ARTICLE IN PRESS

International Journal of Mechanical Sciences 49 (2007) 650660

www.elsevier.com/locate/ijmecsci

Finite element analysis of the inuence of tool edge radius on size effect
in orthogonal micro-cutting process
Kai Liu, Shreyes N. Melkote
The George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0406, USA
Received 15 August 2006; accepted 1 September 2006
Available online 7 November 2006

Abstract
The size effect in metal cutting is evident in the nonlinear scaling phenomenon observed in the specic cutting energy with decrease in
uncut chip thickness. It has been argued by many researchers that this scaling phenomenon is caused mainly by the cutting tool edge
radius, which purportedly affects the micro-cutting process by altering the effective rake angle, enhancing the plowing effect or
introducing an indenting force component. However, the phenomenological reasons why the tool edge radius causes size effect and the
relationship between the tool edge radius and the characteristic length scale associated with the size effect in micro-cutting has not been
sufciently claried. In this paper, a strain gradient plasticity-based nite element model of orthogonal micro-cutting of Al5083-H116
alloy developed recently is used to examine fundamentally the inuence of tool edge radius on size effect. The applicability of two length
scalestool edge radius and the material length scale l in strain gradient plasticityare also examined via analysis of data available in
the literature.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Tool edge radius; Micro-cutting; Size effect; Strain gradient; Finite element

1. Introduction
The size effect in metal cutting is characterized by a
nonlinear increase in the specic cutting energy, i.e. energy
per unit volume with decrease in uncut chip thickness. This
effect is especially prominent when cutting at micron level.
Many researchers have attempted to explain this scaling
phenomenon in terms of material strengthening mechanisms. Backer et al. [1] attributed the size effect to
signicantly reduced number of imperfections encountered
when deformation takes place in a small volume. Hence,
they argued that the material strength would be expected to
increase and approach the theoretical strength when the
uncut chip thickness is decreased. Kopalinsky and Oxley [2]
and Marusich [3] attributed the size effect in machining to
an increase in the shear strength of the workpiece material
due to a decrease in the tool-chip interface temperature as
the uncut chip thickness is decreased. Larsen-Basse and
Oxley [4] attributed the size effect to material strengthening
Corresponding author.

E-mail address: shreyes.melkote@me.gatech.edu (S.N. Melkote).

0020-7403/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmecsci.2006.09.012

due to an increase in the strain rate in the primary shear

zone with decrease in uncut chip thickness. Fang [5]
presented a complex slip line model for orthogonal
machining and attributed the size effect to the material
constitutive behavior of varying shear ow stress with
uncut chip thickness. Recently, Dinesh et al. [6] linked the
size effect observed in micro-/nano-indentation to that in
machining. The increase in hardness of a metallic material
with decrease in indentation depth is a consequence of the
dependence of ow stress of the metal on the strain
gradient in the deformation zone. They suggested that the
size-effect in machining can also be explained by the theory
of strain-gradient plasticity since strain gradients in
machining are very intense. Building upon the work in
Ref. [6], Joshi and Melkote [7] presented an analytical
model for orthogonal cutting that incorporates a material
constitutive law with strain gradient effects. Liu and
Melkote [8] attributed the size effect to strain gradient
strengthening of the material when cutting at small uncut
chip thickness levels. They presented a strain gradient
based nite element model for micro-scale orthogonal
cutting. Their simulation results predict a sizeable strain

ARTICLE IN PRESS
K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

Nomenclature
b
s
se
s0
sref
dij
r
rs
rg
G
a
Z
Zijk
Z0 ijk
e
_
_ o
ep
l

Burgers vector (nm)

ow stress (MPa)
effective stress
deviatoric stress
reference yield stress (MPa)
Kronecker tensor
dislocation density
statistically stored dislocation density
density of geometrically necessary dislocations
shear modulus
empirical constant
effective strain gradient
strain gradient tensor
deviatoric strain gradient tensor
effective (true) strain
effective strain rate
reference strain rate
effective plastic strain
characteristic material length

gradient strengthening effect in microcutting. This model was

recently validated through micro-cutting experiments [8].
While these researchers believed that the increase in
material shear strength with decrease in uncut chip
thickness is responsible for the size effect, others have
argued that the sharp tool assumption commonly made in
analyzing cutting processes causes the nonlinear scaling
effect. Since the edge radius of a cutting tool does become
comparable to or is sometimes greater than the uncut chip
thickness when the uncut chip thickness is reduced to the
micron level, the sharp tool assumption is no longer valid.
The effect of tool edge radius on the cutting process should
therefore be taken into account when making shear
strength calculations.
The existence of a ploughing mechanism, which is
believed to be caused by the cutting edge radius, has been
considered by many researchers [916] to be the primary
cause of size effect. Thomsen et al. [9] proposed the concept
of a separable ploughing force component not contributing
to chip formation and estimated this force through the
intercept obtained by extrapolating the cutting force to
zero uncut chip thickness. Albrecht [10,11] went on to
propose that two separate ploughing mechanisms occur
simultaneously, one on the rake face and the other around
the tool edge radius. Connolly and Rubenstein [13]
modeled ploughing based on the assumption that material
owing under the edge radius was deformed by an
extrusion-like process. Wu [14] modeled the ploughing
force component to be proportional to the volume of
material forced around the blunt edge and under the tool.
Endres et al. [15] rened this model to obtain the
proportional ploughing force component through a parameter estimation routine. Waldorf [16] developed a slip line

