The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/authorsrights
School of Mechanical and Manufacturing Engineering, The University of New South Wales, NSW 2052, Australia
Department of Machine Intelligence & Systems Engineering, Akita Prefectural University, 84-4 Tsuchiya-ebinokuchi, Yirihonjou, Akita 015-0055, Japan
a r t i c l e
i n f o
Article history:
Received 18 October 2013
Received in revised form 4 December 2013
Accepted 9 December 2013
Available online 22 December 2013
Keywords:
A. Fibres
B. Fracture
B. Wear
C. Finite element analysis
Vibration-assisted cutting
a b s t r a c t
This paper develops an elliptic vibration-assisted (EVA) technique to effectively cut bre-reinforced polymer (FRP) composites using a simple tool. A novel vibrator was invented to work at the anti-resonant frequency to realize stable and high variational velocities. A three-dimensional microstructure-based nite
element model was also established to explore the material removal mechanisms in the EVA cutting. It
was found that the application of vibration can signicantly decrease the cutting forces and reduce the
subsurface damage in a workpiece. The vibration in the cutting direction is more effective in reducing
the cutting force, but that normal to the cutting direction has the advantage of chip removal. When
the vibration is applied to both the directions in the EVA cutting, an optimal cutting process can be
reached, providing much smaller cutting forces, a much improved surface integrity, and an extended tool
life. The study concluded that the ratio of the tool-feed-rate to the maximum vibration velocity in the cutting direction, and the ratio of the cutting distance in a single tool vibration cycle to the bre diameter are
the key parameters. To maximise the advantage of the EVA cutting, it is necessary that these two parameters are below their critical values.
2013 Elsevier Ltd. All rights reserved.
1. Introduction
Fibre-reinforced polymer (FRP) composites have been widely
used in industry due to their high strength and stiffness to weight
ratio. However, machining of FRP composite products is quite difcult, because of the signicant difference in mechanical properties of bres and matrix. As a result, a machined FRPs usually
contains various damages, such as bre pull-out, bre fracture, matrix cracking, brematrix debonding and delamination [14]. To
date, most experimental investigations on the machining of FRP
composites are on the following issues: effect of bre or matrix
types [5,6], inuence of bre volume fraction and orientations
[2,7], role of tool materials and geometries [810], contribution
of the depth of cut [11], and selection of processing parameters
[1214]. However, these studies are limited to traditional machining methods, such as turning, milling, drilling and grinding, and are
still facing the poor surface integrity problems highlighted above.
In order to reveal the machining mechanisms, corresponding theoretical analysis has also been carried out, using various modelling
methods [3,4,1519]. The nite element (FE) analysis has also been
conducted, of which some were based on the consideration of
equivalent homogenous materials [2023] and some others
Corresponding author. Tel.: +61 2 9385 6078.
E-mail address: liangchi.zhang@unsw.edu.au (L.C. Zhang).
0266-3538/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compscitech.2013.12.011
involved the microstructures of FRP composites [2426]. Nevertheless, these are still insufcient to reect the real complex structure
of FRPs, especially in the understanding of the dynamic material
removal process in machining.
On the other hand, it has been a common understanding that
grinding is more appropriate for machining FRP composites
[2729], because in grinding the depth of cut of individual cutting
edges is usually smaller than the diameter of a bre [2]. However,
in many cases, grinding is often inefcient. This raises an important
question: Can a FRP composite be cut at a nominally large depth of
cut but with a small toolcomposite interaction to improve the
surface integrity while using a simple tool?
