International Journal of Machine Tools & Manufacture 64 (2013) 20–30

Contents lists available at SciVerse ScienceDirect

International Journal of Machine Tools & Manufacture

journal homepage: www.elsevier.com/locate/ijmactool

Brittle damage and interlaminar decohesion in orthogonal micromachining

of pyrolytic carbon
Vivek Bajpai, Ramesh Singh n
Department of Mechanical Engineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra 400076, India

a r t i c l e i n f o a b s t r a c t

Article history: Engineered features on pyrolytic carbon (PyC) have been reported to improve the functional
Received 18 May 2012 performance of the cardiovascular implants. PyC also finds application in thermonuclear components
Received in revised form due to its unique directional thermal properties. Note that PyC comprises of stacked layers of brittle
21 May 2012
graphite-like material and its machining characteristics differ from plastically deformable isotropic
Accepted 16 July 2012
Available online 31 July 2012
materials due to brittle damage and interlaminar decohesion. Consequently, this study is aimed at
understanding the mechanics of material removal in the plane of transverse isotropy (horizontally
Keywords: stacked laminae) of PyC via a finite element model. A damaged plasticity material model has been used
Pyrolytic carbon to capture the effect of material degradation of a brittle material under machining. Uniaxial tension/
Orthogonal micromachining
compression tests have been carried out to calibrate the damaged plasticity model. A surface based
Damaged plasticity
cohesive bonding has been used between the layers to simulate the interlaminar decohesion which
Surface based cohesive bonding
results in peeling, slipping and delamination during machining. The model predicts the cutting force
and thrust forces under different process conditions. The cutting force predictions from the finite
element model have been validated against the experimental data for different cutting conditions. In
addition, the model also predicts the chip morphology for different machining conditions. The
prediction error in the model lies between 2% and 23%. Parametric studies have also been performed
to understand the effect of the machining parameters, such as rake angle, uncut chip thickness on the
process response. It is found that use of the positive rake angle decreases the cutting forces up to 72%.
& 2012 Elsevier Ltd. All rights reserved.

1. Introduction conventional isotropic materials. The material separation phenom-

Pyrolytic carbon (PyC) is a material known for its excellent
biocompatibility, high strength-to-weight ratio and unique direc-
tional thermal properties which finds application in biomedical
implants and thermal sinks. It has layers of brittle graphite-like
material as shown in Fig. 1 [1]. The layered AB plane is the plane
of transverse isotropy and the C plane is anisotropic. Mechanical
Micromachining can be used to create functional features in PyC.
However, with notable exceptions of few experimental studies
[1–3], most of the experimental and modeling work reported in
the literature on mechanical micromachining (micromilling, micro-
turning, etc.) is for isotropic, homogeneous, and plastically deform-
ing metals [4–7].
To the best of authors’ knowledge, no work has been reported
on the modeling of micromachining process for layered brittle
materials. Due to the layered and anisotropic nature of PyC,
the machining characteristics are significantly different from the
n
Corresponding author. Tel.: þ91 22 25767507(O), þ91 22 25768507(R);
fax: þ 91 22 25726875.
E-mail addresses: vivekbajpai@iitb.ac.in (V. Bajpai), ramesh@me.iitb.ac.in,
rsingh@iitb.ac.in (R. Singh).
enon in micromachining of PyC has yet to be fully understood.
During the orthogonal micromachining of PyC, the chip layer is
peeled and the chip failure occurs due to tension in the bottom
portion and compression in the top portion. This tension compres-
sion results in delamination of the layers in the chip. Note that the
brittle materials degrade under mechanical loading. The degrada-
tion phenomenon can be captured via a damaged plasticity model
wherein the stiffness of the material deteriorates as a function of
strain [8]. The chip formation mechanism uses traction separation
law for interlaminar decohesion which captures the peeling, slipping
and delamination in the layered chip. The damaged plasticity model
is used for capturing the tensile and compressive behavior of the
degraded material. Uniaxial tension and compression tests are
required for calibrating the parameters of the damage model. Finally,
a finite element model for orthogonal cutting in AB plane (plane of
transverse isotropy) is developed based on damaged plasticity and
traction separation to understand the material removal mechanism
in PyC. The proposed model has been validated with orthogonal
micromachining experiments for cutting forces and chip morphol-
ogy. To characterize the effect of machining conditions, a parametric
study has been performed to study the effect of depth of cut, tool
width, rake angle and cutting speed on the brittle failure and the

0890-6955/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijmachtools.2012.07.007
 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30 21

B 1 μm

A
10μm

Fig. 1. Schematic and SEM image of PyC [1].