m
s
t
p
xi
v
Q_
T
To
Tm
A
B
c
n
m
rm
Cp
K
U_

651

friction coefcient
frictional shear stress
limiting shear stress in Coulomb friction model
contact pressure
local coordinates in mesoscale cell
volume of the mesoscale cell
volume heat ux
dimensionless temperature in JohnsonCook
equation T  T o =T m  T o
to ambient temperature
melting temperature
material constant in JohnsonCook equation
material constant in JohnsonCook equation
material constant in JohnsonCook equation
material constant in JohnsonCook equation
material constant in JohnsonCook equation
material density
specic heat capacity
thermal conductivity
material time rate of internal thermal energy

model to describe the effects of tool edge radius and tool

ank area on the cutting force.
Masuko [17] introduced the concept of an additional
force component necessary to cause the cutting edge to
penetrate the workpiece as an indenter. The indenting force
component Rn, shown in Fig. 1, is considered to be
independent of the uncut chip thickness. Therefore, he
concluded that this indenting force was responsible for the
size effect.
The above models assume that the ploughing/indenting
effects are separable. However, great difculty remains in
that this force component cannot be measured directly by
available methods such as the extrapolation technique [9]
or the feed dwell approach [18].
Lucca et al. [19] experimentally examined the dissipation
of mechanical energy in orthogonal ultra-precision ycutting of aluminum alloy 6061-T6 over depths of cut ranging
from 0.01 to 20 mm at a cutting speed of 48 m/min. Their
force data suggest that the process transitions from a
cutting-dominant process to a ploughing/sliding indentation-dominant process. Tool edge radius was seen to have a

R
Rn
Rc

2
Rn

CL

Fig. 1. Illustration of additional indenting force component [17].

ARTICLE IN PRESS
652

K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

signicant effect on the resulting forces by altering the rake

angle.
Qualitative effects of the tool edge radius on the cutting
process mechanics are well documented. It is well known
that the total machining force increases with an increase in
the edge radius. This increase in the machining force is
generally attributed to the ploughing mechanism. Bitans
and Brown [20] showed through their experiments on wax
that a larger edge radius resulted in a larger shear zone. It
can be seen from the work of Nakayama and Tamura [21]
and Kopalinsky and Oxley [2] that a larger edge radius
results in a smaller shear angle for the same uncut chip
thickness. However, size effect is also observed when
cutting with a sharp tool [19] (where the tool edge radius
was less than 10% of the uncut chip thickness). In addition,
Liu and Melkote [8] have shown that other factors such as
the drop in secondary shear zone temperature and strain
gradients can also contribute to the size effect.
Most of the previous studies are based on a mechanistic
modeling approach, which does not realistically represent
the conditions prevalent in cutting with a blunt edge tool.
Only a few reports examining this phenomenon via the
nite element [22,23] and molecular dynamics [24]
approaches have incorporated the tool edge radius effect.
In this paper, a strain gradient plasticity based nite
element model for orthogonal micro-cutting that has been
previously veried [8] for accurate prediction of the cutting
forces and specic cutting energy in microcutting of
Al5083-H116 is used to examine the role of the tool edge
radius on size effect. The phenomenological reasons for the
tool edge radius causing size effect and the relationship
between the magnitude of the tool edge radius and the
intrinsic material dependent length scale introduced by
strain gradient plasticity are investigated.
2. Finite element model
In this paper, a strain gradient based nite element
model of orthogonal micro-cutting developed earlier by
Liu and Melkote [8] is adopted as the simulation platform.
Note that this model has been recently validated through
micro-cutting experiments performed on a strain rate
insensitive aluminum alloy Al5083-H116 with diamond
tools [8]. The nite element model is a fully coupled
thermalmechanical model developed in the commercially
available software, ABAQUSs/Standard (version 6.4). For
the sake of completeness, this section briey reviews the
following aspects of the nite element model: (a) constitutive model, (b) modeling of toolchip interaction, (c)
modeling of chip separation, and (d) modeling of heat
transfer. Several key techniques and the overall simulation
approach are reviewed briey.
2.1. Constitutive modeling

behavior under highly localized inhomogeneous deformation. This theory links Taylors model of dislocation
hardening to a non-local theory of plasticity in which the
density of geometrically necessary dislocations is expressed
as a non-local integral of the strain eld. The Taylor
dislocation model denes the shear ow stress t in terms of
p
the dislocation density r as t aGb r, where the
dislocation density r is composed of the density of
statistically stored dislocations, rs, and the density of
geometrically necessary dislocations, rg.
The density of statistically stored dislocation can be
determined from the uniaxial stressstrain law in the
absence of strain gradient effects as
p
s 3aGb rs sref f ,
(1)
where sref is the reference yield stress in uniaxial tension.
The density of geometrically necessary dislocations is
related to the effective strain gradient Z as,
rg 2Z=b.