Vibration-assisted cutting may provide a satisfactory answer to
the above question, because this kind of cutting methods adds a
displacement of a micro-scale amplitude with an ultrasonic frequency to the tip motion of a cutting tool. The process effectiveness
has been experimentally evidenced by the machining of many single phase materials such as metals and ceramics [30,31]. The
advantage is that the ultrasonic vibration alters the tip trajectory
of a tool, which consequently makes the instant depth of cut much
smaller than a bre diameter. This may in turn improve the surface
integrity as pointed out by Zhang and Xu [2,32]. However, the
immediate challenge is as follows: (1) what vibration amplitude,
frequency and tool tip trajectory would be appropriate for a high
performance FRP cutting? (2) how can the material removal
104
xTool t a cos2pft
zTool t b cos2pft w
mx t 2pfa sin2pft m
v
Workpiece
Chip
Tool
b
a X
ap
Ellipticalvibration
Locus of
cutting edge
mz t 2pfb sin2pft w:
Matrix
Fibre
(a)
Fibre
Matrix
Interface
EHM
Generator
Amplifier
C
Tool
A
Y
X
Tool
FPZ
(b)
Fig. 1. Illustration of (a) EVA cutting of FRP composite and (b) its FE micro-scale
model.
Z
FPX
PZT actuator
Y
X
105
100
Mode1
Mode2
Antiresonance point
Cutting tool
Gain dB
80
60
40
Y
Z
Resonance point
Locus of
elliptic vibration
(a)
20
17.2
17.3
17.4
17.5
17.6
Frequency Hz
(b)
four alternating current signals, VA, VB, VC and VD, were applied to
the PZT actuators, respectively. If their frequencies were set at
the same value of f and their amplitudes were Vp, then VA =
Vpsin(2pft), VB = Vpsin(2pft + p), VC = Vpsin(2pft + w) and VD =
Vpsin(2pft + w + p), where the phase shift between VA and VC was
w, that between VA and VB was p, and that between VC and VD
was p. When f is set to be the same as or close to the resonant
or anti-resonant frequencies of the third-bending-mode of the
assembled body (including the actuators, cylindrical body and cutting tool), the vibrator will vibrate in two modes simultaneously.
As a result, the tip of the cutting tool at the end of the vibrator will
vibrates elliptically. It should be mentioned that to obtain elliptic
vibrations in a exible range, the cylindrical body should be supported at the nodal points of its bending vibration modes. To better
balance the cutting force and improve the stiffness of the vibrator
during the cutting process, preloads of FPX in cutting direction and
FPZ in vertical direction were applied at the free end. In addition, a
wave function generator (WF1946B by NF Corporation) and its corresponding ampliers (HSA4052) were used as the power supply
to produce sinusoidal signals with the same frequencies and required phase shifts.
3.2. Performance of the vibrator
To determine the dimensional design of the ultrasonic vibrator,
FE analysis was carried out with the condition of fAB = fCD, where fAB
and fCD are the frequency of the third-bending-mode in the
xy-plane (Mode 1) and yz-plane (Mode 2) respectively. Fig. 3a
shows the results of the frequency response using a PZT device
analysis software (PIEZO by Dynus Co., Ltd.), under the condition
of Vp = 1 V, fAB = fCD = 17.38 kHz and w = 90. Clearly, an elliptic
motion occurred on the tip of the cutting tool. The vibrator was
then manufactured and Fig. 3b shows the resonant frequencies
measured by an impedance analyser (4294A by Agilent Co., Ltd.).
Obviously, the resonant frequencies of the Mode 1 and 2 were
17.386 kHz and 17.382 kHz, respectively. Therefore, the designed
vibrator has met the requirements.
Fig. 3a also shows that the impedances for the two modes
reached their maxima at 17.43 kHz, indicating that the power consumption would be the lowest when the voltages with these frequencies were applied. This is the anti-resonance phenomenon. It
has been a common understanding that in general a high power
is required to obtain high vibration velocities in vibration-assisted
machining. The vibrators in the current manufacturing practice are
mostly arranged to vibrate at the resonant frequencies, leading to a
big electric-mechanical loss, sharp temperature rise and unstable
velocities as sacrice [36]. In contrast, working at the anti-resonant
frequency can not only avoid these problems but also reach the
same high but more stable vibration velocity [33,3739]. The earlier works by the authors have proven that vibrating at the antiresonance frequency outstands that at the resonance frequency
under the high power condition [4043]. In this study, therefore,
the vibrator was set to vibrate at the anti-resonance frequency of
17.43 kHz. In addition, to keep the vibration characteristics unchanged under different cutting forces, preloads on the free end
of the vibrator were set as FPX = FPZ = 30 N according to [41,42,44].