Fig. 2. Schematic of chip formation mechanisms: (a) peeling and bending in positive rake angle; (b) slipping, bending and crushing in negative rake angle.

interlaminar decohesion which affects the cutting forces and chip
morphology.
2. Physical description of the micromachining process in PyC
induces significant amount of crushing due to the brittle nature of
PyC. The ploughing phenomenon observed in the plastically flowing
material is replaced by crushing. The rubbing between the flank and
workpiece has also been modeled.

In the orthogonal micromachining of PyC in the plane of 3. Material characterization

transverse isotropy, the following mechanisms take place: peeling,
slipping, delamination, bending rubbing and ploughing/crushing.
Figs. 2(a) and (b) show the mechanism of the chip formation for
positive and negative rake angles, respectively.
Peeling and slipping occur in the between the two laminae close
to the tool tip due to interlaminar decohesion. The interlaminar
peeling/slipping depends upon the magnitude of the traction force in
the cutting and thrust directions and the rake angle. As shown in
Fig. 2(a), the bending stresses are generated in the chip due to the
peeling of the layer while cutting with a positive rake angle tool. The
bending of the chip induces compression in the top and tension in
the bottom section. Fig. 2(b) shows cutting with a negative rake angle
which results in the slipping of the chip. Contrary to the peeling
mechanism, the tensile stresses are induced in the upper section and
the compressive in the lower section of the layer. These stresses
eventually lead to the material separation. The above two cases result
in the delamination of the chip during bending, magnified view of
Fig. 2 shows chip delamination.
A finite element model capturing the abovementioned phenom-
ena could enhance the understanding of micromachining process
mechanism of layered brittle materials. The proposed model con-
siders the effects of tool nose radius and rake angle on the cutting
forces and chip morphology. Note that the negative rake angle
3.1. Mechanical properties
Pyrolytic carbon is a layered anisotropic material produced by the
decomposition of hydrocarbon gases. A layered structure makes it
stronger in certain orientations. Some of the mechanical properties of
PyC used in the simulations have been taken from the material
datasheet [13]. Flexural strength and fracture toughness (for different
orientations) have been characterized and reported in the literature
[9–12]. As mentioned previously, the damaged plasticity material
model needs postfailure behavior of the material under uniaxial
tension and compression, which has not been reported in the
literature. Hence, a number of tests were conducted on the material.
The stacked layers of the material and the cohesive bonding have
been modeled separately in the proposed model. The mechanical
properties of a single layer are required to simulate the machining
response of the stacked laminae. The layered material is tested under
tension and compression to estimate the properties of a single layer.
3.2. Material testing under uniaxial tension and compression
The damaged plasticity model requires material response
under the elastic limit as well as the postfailure response beyond
22 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30
cracking strain in the strain softening regime in tension. Three
dog-bone specimens of PyC were prepared for the tensile testing. The
specimens were tested on a universal testing machine (INSTRON
5566As) with a 5 kN load cell. Fig. 3(a) shows the test specimen
before and after testing, it is clear from Fig. 3(a) that the specimen
failed in tension at the narrower section in the middle. The test
revealed that the layered material failed without interlaminar slip-
ping, hence, the material response of the layered structure can be
used to estimate the tensile behavior of a single lamina. Similar to the
material behavior in tension, the damaged plasticity model requires
the material response under compression in three different regimes,
namely, elastic, compression hardening and strain softening. Three
specimens of pyrolytic carbon of dimensions (6 mm  2.4 mm 
1 mm) were prepared for the compression tests in two mutually
perpendicular directions (AB and C). The AB plane comprises of
horizontally stacked laminae which does not allow interlaminar
slipping. In addition, the height of the specimen was kept very small
(1 mm) to ensure crushing and avoid buckling (see Fig. 3(b)). In the
absence of interlaminar slipping and buckling, it is expected that the
compression test in the AB plane will yield the stress–strain behavior
of a single layer.
To confirm that the material response of the single layer is correct
and can be used to capture the response of the stacked laminae,
compression tests were carried out in C plane in conjunction with a
finite element simulation of the test. It is expected that the
compression in C plane will lead to buckling of the layers and
de-bonding which could be simulated and benchmarked against the
experimental data. Fig. 3(c) shows the sample before and after the
compressive test in C plane. The crushed material shows partly
buckled layers due to de-bonding of the laminae. A finite element
simulation of the compressive test was conducted in C plane wherein
several layers have been glued together via surface based cohesive
Fig. 3. Test specimen before and after test for: (a) tension (AB plane); (b) compression (AB plane); (c) compression (C plane).
bonding to account for the interlaminar strength. Fig. 4(a) shows the
stress distribution in the direction of the compressive load and the
corresponding deformation in the layers. The results show that the
interlaminar decohesion, buckling and crushing of the layers occur in
C plane. The stress–strain response obtained from the simulation of
compression in
C plane is in good agreement with the experimental data as shown
in Fig. 4(b). It can be inferred from the abovementioned results that
the single layer properties determined from the tests in AB plane are
reasonably accurate and can be used in the calibration of damaged
plasticity model parameters discussed in detail later during the
description of damaged plasticity model.
4. Finite element modeling
Abaquss Explicit finite element code has been used in the
simulation of the orthogonal machining of PyC to capture the
interlaminar decohesion (peeling, slipping and debonding), bend-
ing, crushing and rubbing as described previously. The tool has
been considered as a rigid body in this analysis. Fig. 5 shows the
flow chart of FE modeling. The model contains two major
modules, numerical formulation and chip formation mechanism.
The numerical formulation contains geometry, interactions, con-
straints, loading/boundary conditions and material properties.
The chip formation mechanism used in the model is based on
peeling due to interlaminar decohesion and the degradation/
failure of the chip via surface based cohesive zone and damaged
plasticity, respectively. The damaged plasticity model is used to
define the material behavior in tension and compression and the
surface based cohesive zone model is used to define chip forma-
tion in peeling and delamination within the chip after separation.
 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30 23