(2)

Based on these equations, a ow stress equation

accounting for the effect of geometrically necessary
dislocations can be written as
q
s sref f 2 ; _; T  lZ,
(3)
where l is the material length scale and is given by,
 2
G
2
l 18a
b.
sref

(4)

The ow stress function, f, is assumed to be of the

JohnsonCook [26] form as follows:


_

n
_
(5)
f ; ; T A B 1 C ln
1  T  m .
_ o
The strain gradient tensor is dened as
Zijk uk;ij ik;j jk;i  ij;k .

(6)

The effective strain gradient, which is a measure of the

density of geometrically necessary dislocations, is dened
as
q
Z 14Z0ijk Z0ijk ,
(7)
where the third-order deviatoric strain gradient tensor Z0ijk
is given by
Z0ijk Zijk  14dik Zjpp djk Zipp .

(8)

Using Taylor series expansion of the strain components

[25], the deviatoric strain gradient Z0ijk is obtained from the
following non-local volume integral,
Z


Z0ijk
ik xj jk xi  ij xk  14dik xj djk xi pp dv
v

Z

1
xm xk dv

A Taylor-based non-local theory of plasticity proposed

by Gao and Huang [25] is used to represent the material

In the current nite element implementation of this

model, the above integral is reduced to an area integral

ARTICLE IN PRESS
K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

653

(due to the orthogonal cutting assumption) and is

evaluated over a mesoscale cell dened around each
Gaussian integration point and contained within the
corresponding element [27].
The constitutive equations of Taylor-based non-local
theory of plasticity [25] can be summarized in rate form as
follows:

The fraction of dissipated energy converted into heat due

to plastic deformation and friction is assumed to be 0.9.
Heat generated due to friction is distributed via a weighting
factor of 0.5 between the two contact surfaces.

s_ kk 3k_kk ,

The pure deformation method [3,8] of chip formation is

implemented in this work via the adaptive remeshing
technique, thus avoiding a chip separation criterion. The
cutting process is linked to a metal forming type of
operation, with material owing around the tool. There is
no pre-dened parting line and therefore the shape of the
chip is not pre-determined. Instead, as the tool advances,
nodes of the workpiece move along the tool surfaces and
elements may deform strongly close to the tool tip. The
severely distorted elements are replaced during the remeshing step by new regular shaped elements. The material
that overlaps with the tool is removed in the remeshing step
(see Fig. 2).
An 8-node bi-quadratic displacement and bilinear
temperature element is chosen to approximate the workpiece geometry. The remeshing module consists of a preprocessor coded in FORTRAN 77 and the automatic mesh
generator feature of ANSYSs.
During simulation, the amount of penetration between
the tool and workpiece contact pair is checked at each time
step to determine the interference depth, which is used as
the remeshing criterion. Once the remeshing criterion is
satised, the outline of the workpiece is stored and the
automatic mesh generating module of ANSYSs is used to
create a new mesh for this region. Subsequently, the
solution is mapped from the old mesh to the new mesh. The
diffuse approximation method of solution mapping,
described in detail elsewhere [29], is used.
It is necessary to have a very ne mesh in the primary
and secondary deformation zones to resolve the relatively
steep stress and strain gradients present in these zones.
Therefore, the mesh pattern generated in each remeshing
step was designed to be denser in the vicinity of the two
deformation zones and coarse in regions away from these
zones (see Fig. 3). With the smallest element size of 0.06 mm
in the cutting zone for uncut chip thickness of 0.5 mm, this
approach reduces the number of elements by a factor of 10

(10)


3s0
s_ 0ij 2G _0ij  4sij

6Gs0kl _0kl s2ref lZ

3Gse s2ref ff 0

if se s and s_ e X0;
if se os and s_ e o0:

2G_0ij

(11)
The key feature of the Taylor-based non-local theory of
plasticity is that it does not involve higher order terms and
preserves the structure of classical continuum mechanics.
2.2. Toolchip interaction
Accurate representation of the interaction between the
tool and chip is vital for obtaining a reliable and realistic
simulation. However, the friction characteristic at the
toolchip interface is difcult to determine since it is
inuenced by many factors such as cutting speed, contact
pressure, and temperature. Extensive studies have been
performed on the mechanics of interaction along the
toolchip interface and several models have been developed. Of these, Zorevs model [28], which considers two
distinct regions of sliding and sticking at the toolchip
interface, is widely used. In the sliding region, the shear
stress is a fraction of the normal contact pressure, p. This
model, also termed here as the extended Coulomb friction
model, is used here to model toolchip interaction and is
given as
s mp when mpot sliding;
s t

when mpXt sticking:

(12)

2.3. Heat transfer model

In the nite element model, heat generation due to
plastic deformation and friction at the toolchip interface
is modeled as a volume heat ux. Heat conduction is
assumed to be the primary mode of heat transfer, which
occurs in the workpiece and tool.
The governing equation of heat transfer is as follows:
Z
Z
Z
rm U_ dV
q dS
Q_ dV ,
(13)
V

2.4. Chip separation modeling

where V is the volume of solid material with surface area S,

U_ is the material time rate of internal thermal energy, rm is
the mass density, q is the heat ux per unit area of the body
owing into the body, and Q_ is the heat supplied externally
into the body per unit volume.

Fig. 2. Illustration of material separation using the pure deformation

method.

ARTICLE IN PRESS
654

K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

Fig. 3. Finite element model conguration.

or more and greatly reduces the computational cost while

yielding a sufciently high resolution in the solution.
The typical workpiece dimensions used in the simulations are 3 mm  1 mm. As seen in Fig. 3, the bottom of the
workpiece is fully constrained while the top and right sides
of the tool are xed in the Y direction. A velocity load is
applied to the tool in the X direction to simulate the cutting
speed.
The workpiece deformation behavior is modeled as
elasto-plastic and the tool is considered to be rigid.
A ow chart summarizing the overall simulation
approach is shown in Fig. 4. Additional implementation
details of the model can be found in Ref. [27].

Start

Initial model definition

Node and element files

ABAQUS Input files

generation for new job

ABAQUS/ STANDARD

3. Orthogonal cutting simulation experiments

Orthogonal micro-cutting simulations were conducted
using the strain gradient based nite element model for a
strain rate-insensitive material, Al5083-H116. The model
incorporates ow stress dependence on strain, strain
gradient and temperature. Note that earlier work by the
authors [8] has shown that the strain gradient is the
dominant factor that contributes to size effect at small
uncut chip thickness levels (p10 mm) in this material. In
addition, temperature dependence of ow stress was found
to be dominant in causing size effect at relatively high
cutting speeds (4200 m/min) and large uncut chip thickness (420 mm). This latter size effect mechanism is
attributed to material strengthening due to a drop in the
secondary shear zone temperature with decrease in uncut
chip thickness [8].
The cutting conditions, tool geometry and friction
parameters used in the simulations are listed in Table 1.
Cutting tools with two different edge radii, 5 and 20 mm,
are used in the cutting simulations. The tool with an edge
radius of 5 mm is used in cutting simulations of 320 mm

Remeshing
module

Solution
Mapping
Module

Remeshing
criteria satisfied?
Y
Terminate analysis, Output
restart files
(.odb, .prt, .mdl, .res, .stt)

Extract geometry profile of

tool and workpiece

Fig. 4. Overall simulation approach.

uncut chip thickness. The tool with an edge radius of 20 mm

is used in cutting simulations of 7.575 mm uncut chip
thickness. For each edge radius, simulations have been
performed with and without the strain gradient term lZ in
Eq. (3) at a cutting speed of 200 m/min. In these
simulations, the ratio of uncut chip thickness t to edge
radius r (t/r ratio), varies from 0.375 to 3.75 for the 5 mm
edge radius case and from 0.6 to 4.0 for the 20 mm edge
radius case. A constant friction coefcient of 0.14 and

ARTICLE IN PRESS
K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

critical shear stress of 135 MPa were used in all the

simulations.
Flow stress data for the rate-insensitive Al5083-H116
alloy are obtained from the hot torsion experiments of
Zhou and Clode [30]. A modied JohnsonCook ow
stress equation (Eq. (14)) was used to t the ow stress data
of Zhou and Clode shown in Fig. 5 with a zero strain rate
hardening exponent. Coefcients of the modied JohnsonCook model are listed in Table 2. The other material
properties of Al5083-H116 and the diamond tool used in
the nite element simulations are given in Tables 3 and 4.