To collect the vibration characteristics of the tool tip, a measuring system consisting of two laser Doppler vibrometers (LV-1610
by Ono Sokki) [44] was constructed. Fig. 4a shows the results under
Vp = 50 V and f = 17.43 kHz. Clearly, the cutting tool vibrates with
sinusoidal patterns along the cutting direction with a = 1.76 lm
and along the vertical direction with b = 1.45 lm when w = 90.
The synthesis of the two vibrations is a typical elliptic vibration
and the tool tip motion follows this pattern. The shape of the ellipse also varies with the change of the phase shift. For instance,
the elliptic pattern is nearly linear when w = 10 but becomes circular when w = 150. The inuence of the voltage amplitude Vp on
the vibration amplitude is shown in Fig. 4b. As can be seen, Vp signicantly affects the vibrations in both the cutting and vertical
directions. The amplitude a increases from 0.69 lm to 2.79 lm
and b increases from 0.55 lm to 2.31 lm linearly as Vp increases
from 10 V to 90 V. This shows an important fact that a and b can
be altered accurately by changing the value of Vp.
For the other types of vibration-assisted cutting, the actuators
A&B (Cases 1) and C&D (Cases 2) were excited separately and
Fig. 5 shows the vibration traces of the tool tip. The vibration
amplitudes were a = 1.95 lm and b = 0.12 lm in Case 1, and
a = 0.08 lm and b = 1.31 lm in Case 2. Since only one thirdbending-mode was excited by applying 180 phase shift voltages
to the parallel actuators, unidirectional vibrations in cutting direction (CDVA cutting) and vertical direction (NDVA cutting) were
generated in Cases 1 and 2, respectively. Consequently, they are
used for the following cutting experiments.
4. Setup and conditions for experimental and numerical
analyses
The workpieces used in the study were unidirectional carbon bre-reinforced polymer (CFRP). The conditions for the simulation
and experimental cutting tests are listed in Table 1. Previous
experiments [2] have shown that when cutting CFRP, both the bre
and matrix exhibit brittle fracture. In the present FE analysis,
therefore, the brittle cracking constitutive model in ABAQUS Explicit was used. Table 2 shows the properties used in the simulation.
106
=90
=10
a
=50
Vibration
direction
1m
1s
1m
1m
=150
=90
=180
1m
1m
1m
(a)
3.5
Amplitude a
Amplitude b
Vibration amplitude m
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0
20
40
60
80
100
Input voltage Vp V
(b)
Fig. 4. Inuence of (a) the phase shift and (b) the voltage amplitude on elliptic vibration.
2m
1m
Fig. 5. Actuator exciting methods and the locus of vibration on the tool tip.
Table 1
Simulation and experimental conditions.
Tool and workpiece
Tool material
Tool clearance angle a ()
Tool rake angle c ()
Cutting edge radius re (lm)
Workpiece material
Fibre orientation h ()
Fibre diameter D (lm)
Fibre volume fraction Vf (%)
Cutting conditions
TiAlN/TiN coated tungsten carbide
7
5
2
Unidirectional CFRP
90
7
60
17.43
2080
90
30
30
10150
0.315
None
Carbon bre
Epoxy
Interface
EHM
q (kg/m3)
Yongs
modulus
E (GPa)
Poissons
ratio m
Tensile
strength
rt (MPa)
Shear
strength
rs (MPa)
1750
1220
1220
1530
230
3.2
3.2
135
0.2
0.35
0.35
0.318
2000
85
50
380
50
25
Density
107
In the FRP cutting process, the fracture of bres not only dominates the cutting force and plays an important role in the chip formation, but also determines the surface nish and subsurface
quality. As such, we will focus on the fracture of bres in our cutting analysis.
between the tool and workpiece was taken as 0.25 based on the
experimental measurements available [2].