damage behavior of the material in tension and compression can

Flat compression be explicitly modeled. The built-in damaged plasticity model in
plate Abaqus Explicits has been modified for modeling the damaged
Workpice for elasticity in combination with tensile and compressive plasticity
compression in the plane of transverse isotropy (AB plane) of PyC to represent
its inelastic behavior.
Fixed nodes The model is a continuum, plasticity-based, damage model for
brittle materials. It assumes that the two major failure mechanisms
are tensile cracking and compressive crushing. The progression of
the failure surface is controlled by two hardening variables, equiva-
Delamination lent plastic strain in tension (e~ pl ~ pl
t ) and compression (e c ) which
and buckling govern the failure mechanism. The postfailure stresses in tension
(t)/compression (c) is given by Eq. (1):

1 st=c ¼ ð1dt=c ÞE0 ðet=c e~ ck=in

t=c Þ ð1Þ

2
Magnified view 3
dt=c ¼ dt=c ðe~ pl
t=c , y,f i Þ,. . .,ð0 r dt=c r 1Þ ð2Þ

where st/c post failure stress, dt/c degradation in the original stiffness
80 (E0) of the material, et/c is the stain, e~ ck
t is cracking strain in tension

70
Simulation and e~ in
c is inelastic strain in compression. The subscripts ‘t’ and ‘c’
stand for tension and compression, respectively.
Experiment
60
4.2. Calibration of the material model parameters
Stress (Mpa)

50

40 The uniaxial tension and compression tests yielded the stress–

strain relationship till failure. The damaged plasticity model needs
30
the following set of values to simulate the physical phenomenon of
20 orthogonal machining:

10 1. Modulus of elasticity (E0).