 
T  To m
s A Bn 1 _C 1 
.
(14)
Tm  To
4. Results and discussion
For the simulations without strain gradient effect and for
different t/r ratios, the von Mises stress distributions in the
workpiece are shown in Fig. 6. In all cases it can be seen
that the zone of plastic deformation extends below the
cutting edge.
Table 1
List of cutting and simulation conditions used
Workpiece material

Al5083-H116

Tool
Rake angle (1C)
Clearance angle (1C)
Tool edge radius (mm)
Cutting speed (m/min)
Width of cut (mm)
Uncut chip thickness (mm)

Diamond
5
5
5, 20
200
1
320 (for 5 mm edge radius),
7.575 (for 20 mm edge radius)
0.14
135

Friction coefcient
Critical shear stress (MPa)

The specic cutting energy was computed by dividing the

total force acting on the tool in the cutting direction by the
product of the workpiece width (unity) and uncut chip
thickness. It can be seen in Fig. 7 that the specic cutting
energy increases by approximately 100% when the t/r ratio
decreases from 4 to 0.6, which indicates that the size effect
is captured in the simulations using a tool with edge radius
even though the strain gradient effect is not included.
Comparison of the predicted specic cutting energy for
the edge radius tool with strain gradient effect and the
predicted specic cutting energy without strain gradient
effect, shown in Fig. 8, reveals that when simply accounting
for the tool edge radius effect only a fraction of the total
increase in specic cutting energy with decrease in uncut
chip thickness is captured. The model prediction with
strain gradient effect for the tool with edge radius shows a
much greater increase in the specic cutting energy.
It is possible that the nonlinear scaling effect observed in
Fig. 7 is due to the material strengthening effect caused by
temperature drop at the toolchip interface with decrease
in uncut chip thickness. Further examination of this
possibility reveals that the maximum secondary deformation zone temperature decreases by 49 1C for the 5 mm edge
radius case when the uncut chip thickness is reduced from
20 to 3 mm and by 41 1C for the 20 mm edge radius when the
uncut chip thickness is reduced from 75 to 7.5 mm. It is
evident that for both cases the temperature drops are
not large enough to account for the size effect observed in

Table 2
Modied JohnsonCook ow stress model coefcients for Al5083-H116
A (MPa)

B (MPa)

167

300

0.12

0.89

200.0
.
e = 18.04 1/s
160.0

Stress (MPa)

T=350C
T=400C

120.0

T=450C
T=500C

80.0

T=550C

experimental

40.0

Predicted
0.0
0.0

0.4

655

0.8

1.2

1.6

Strain

Fig. 5. Flow stress data for Al5083-H116 [30].

2.0

ARTICLE IN PRESS
K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

Fig. 7. This suggests that the observed nonlinearity is most

likely due to other effects.
To investigate the underlying reasons for the size effect
captured in the simulations with the edge radius tool and in
the absence of the strain gradient effect, simulations were

Table 3
Material properties of Al5083-H116 [31]
Density (kg/m3)
Specic heat (J/kg 1C)
Thermal conductivity (W/m K)
Coefcient of thermal expansion (mm/m 1C )
Melting temperature (1C)
Yield strength (MPa)
Youngs modulus (GPa)
Shear modulus (GPa)
Poissons ratio

2660
900
117
12.6
591638
228
71
26.4
0.33

Table 4
Material properties of diamond tool [31]
Density (kg/m3)
Specic heat (J/kg 1C)
Thermal conductivity (W/m k)
Coefcient of thermal expansion (mm/m 1C )
Melting temperature (1C)
Youngs modulus (GPa)
Poissons ratio

3500
471.5
1500
2.0
4027
850
0.1

run to analyze the nature of material deformation around

the tool tip at two uncut chip thickness values (7.5 and
75 mm) and 200 m/min cutting speed with a sharp tool and
a tool with 20 mm edge radius. Specically, the active
plastic yielding zone around the tool tip as well as the
toolchip contact lengths were analyzed. The tool and
workpiece material properties, boundary and initial conditions and frictional conditions were kept the same as
noted before. The resulting active plastic yielding zones for
the four simulation cases are shown in Figs. 912.
The active plastic yielding zones shown in Figs. 912
were identied by dening an element group that includes
only the elements undergoing plastic deformation at steady
1400
Specific cutting energy (MPa)

656

Edge radius 20 m

1200

Edge radius 5 m

1000
800
600
400
200
0
0

t / r ratio

Fig. 7. variation of specic cutting energy with or ratio for two edge radii,
without strain gradient effect.

Fig. 6. Steady state von Mises stress contours for edge radius of 5 mm, cutting speed of 200 m/min, w/o strain gradient effect.

ARTICLE IN PRESS
K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

state. It can be seen that the plastic shear zones for the
sharp tool are thinner than for the edge radius tool, which
are wider and extend toward the rake face of the tool. The
differences in plastic shear zone size and shape show that

Specific cutting energy (MPa)

1800
Sim. radiused edge tool w/o SG
Sim. radiused edge tool with SG

1600
1400
1200
1000
800
600
400
200
0
0

10

15

20

25

Uncut chip thickness (m)

Fig. 8. Variation of specic cutting energy versus uncut chip thickness for
edge radius tool, with and without strain gradient (SG) effect.