In the cutting experiments, a micrograin grade TiAlN/TiN
coated tungsten carbide insert (rake angle 5 and clearance angle
7) was used as the cutting tool, and the workpieces (size:
80 mm 40 mm 3 mm) were prepared from a unidirectional
CFRP laminate plate. The experiments were carried out on a CNC
surface grinder (SGT-315RPA by Nagase Integrecs Co. Ltd.) as
shown in Fig. 6, and the cutting procedure was as follows.
Firstly the vibrator was loaded on the worktable and its position
was adjusted to make the cutting tool in parallel with the
CNC spindle. The workpiece was fastened on the spindle with its
surface to cut to be perpendicular to the cutting tool. No
coolant was used in all the experiments and the bres were
vertical to the cutting direction (i.e., h = 90). To measure the average cutting and vertical forces, a Kistler 3D dynamometer (9256A1,
with 5 kHz natural frequency) was mounted beneath the vibrator
[45,46].
Fig. 7 shows the bre fracture of the second bre under the cutting conditions of ap = 30 lm and v = 1 m/min. When the ultrasonic
vibration was applied, the frequency used was f = 17.43 kHz. With
the CDVA cutting mode, a = 2.07 lm and b = 0 lm; under the
NDVA cutting mode, a = 0 lm and b = 1.67 lm; and under the
EVA cutting mode, a = 2.07 lm and b = 1.67 lm.
Fig. 7a presents the bre fracture in the traditional cutting process. It is clear that the wedging of the cutting edge does not remove the fractured bres effectively, but pushes the broken bre
segments into the zone in front of the cutting edge. This leads to
obvious bre-bending ahead of the cutting zone, brings about severe damages to the neighbouring bres and the matrix material,
and causes the bres to crack even in the deep subsurface.
When a vibration is applied to the feeding direction, i.e., the
CDVA cutting mode shown in Fig. 7b, it can be seen that some
localised deformation zones appear, giving rise to certain smaller
bre fragments. The extent of the bre-bending is reduced, reected by the straightness of the bres. This is because of the following facts: (1) In the CDVA cutting mode, the maximum
vibration velocity in the cutting direction, 2pfa= 22.97 m/min, is
much larger than the tool feed rate, v = 1 m/min; and (2) the cutting distance in each vibration cycle, D = v/f = 0.96 lm, is much
smaller than the bre diameter D. As a result, the vibration breaks
a feed step down to many intermittent cutting actions. However,
bre fracture in the deep subsurface still occurs, caused by the
reciprocating motion of the cutting edge.
When a vibration is applied to the vertical direction, i.e., the
NDVA cutting mode in Fig. 7c, the bre facture becomes localised,
because the tool-bre friction caused by the vertical vibration motion of the cutting edge brings about additional tension and compression cycles in the bre, particularly at the bre-tool contact
surface. This role of such friction-induced tensioncompression is
twofold: (i) it facilitates the localised fracture of the bre by
Fig. 7. Fibre fracture in (a) traditional cutting, (b) CDVA cutting, (c) NDVA cutting and (d) EVA cutting. The colours indicate the levels of Von Mises stresses during the cutting.
(For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
108
Workpiece holder
Workpiece holder
CFRP
CFRP
Chips
Chips
Chips
Tool
Tool Chips
Feed direction
Feed direction
(a)
(b)
Workpiece holder
Workpiece holder
CFRP
CFRP
Chips
Tool
Chips
Chips
Tool Chips
Feed direction
Feed direction
(c)
(d)
Fig. 8. Some snapshots in (a) traditional cutting, (b) CDVA cutting, (c) NDVA cutting and (d) EVA cutting of CFRP composites. The insert at the right bottom of each image is
the chips collected.