2. Postfailure stress in tension as a function of cracking strain
0
0 0.05 0.1 0.15 0.2 0.25 (st v/s e~ ck
t ).
3. Postfailure stress in compression as a function of inelastic or
Strain
crushing strain (sc v/s e~ in
c ).
Fig. 4. (a) Simulation of compression test in C plane; (b) comparison of stress– 4. Degradation factor (dt/c) in the elastic stiffness in tension
strain plot for simulation and experiment. and compression versus cracking strain and inelastic strain,
respectively.
Chip Formation Numerical
FE Modeling
and Failure Formulation The mechanical responses captured the tensile and compres-
sive tests are shown in Fig. 6(a) and (b), respectively. It shows that
Peeling(Traction Geometry (Interactions a maximum stress of 46 MPa at a strain of 0.0262 is obtained
separation formulation) and constraints) under tension whereas a stress of 81 MPa at a strain of 0.136 is
Chip failure (Damaged Loading and Boundary obtained under compression. Note that there are two distinct
Model Validation
plasticity model in Conditions regions of interest in the compression tests, namely, compression
compression and tension Material Properties
mode) hardening (2–3 in Fig. 6(b)) and strain softening (3–4 in Fig. 6(b)).
Delamination(Traction Parametric In the case of tension test, only strain softening behavior has been
separation formulation) Study observed (2–3 in Fig. 6(a)).
It is required to provide the postfailure stress and damage as a
Fig. 5. Flow chart of the FE modeling.
function of cracking strain ðe~ ck ~ in
t Þ and inelastic strain ðe c Þ in case of
tension and compression, respectively. Figs. 6(c) and (d) show
plot of postfailure stress and damage v/s creaking/inelastic strain
The model is validated with the experimental results and used to under tension and compression, respectively. The cracking and
study the effect of cutting parameters on cutting forces and chip inelastic strains have been calculated by the following formula:
morphology.
e~ ck el
t ¼ et eot ð3Þ

4.1. Damaged plasticity model

e~ in el
c ¼ ec eoc ð4Þ
The primary mechanism of the chip formation is interlaminar
el
decohesion between the bottom-most layer of the chip and where e ¼ st =E0
ot
topmost layer of the workpiece. The top and bottom portions of
the separated laminae in the chip experience compression and 4.3. Surface based cohesive zone model
tension, respectively, in peeling and vice versa in slipping. Abaqus
Explicits has a damaged plasticity model to capture the loss of As mentioned previously, PyC comprises of multiple laminae
stiffness beyond the elastic regime. Consequently, the progressive stacked together. A surface based cohesive zone model has been
24 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30

Fig. 6. Stress–strain response of pyrolytic carbon under: (a) tension; (b) compression; post failure response of stress and damage v/s (a) cracking strain under tension;
(b) inelastic strain under compression.

traction initiation and propagation can be defined in a variety of ways. In this

study, MAXS (Maximum nominal stress) damage model has been
chosen to model the inter-facial failure stress limit. Damage is
t°n (t°s,t°t)
assumed to initiate when the maximum nominal stress ratio reaches
a value of one. This criterion can be represented as:
( )
/t n S t s t t
max , , ¼1 ð6Þ
t 0n t 0s t 0t