657

the edge radius of the tool changes the material ow

pattern around the tool tip by expanding and widening the
plastic shear zone. This in turn requires higher energy
dissipation and thus offers a likely explanation for the size
effect evident in Fig. 7 even in the absence of strain
gradient strengthening.
The toolchip contact lengths for the four cases are also
listed in Table 5. It is observed that at 75 mm uncut chip
thickness, the sharp tool and the edge radius tool yield
comparable contact lengths. The contact length for the
edge radius tool is only about 10% higher than for the
sharp tool. However, the difference in contact lengths
becomes larger at 7.5 mm uncut chip thickness, where the
toolchip contact length for the edge radius tool is three
times higher than for the sharp tool. This suggests that
much higher energy is consumed in frictional interaction at
the toolchip interface at the small uncut chip thickness of
7.5 mm when using a blunt tool. This results in a relatively
higher cutting force and consequently higher specic
cutting energy. Therefore, an additional reason for the
size effect observed in Fig. 7 in the absence of strain

Fig. 9. Active plastic yielding region for uncut chip thickness of 7.5 mm and sharp tool, 200 m/min cutting speed, without strain gradient effect.

Fig. 10. Active plastic yielding region for uncut chip thickness of 7.5 mm and tool with edge radius, 200 m/min cutting speed, without strain gradient effect.

ARTICLE IN PRESS
K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

658

Fig. 11. Active plastic yielding region for uncut chip thickness of 75 mm and sharp tool, 200 m/min cutting speed, without strain gradient effect.

Fig. 12. Active plastic yielding region for uncut chip thickness of 75 mm and tool with edge radius, 200 m/min cutting speed, without strain gradient effect.

Table 5
Simulated toolchip contact lengths
Simulated toolchip contact length (mm)

Sharp tool
Tool with 20 mm edge
radius

7.5 mm uncut chip

thickness

75 mm uncut chip
thickness

10
30

110
130

gradient effects is the increased toolchip frictional

interaction for the edge radius tool at smaller uncut chip
thickness values.
4.1. Discussion of length scales in micro-cutting
The length scale commonly associated with size effect in
micro-cutting is the tool edge radius. The belief is that the
nonlinear increase in specic cutting energy occurs at an
uncut chip thickness value close to the edge radius of the
cutting tool. However, it has been shown here and in other
recent work by the authors [8] that alternate material

strengthening-based mechanisms such as strain gradient

plasticity, which introduces an intrinsic material dependent
length scale l (given by Eq. 4), can also contribute to the
size effect. The strain gradient effect is signicant only
when the product of the material length scale l and effective
strain gradient Z is comparable to or greater than the
f 2 ; _; T term in Eq. (3). In other words, under conditions
where the strain gradient effect is dominant, the nonlinearity in specic cutting energy should occur at or near
the material length scale l. The purpose of the following
discussion is to examine the validity of the two length
scalestool edge radius and the material length scale l
for explaining the size effect evident in data available in the
literature.
Since the increasing trend in specic cutting energy
appears as a smooth curve, a criterion is needed to dene
the starting point of the nonlinear increase in specic
cutting energy vs. uncut chip thickness curve. In this work,
the uncut chip thickness at which the curvature of the
specic cutting energy curve is maximum is considered to
be the starting point of the nonlinearity. It should be noted
that the curvature calculation is affected by the scales of
the x- and y-axes. Consequently, the specic cutting energy

ARTICLE IN PRESS
K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

data taken from the literature is plotted in a normalized

coordinate system such that both axes are scaled to a [0, 1]
range. This is done by normalizing each axis by the
maximum value for that axis.
Table 6 summarizes the length scale analysis performed
using data from the literature. The measured length scale in
the table represents the uncut chip thickness at which the
specic cutting energy starts to deviate from linearity. This
is established using the maximum curvature criterion
dened earlier. Note that the material length scale l is
estimated from Eq. (4). In some cases, sufcient information about the workpiece material is not available in the
referenced papers and consequently the material length
scale cannot be computed reliably. These cases in the table
have no entry in the cell for material length scale.
It can be seen from Table 6 that the starting point of the
nonlinearity in specic cutting energy determined from the
maximum curvature criterion does not always correspond
to the tool edge radius. For several cases including Pure
Zinc, Al2024, Al6061-T6 at 300 m/min, and Al5083-H116,
it is signicantly different from the tool edge radius.
However, it can be seen that the starting point of the
nonlinearity is generally closer to the material length scale l
especially for data generated at low cutting speed, where
the strain gradient effect is expected to be dominant [8].
For example, this observation can be made from the data
for Pure Zinc, Al2024, Al6061-T6 at a cutting speed of
48 m/min, and Al5083-H116 at 10 m/min.
Another interesting observation that can be made from
an examination of the data in Table 6 is the relatively small
change in the measured length scale with large changes in
the tool edge radius at a given cutting speed. For example,
for Al2024, a change in the tool edge radius from upsharp
to 101.6 mm produces a minimal change in the measured
length scale. Similar observations can be made in the case
of Pure Zinc where a ve-fold change in the tool edge
radius (from 5 to 25 mm) produces very little change in the

