980, pp. 218}95.
[7] El-Khabeery MM, Fattouh M. Residual stresses distribution caused by milling. International Journal of Machine
Tools Manufacture 1989;29(3):391}401.
[8] Schreiber E, Schlicht H. Residual stress after turning of hardened components. The International Conference On
Residual Stresses. Garnish-Partenkirchen (FRG), 1986, 853}60.
[9] Matsumoto Y, Barash MM, Liu CR. E!ects of hardness on the surface integrity of AISI 4340 steel. Transactions of
the ASME, Journal of Engineering for Industry 1986;108:169}75.
[10] Wu DW, Matsumoto Y. The e!ect of hardness on residual stresses in orthogonal machining of AISI 4340 steel.
Transactions of the ASME, Journal of Engineering for Industry 1990;112(3):245}52.
C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086 1085
[11] Tonsho! HK, Wobker HG, Brandt D. Tribological aspects of hard turning with ceramic tools. Journal of the
Society of Tribologists and Lubrication Engineers 1995;51(2):163}8.
[12] Okushima K, Kakino Y. The residual stresses produced by metal cutting. Annals of the CIRP 1971;10(1);13}4.
[13] Lajczok MR. A study of some aspects of metal machining using the "nite element method. PhDDissertation, North
Carolina State University, 1980.
[14] Natarajan R, Jeelani S. Residual stresses in machining using "nite element method. Computers in Engineering,
Computer Software and Applications ASME, New York, 1983;3:19}20.
[15] Liu CR, Lin ZC, Barash MM. E!ects of plane strain and plane stress conditions on stress "led in the workpiece
during machining * an elasto-plastic "nite element analysis. High Speed Machining 1984;2:167}80.
[16] Liu CR, Lin ZC, Barash MM. Thermal and mechanical stresses in the workpiece during machining. High Speed
Machining 1984;2:181}91.
[17] Strenkowski JS, Carroll JT. A"nite element model of orthogonal metal cutting. Transactions of the ASME, Journal
of Engineering for Industry 1985;107:349}54.
[18] Lin ZC, Lin YY, Liu CR. E!ects of thermal load and mechanical load on the residual stress of a machined
workpiece. International Journal of Mechanical Sciences 1991;33(4):263}78.
[19] Lin ZC, Lai WL, Lin HY, Liu CR. Residual stress with di!erent tool #ank wear lengths in the ultra-precision
machining of Ni-P alloys. Journal of Materials Progress Technology 1997;65:116}26.
[20] Shih AJ. Finite element simulation of orthogonal metal cutting. Transactions of the ASME, Journal of Engineering
for Industry 1995;117:84}93.
[21] Obikawa SH, Shirakashi T, Usui E. Application of computational machining method to discontinuous chip
formation. Transactions of the ASME, Journal of Manufacture Science and Engineering 1997;119:667}74.
[22] Guo YB, Dornfeld DA. Finite element analysis of drilling burr minimization with a backup material. Transactions
of NAMRI/SME 1998;XXVI:207}11.
[23] Hibbit K, Sorenson, Inc., 1996, ABAQUS/Explicit User's Manual, ver.5.6, Providence, RI.
[24] Kattus JR. Aerospace structural metals handbook, 1975.
[25] Guo YB, Dornfeld DA. Finite element modeling of drilling burr formation process. ASME Transactions, Journal of
Manufacture Science and Engineering, in press.
[26] Huang JM, Black JT. An evaluation of chip separation criteria for the FEM simulation of machining. ASME
Transactions Journal of Manufacture Science and Engineering 1996;118:545}53.
[27] Iwata K, Osakada K, Terasaka Y. Process modeling of orthogonal cutting by the rigid}plastic "nite element
method. ASME Transactions, Journal of Engineering Material and Technology 1984;106:132}8.
[28] Shirakashi T, Usui E. Simulation analysis of orthogonal metal cutting mechanism. Proceedings of the International
Conference. Production Engineering, Tokyo, Part 1, 1974, p. 535}40.
[29] Hsu HC. An elasto-viscoplastic "nite element model of orthogonal metal cutting for residual stress prediction. PhD
Dissertation, N.C. State University, 1992.
[30] Kronenberg. Machining science and application, Oxford: Pergamon Press, 1966.
1086 C.R. Liu, Y.B. Guo / International Journal of Mechanical Sciences 42 (2000) 1069}1086