δ°n (δ°s,δ°t) δfn (δfs δft ) separation The symbol /S represents the Macaulay bracket which is used
to signify that a pure compressive deformation or stress state
Fig. 7. Typical traction–separation response for surface based cohesive bonding. does not initiate damage. The following numerical values have
been used in the simulation for cohesive zone modeling:
Knn ¼5500 MPa, Kss ¼2180 MPa, Ktt ¼ 2180 MPa, tn ¼ 9.6 MPa,
used to represent the bonding between the layers. The surface
ts ¼6.9 MPa and tt ¼6.9 MPa [13].
based cohesive zone defines the surface-to-surface interaction which
can be used to model the interlaminar decohesion (peeling, slipping
and delamination) directly in terms of traction versus separation. 4.4. Numerical formulation
The cohesive behavior assumes a linear elastic traction–separation
law prior to damage. It assumes that failure of cohesive bond is The cutting geometry has two parts: the chip comprising of a
characterized by progressive degradation of the stiffness of the bond, stack of laminae above the predefined depth of cut and the bulk
for which it is necessary to define a damage process. The nominal material (the part below the chip) as shown in Fig. 8(a). The bulk
traction stress vector, t, consists of three components: tn, ts and material has a layered structure only for the top few layers and
tt, which represent the normal and two shear tractions, respectively. the remaining portion is considered to be monolithic to improve
dn, ds and dt represent the corresponding separations. the computational efficiency. The mesh elements used in this study
The elastic behavior can be expressed as: were 3D eight noded linear brick reduced integration elements
8 9 2 38 9 (C3D8R). The generalized boundary and loading conditions are
< tn >
> = K nn K ns K nt < > dn >
= shown in Fig. 8(a). The bottom of the workpiece is grounded (i.e.,
6 K st 7
t ¼ t s ¼ 4 K ns K ss 5 ds ¼ K d ð5Þ u¼0, v¼0, w¼0), whereas the tool is given a constant velocity in the
>
:t > ; >
:d > ;
t K nt K st K tt t X direction and all the other translational and rotational degrees of
freedom have been constrained. Note that the cutting velocities used
Typical traction separation response is given in Fig. 7, where t 0n , in the simulations are relatively slow and no significant temperature
t 0s and t 0t represent the peak values of the nominal stress when the rise has been observed experimentally, hence the dependence of the
deformation is either purely normal to the interface or purely in the properties on temperature has not been considered.
first/second shear direction, respectively. dn, ds and dt present corres- Fig. 8(b) shows the interactions among various parts in the
ponding maximum contact separation, if the separation is purely in assembly. Interaction 1 is the fictional contact between the nodes
the normal direction or in the first or second shear direction. Damage of the chip and bulk with the tool rake face. It facilitates formation
 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30
Tool velocity
Depth of cut
Chip layer
Bulk material
Fixed nodes (u=0, v=0, w=0)
Chip layer
Workpiece/ bulk
Fig. 8. (a) Boundary conditions and initial conditions; (b) interaction between nodes and surfaces.
Ch ip
Compression
Tool
Peeling Tension Ploughing/ crushing
Fig. 9. FE simulation of orthogonal micro-machining at 30 mm depth of cut, 5 mm/min cutting speed and 0.3 mm tool width and rake angle: (a) 201; (b)  51.
of curved chip and ploughing/crushing depending upon the rake
angle. Interaction 2 is the cohesive interaction and defined between
the layers. Additionally, the interaction 2 defines the internal friction
between the chip layers. Interaction 2 facilitates decohesion of the
chip layer and prevents the chip to penetrate into the bulk material.
5. Simulation results and model validation
Finite element simulations were performed to capture mechanics
of material removal during orthogonal machining of PyC. The model
is validated against experimental results for cutting forces and chip
morphology. The experimentally measured values were compared
with the simulation results for a range of experimental conditions.
5.1. Simulation results for material separation
Figs. 9(a) and (b) show primary chip formation mechanisms:
peeling and slipping in conjunction with bending of the chip at
positive and negative rake angles, respectively. The positive rake
angle yields long and curly chip whereas the negative rake angle
produces relatively smaller fragmented chips. The positive rake
angle exhibits peeling phenomenon for chip formation as compared
25
Depth of cut
Stacked layers
3
Tool nose
2
1
Reference point
with cutting velocity
Magnified view
Tool
Reference Point
Interaction 1
Interaction 2
Ch ip
Tool
Tension
Peeling Slipping Compression
to predominantly slipping phenomenon observed at negative rake
angles. The crushing of the machined surface below the tool is
observed due to the brittle nature of the work material. The model
also captures the delamination/debonding between layers of the
material in the chip layer if the stresses exceed the cohesive
strength. Figs. 10(a) to (d) show progressive debonding between
the layers in the chip during the machining process. It is clear
that significant debonding occurs in the middle of the chip. The
debonding in the middle is due to bending of the chip as it bending
results in compression and tension at upper and lower portion of
the chip during machining.
The mesh sensitivity of the model has been performed to
check any undesired effect of the element size on the results. Two
different element sizes have been considered for the sensitivity
analysis. The first mesh comprises of coarse elements, 1 mm in
height, 50 mm in width and 50 mm in length and the second mesh
has refined elements of 1 mm in height, 8 mm in width and 8 mm
in length. Fig. 11 shows a comparison of the force predictions for
the two different mesh sizes at 30 mm depth of cut, 300 mm of tool
width, 101 rake angle and 5 mm/min cutting speed. It is clear from
Fig. 11 that both the predictions are within the experimental error
bar. The percentage difference between the two predictions is
limited to 10%. Hence, it can be inferred that the mesh dependency
26 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30

Fig. 10. Progressive debonding between layers.

Experiment
1.6
Sim (1x50x50)
Tool holder
10% Sim (1x8x8)
1.2
9%
Tool
0.8
Workpiece
Workpieceholder
0.4

0
Cutting force Thrust force Dynamometer

Fig. 11. Effect of mesh size on cutting forces.