659

measured length scale (from 54.1 to 52.1 mm). However, a

signicant change in cutting speed does produce a noticeable change in the measured length scale. This can be
clearly seen for Al6061-T6 and Al5083-H116 and is
consistent with the arguments made in Ref. [8] that the
contribution to size effect of material strengthening due to
the temperature drop in the secondary deformation zone
with decrease in uncut chip thickness is more signicant at
higher cutting speeds.
For cutting conditions where multiple factors potentially
contribute to the size effect, the starting point of the
nonlinearity may be affected by multiple factors such as the
tool edge radius, the material length scale l and the
temperature drop in the secondary deformation zone.
5. Conclusions

The tool edge radius accounts for only part of the size
effect in micro-cutting. The edge radius affects the
material deformation process and thereby contributes to
the size effect in the following ways: (i) by changing the
material ow pattern around the tool tip by expanding
and widening the plastic shear zone, and (ii) causing
higher energy dissipation due to increased toolchip
contact length at smaller uncut chip thickness values.
Under cutting conditions where the strain gradient effect
is dominant, such as at low cutting speeds, small uncut
chip thickness and negligible tool edge radius, the
nonlinear increase in specic cutting energy occurs near
the material length scale l.
Analysis of size effect evident in data available in the
literature clearly suggests that, for a given cutting speed,
the nonlinear increase in specic cutting energy often
does not occur at an uncut chip thickness value equal to
or even close to the tool edge radius.
Analysis of existing data also suggests that increasing
the cutting speed produces a noticeable change in the

Table 6
Comparison of length scales in micro-cutting
Material

Measured length
scale (mm)

Total edge radius

(mm)

Pure Zinc [34]

54.1
52.1
90.1
38.1
42.1
42.1
0.95
56.1
62.1
22
0.9
2.01
10
1.44
24.1

5
25
134
Upsharp
25.4
101.6
0.2
396.9
793.8
6
0.25

Al2024 [35]

Al6061-T6 [32,36]

AISI 1045 steel [2]

TeCu [32]
Oxygen-free Cu [33]
Brass [21]
Al5083-H116 [8]

34
0.0650.1
7

Material length scale

l (mm)

52.8
52.8
52.8
3.7
3.7
3.7
27.6

5.7
5.7

Cutting speed
(m/min)

Uncut chip thickness range

(mm)

56.4
56.4
56.4
56.4
56.4
56.4
48
300
300
420
7.6
6
0.1
10
200

17508
17508
17508
8.8508
8.8508
8.8508
0.0120
11.9498
11.9498
5140
0.0120
0.5110
250
0.510
20200

ARTICLE IN PRESS
K. Liu, S.N. Melkote / International Journal of Mechanical Sciences 49 (2007) 650660

660

uncut chip thickness value at which the nonlinearity in

specic cutting energy begins, thus suggesting the
dominant role of material strengthening due to temperature drop in the secondary deformation zone on the
size effect at higher cutting speeds.

Acknowledgement
This work was supported by the National Science
Foundation through Grant DMI-0300457.

References
[1] Backer WR, Marshall ER, Shaw MC. The size effect in metal cutting.
Transactions of ASME 1952 7461.
[2] Kopalinsky EM, Oxley PLB. Size effects in metal removal processes.
Institute of Physics Conference Series 1984;70:38996.
[3] Marusich TC. Effects of friction and cutting speed on cutting force.
In: Proceedings of IMECE (ASME), November 1116, New York,
Paper#: MED-23313, 2001.
[4] Larsen-Basse J, Oxley PLB. Effect of strain rate sensitivity on scale
phenomena in chip formation. In: Proceedings of the 13th international machine tool design and research conference, University of
Birmingham, 1973. p. 20916.
[5] Fang N. Slip-line modeling of machining with a round-edge tool
part II: analysis of the size effect and the shear strain rate. Journal of
the Mechanics and Physics of Solids 2003;51(4):74362.
[6] Dinesh D, Swaminathan S, Chandrasekar S, Farris TN. An intrinsic
sizeeffect in machining due to the strain gradient. In: Proceedings of
2001 ASME IMECE, November 1116, NY, 2001. p. 197204.
[7] Joshi SS, Melkote SN. An explanation for the size-effect in machining
based on strain gradient plasticity. Journal of Manufacturing Science
and Engineering, Transactions of the ASME 2004;126(4):67984.
[8] Liu K, Melkote SN. Material strengthening mechanisms and their
contribution to size effect in micro-cutting. In: Proceedings of 2005
ASME international mechanical engineering congress and exposition,
Orlando, FL, November 2005. p. 18.
[9] Thomsen EG, Lapsley JT, Grassi RC. Deformation work absorbed
by the workpiece during metal cutting. Transactions of ASME
1953;75:591603.
[10] Albrecht P. New developments in the theory of metal cutting process:
Part Ithe ploughing process in metal cutting. ASME Journal of
Engineering for Industry 1960;82:34857.
[11] Albrecht P. New developments in the theory of metal cutting process:
part IIthe theory of chip formation. ASME Journal of Engineering
for Industry 1961;83:55768.
[12] Armarego EJA, Brown RH. On the size effect in metal cutting.
International Journal of Production Research 1962;1(3):7599.
[13] Connolly R, Rubenstein C. The mechanics of continuous chip
formation in orthogonal cutting. International Journal of Machine
Tool Design Research 1968;8:15987.
[14] Wu DW. Application of a comprehensive dynamic cutting force
model to orthogonal wave generating processes. International
Journal of Mechanical Science 1988;30(8):581600.