Fig. 12. Experimental setup for orthogonal machining of PyC.

is not very pronounced. Any further reduction in the element size

will increase the computational time significantly. Consequently, the Minidynes 9256C2), whereas, the chip morphology was studied
mesh comprising of the elements (1 mm  50 mm  50 mm in size) under a scanning electron microscope (SEM) and optical microscope.
have been used in all the simulations.
5.3. Model validation for cutting forces and chip morphology

5.2. Experimental work
The experimental setup of orthogonal micro-machining is shown
in Fig. 12. Cutting operations were performed on the micromilling
machine MIKROTOOLSs. The machine spindle had been replaced by
a tool holder. The tool holder was capable of positioning the tool at
different rake angles. The positioning resolution of the micro-milling
machine is 0.1 mm and the accuracy is 71 mm. TiAlN coated
tungsten carbide (WC) THINBITs grooving tools were used in the
experiment. The process parameters varied in the experiments were
depth of cut, tool width, rake angle and cutting speed. The cutting
forces were measured via a three-component dynamometer (Kistler
The model is validated with the experimental results at different
depths of cut, tool widths, rake angles, and cutting speeds. Figs. 13(a)
and (b) show a comparison of the force signals obtained from the
experiment and the simulation.
Figs. 14(a) through (d) show the model validation at different
cutting conditions. Fig. 14(a) shows the effect of the depth of cut. It
can be seen that the model captures the effect of increase in the
depth of cut (uncut chip thickness). The prediction errors in cutting
forces at 30 mm and 50 mm are 3% and 18%, respectively. Fig. 14(b)
shows that the effect of tool width is not very pronounced and the
experimental values remain almost constant for the tool widths
studied in this experiment. The maximum prediction errors in
 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30 27

1.4 0.8
1.2
1 0.6
0.8
0.4
0.6
0.4 0.2
Cutting force (Experimental) Thrust force (Experimental)
0.2
Cutting force (Simulation) Thrust force (Simulation)
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Fig. 13. Comparison of force signature between simulation and experimental results: (a) cutting force; (b) thrust force at 201 rake angle, 20 mm/min cutting speed, 30 mm
depth of cut and 1 mm tool width.