[15] Endres WJ, Devor RE, Kapoor SG. A dual mechanism approach to
the prediction of machining forcesParts 1 and 2. ASME Journal of
Engineering for Industry 1995;117:52642.
[16] Waldorf DJ. Shearing, Ploughing and wear in orthogonal machining,
Ph.D. thesis, University of Illinois at UrbanaChampaign, 1996.
[17] Masuko M. Fundamental research on the metal cutting: second
report. Bulletin of Japan Society of Mechanical Engineers 1956;
22(118):371.
[18] Colwell LV. Methods for sensing the rate of tool wear. Annals of
CIRP 1971;19:647.
[19] Lucca DA, Seo YW, Rhorer RL. Energy dissipation and toolworkpiece contact in ultra-precision machining. Tribology Transactions
1994;37(3):6515.
[20] Bitans K, Brown RH. An investigation of the deformation in
orthogonal cutting. International Journal of Machine Tool Design
Research 1965;5:15565.
[21] Nakayama K, Tamura K. Size effect in metal-cutting force. American
Society of Mechanical Engineers 1968;67-Prod-9, 1967:18.
[22] Kim KW, Lee WY, Sin HC. A nite element analysis of machining
with the tool edge considered. Journal of Material Processing
Technology 1999;86:4555.
[23] Yen YC, Jain A, Altan T. A nite element analysis of orthogonal
machining using different tool edge geometries. Journal of Material
Processing Technology 2004;146(1):7281.
[24] Komanduri R, Chandrasekaran N, Raff LM. Effect of tool geometry
in nanometric cutting: a molecular dynamics simulation approach.
Wear 1998;219:8497.
[25] Gao H, Huang Y. Taylor-based nonlocal theory of plasticity.
International Journal of Solids and Structures 2001;38:261537.
[26] Johnson GR, Cook WH. A constitutive model and data for metals
subjected to large strain, high strain rates and high temperatures. In:
Proceedings of the Seventh International Symposium on Ballistics,
1983. p. 5417.
[27] Liu K. Process modeling of micro-cutting including strain gradient
effect, Ph.D. thesis, Mechanical Engineering, Georgia Institute of
Technology, 2005.
[28] Zorev NN. Inter-relationship between shear processes occurring
along tool face and shear plane in metal cutting. International
Research in Production Engineering, 1963. p. 429.
[29] Hamel V, Roelandt JM, Gacel JN, Schmit F. Finite element modeling
of clinch forming with automatic remeshing. Computers and
Structures 2000;77:185200.
[30] Zhou M, Clode MP. Constitutive equations for modeling ow
softening due to dynamic recovery and heat generation during plastic
deformation. Mechanics of Materials 1998;27:6376.
[31] MatWeb material database /http://www.matweb.comS.
[32] Lucca DA, Rhorer RL, Komanduri R. Effect of tool edge geometry
on energy dissipation in ultraprecision machining. Annals of the
CIRP 1993;42(1):836.
[33] Furukawa Y, Moronuki N. Effect of material properties on ultra
precision cutting process. Annals of the CIRP 1988;37(1):1136.
[34] Schimmel RJ, Endres WJ. Application of an internally consistent
material model to determine the effect of tool edge geometry in
orthogonal cutting. Transactions of the ASME 2002;124:53643.
[35] Kountanya RK. Process mechanics of metal cutting with edge
radiused and worn tools, Ph.D. thesis, University of Michigan, 2002.
[36] Waldorf DJ, DeVor RE, Kapoor SG. An evaluation of ploughing
models for orthogonal machining. Journal of Manufacturing Science
and Engineering 1999;121:5507.