Fig. 14. Model validation with varying (a) depth of cut of 30 mm and 50 mm at 300 mm tool width, 201 rake angle and 5 mm/min cutting speed; (b) tool width of 300 mm
and 500 mm at 10 mm depth of cut, 201 rake angle and 5 mm/min cutting speed; (c) rake angle of  101 and 201 at 30 mm depth of cut, 300 mm tool width and 5 mm/min
cutting speed; (d) cutting speed of 5 mm/min and 20 mm/min at 30 mm depth of cut, 300 mm tool width and 201 rake angle.
28 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30
Fig. 15. Comparison of chip morphology at positive rake angle (201): (a) simulation; (b) experiment; comparison of chip morphology at negative rake angle (  101):
(c) simulation; (d) experimental.
cutting and thrust forces are less than 15%. The rake angle has
significant effect on the cutting and thrust forces. It is evident from
Fig. 9 that the peeling mechanism occurs at only positive rake angles
whereas significant crushing, slipping and subsurface compression
is observed at negative rake angles resulting in very high thrust
forces. The model captures the chip slipping, crushing and subsur-
face compression and the prediction errors in cutting and thrust
forces at a negative rake angle of  101 are 4% and 16%, respectively.
The prediction errors improve significantly (o6%) at positive rake
angle where peeling is the dominant mechanism of chip formation.
The effect of cutting speed is negligible which is captured by this
model.
Figs. 15(a) and (b) show the chip morphology captured during the
experiment and that predicted via simulation at positive rake
angle (201). Fig. 15(a) shows that long curved chips are obtained
during machining with positive rake angle tools. It is evident from
Figs. 15(a) and (b) that the model captures the peeling and the curled
chip formation. The chip radius is captured within an accuracy of 6%.
Fig. 15(c) and (d) show the chip formation with negative rake angle.
The simulation shows that a significant portion of the chip layer in
front of the tool rake face gets crushed. The experimental results
show fragmented chip formation due to crushing of the material as
predicted by the model. Note that rake angle has a huge influence on
the thrust force due to an increase in the ploughing component with
an increase in the negative rake angle. Since PyC is a brittle material
and cannot flow, ploughing results in crushing of the chip. The
cutting force also increases during transition from positive to negative
rake angle.
6. Parametric study of the effect of process variables
Fig. 16(a)–(d) show the parametric plots of cutting and thrust
forces in the plane of transverse isotropy (AB plane). Fig. 16(a) shows
the effect of depth of cut on the cutting and thrust forces. A 40%
reduction in the cutting force is observed when the depth of cut
changes from 10 mm to 30 mm, followed by an increase of 30% when
the depth of cut transitions from 30 mm to 50 mm. A similar trend is
seen in case of thrust force as well. A reduction of 14% in thrust force
is observed when the depth of cut changes from 10 mm to 30 mm,
and an increase of 41% when the depth of cut increases from 30 mm
to 50 mm. The reduction in cutting forces with an increase in the
depth of cut from 10 mm to 30 mm is counterintuitive and it can be
attributed to the effect of tool edge radius which is about 7 mm.
Fig. 17 shows that the cutting mechanism is different at 10 mm depth
of cut where peeling is absent and the chip slips and gets crushed
instead of peeling which results in higher cutting and thrust forces at
10 mm depth of cut. Fig. 16(b) shows the effect the tool width on the
cutting and the thrust force. The cutting force shows an increase
of 18% as opposed to a decrease of 2% in the case of thrust force.
This can be explained by the peeling phenomena at positive rake
angles where the thrust forces appear due to the reaction from the
surfaces in the absence of ploughing and/or crushing. The reaction
does not change significantly with the tool width and, therefore, the
thrust force also remains fairly constant. Similar behavior has been
observed during micromachining experiments of PyC [1]. Fig. 16(c)
shows the effect of the rake angle on the cutting forces. An increase
of 9% in the cutting force is observed when the rake angle is changed
from 101 to  51. No appreciable change in the cutting force is
observed between  51 and 01. A reduction of 11% and 38% in the
cutting forces are observed if the rake angle changes from 01 to 101
and further from 101 to 201, respectively. Similar behavior has been
observed for the thrust forces also. This can be attributed to the
reduced ploughing at positive rake angles. The chip crushing is
reduced significantly and the peeling phenomenon begins as the rake
angle becomes positive (Figs. 9 and 15 clearly show the chip for-
mation mechanisms at positive and negative rake angles). Fig. 16(d)
shows the effect of the cutting speed on the cutting forces. There is
no appreciable change in the forces with the cutting speed which has
been observed experimentally as well [1].
 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30 29

1.8 2
Cutting forces
1.6
Cutting force Thrust force
1.4 1.5

Force (N)
Thrust force

Force (N)
1.2
1
1
0.8 0.5
0.6
0.4 0
10 30 50 300 500
Depthof cut (µm) Tool width (µm)

1.8 1.8
1.6 1.6
1.4 1.4 Cutting force
Force (N)

Force (N)
1.2 1.2 Thrust force
Cutting force
1 1
Thrust force
0.8 0.8
0.6 0.6
0.4 0.4
-10 -5 0 10 20 5 20
Rake angle (degree) Cutting speed (mm/min)

Fig. 16. Parametric force plots for effect of: (a) depth of cut of 10 mm, 30 mm and 50 mm at 300 mm tool width, 201 rake angle and 5 mm/min cutting speed; (b) tool width
of 300 mm and 500 mm at 10 mm depth of cut, 201 rake angle and 5 mm/min cutting speed; (c) rake angle of  101,  51, 01, 101 and 201 at 30 mm depth of cut, 300 mm tool
width and 5 mm/min cutting speed; (d) cutting speed of 5 mm/min and 20 mm/min at 30 mm depth of cut, 300 mm tool width and 201 rake angle.

Fig. 17. Cutting mechanism at (a) 10 mm depth of cut; (b) 50 mm depth of cut.

Based on the parametric study, the ideal machining conditions
can be established. Note that peeling is the preferred way of
material removal as opposed to slipping and crushing which lead
to poor machined surface. The tool edge radius affects the cutting
mechanism and, if the edge radius is of the same order as the depth
of cut, chip slipping and crushing will occur. Consequently, the
depth of cut (uncut chip thickness) should be sufficiently larger
(more than three times as large) than the tool edge radius. Similar to
the tool edge radius, a positive rake angle is very important to
achieve peeling. Fig. 16(c) clearly shows that a positive rake angle
induces peeling resulting in lower cutting forces and better surface
quality due to reduced crushing.
7. Conclusions
This paper is focused on investigating the mechanics of material
removal in orthogonal micromachining of PyC. PyC is a layered
brittle material, hence, the material behavior and chip formation
mechanisms are different than metals and isotropic materials. The
damaged plasticity model presented in the current work provides
insight into the material behavior. Brittle materials, such as PyC
exhibit degradation of stiffness (progressive damage), cracking in
tension and crushing in compression due to the machining forces.
The surface-based cohesive model helps to understand the under-
lying mechanisms of interlaminar decohesion which can predict
three distinct phenomena: peeling, slipping and debonding which
individually or collectively result in the chip formation. In addi-
tion, the material crushing has also been simulated. It has been
observed that interlaminar decohesion can result in peeling at
positive rake angles and higher depths of cut. At negative rake
angles and uncut chip thicknesses of the order of edge radius, the
observed mode of interlaminar decohesion is slipping. The mate-
rial pile-up in slipping results in crushing and formation of
fragmented chips. A reduction of up to 72% has been observed if
a positive rake angle is used as opposed to a negative rake angle.
The long curly chips obtained due to peeling and absence of
crushing ensures good surface finish with positive rake angles at
30 V. Bajpai, R. Singh / International Journal of Machine Tools & Manufacture 64 (2013) 20–30
higher depths of cut. The current work has helped in enhancing
the understanding the fundamental process mechanics of mate-
rial removal in a layered brittle material.
Acknowledgement
The authors gratefully acknowledge the financial support from
Council of Scientific and Industrial Research (CSIR) [grant number
22(0552)/ 11/EMR-II], Govt. of India.
References
[1] V. Bajpai, R.K. Singh, Orthogonal micro-grooving of anisotropic pyrolytic
carbon, Materials and Manufacturing Processes 26 (12) (2011) 1481–1493.
[2] V. Bajpai, G. Salhotra, R.K. Singh, Micromachining Characterization of Aniso-
tropic Pyrolytic Carbon, Proceedings of the Institution of Mechanical Engi-
neers, Part B: Journal of Engineering Manufacture 225 (9) (2011) 1591–1605.
[3] R.B. Schroeter, R. Kratochvil, J.D.O. Gomes, High-speed finishing milling of
industrial graphite electrodes, Journal of Materials Processing Technology
179 (2006) 128–132.
[4] F. Gu, S.G. Kapoor, R.E. DeVor, P. Bandyopadhyay, An enhanced cutting force
model for face milling with variable cutter feed motion and complex
workpiece geometry, Journal of Manufacturing Science and Engineering 119
(4A) (1997) 467–475.
[5] W.S. Yun, D.W. Cho, An improved method for the determination of 3D cutting
force coefficients and runout parameters in end milling, International Journal
of Advanced Manufacturing Technology 16 (2000) 851–858.
[6] J. Wang, Y. Gong, G. Abba, J.F. Antoine, J. Shi, Chip formation analysis in
micromilling operation, International Journal of Advanced Manufacturing
Technology 45 (2009) 430–447.
[7] L. Ravi, V. Bajpai, R. Singh, S.S. Joshi, Characterization and modeling of Burr
formation in micro-end milling, Precision Engineering 35 (4) (2011) 625–637.
[8] J. Lubliner, J. Oliver, E. Onate, A plastic-damage model for concrete, Interna-
tional Journal of Solids and Structures 25 (1989) 299–329.
[9] J.J. Gebhardt, J.M. Berry Mechanical properties of pyrolytic graphite. Technical
Information Series 66th Annual Meeting The American Ceramic Society
Chicago, Space Science Laboratory, Missile and Space Division 437144, 1964.
[10] W.H. Smith, D.H. Leeds, Modern Materials, 7, Academic Press, New York,
1986 165–80.
[11] M. Sakai, R.C. Bradt, D.B. Fischbach, Fracture toughness anisotropy of a
pyrolytic carbon, Journal of Materials Science 21 (1986) 1491–1501.
[12] J.J. Gebhardt, J.M. Berry, Mechanical failure phenomena in pyrolytic graphite
at room temperature, Journal of The American Ceramic Society 48 (7) (1965)
350–356.
[13] Graphite Store: Product Data Sheet  PG-SN, Available online at /http://
www.graphitestore.com/pop_up_grades.asp?gr_name=PG-SNS, (access on 14/
04/2012).