Cutting direction
Cutting direction
(a)
(b)
Cutting direction
Cutting direction
(c)
(d)
Fig. 9. Surfaces of CFRP composites machined by (a) traditional cutting, (b) CDVA cutting, (c) NDVA cutting and (d) EVA cutting.
2.0
Simulation: Fx
Simulation: Fz
Experiment: Fx
Experiment: Fz
1.6
1.2
0.8
0.4
0.0
Traditional
CDVA
NDVA
EVA
Cutting methods
Fig. 10. Cutting forces: FE predictions vs. experimentally measured results.
70
Fx: v chang
Fx: a chang
Fz: v chang
Fz: a chang
Cutting force N
60
50
40
30
20
10
0
0
10
20
30
40
50
Rate: 2fa/v
(a)
90
Traditional cutting Fx
Traditional cutting Fz
EVA cutting Fx
EVA cutting Fz
Cutting force N
75
60
30
=D
0.5
1.0
2fa=v
1.5
2.0
2.5
Rate: /D=v/fD
0
45
15
109
12
16
20
(b)
Fig. 11. Inuence of (a) 2pfa/m and (b) D/D (feed rate) on cutting forces.
exerting an additional tensile stress to the bre because the vibration amplitude is very small; and (ii) it can cause the bre to debond or crack vertically because of the compression as shown by
the upsetting phenomenon of the rst bre (from the right) beneath the clearance surface of the tool. On the other hand, the vertical vibration does not change the horizontal wedging force of the
tool tip. As a result, the stresses in the bres ahead of the cutting
zone are greater than those under the CDVA mode.
110
Rake face
50m
50m
50m
Flank face
(a)
(b)
(c)
Fig. 12. Wear of the cutting tool. (a) before cutting; (b) after traditional cutting of 3.5 h and (c) after EVA cutting of 3.5 h.
Acknowledgements
The authors appreciate the Australian Research Council for its
nancial support. This work was supported by an award under
the Merit Allocation Scheme on the NCI National Facility.
References
[1] Davim JP, Rubio JC, Abrao AM. A novel approach based on digital image
analysis to evaluate the delamination factor after drilling composite laminates.
Compos Sci Technol 2007;67(9):193945.
[2] Wang XM, Zhang LC. An experimental investigation into the orthogonal cutting
of unidirectional bre reinforced plastics. Int J Mach Tool Manuf 2003;43(10):
101522.
[3] Zhang LC, Zhang HJ, Wang XM. A force prediction model for cutting
unidirectional brereinforced plastics. Mach Sci Technol 2001;5(3):293305.
[4] Zhang LC. Cutting composites: a discussion on mechanics modelling. J Mater
Process Technol 2009;209(9):454852.
[5] Wang DH, Ramulu M, Arola D. Orthogonal cutting mechanisms of graphite/
epoxy composite 1. unidirectional laminate. Int J Mach Tool Manuf 1995;
35(12):162338.
[6] Koenig W, Wulf C, Grass P, Willerscheid H. Machining of bre reinforced
plastics. CIRP Ann Manuf Technol 1985;34(2):53748.
[7] Hintze W, Hartmann D, Schutte C. Occurrence and propagation of
delamination during the machining of carbon bre reinforced plastics
(CFRPs) an experimental study. Compos Sci Technol 2011;71(15):171926.
[8] Marques AT, Durao LM, Magalhaes AG, Silva JF, Tavares JMRS. Delamination
analysis of carbon bre reinforced laminates: evaluation of a special step drill.
Compos Sci Technol 2009;69(14):237682.
[9] Shyha IS, Aspinwall DK, Soo SL, Bradley S. Drill geometry and operating effects
when cutting small diameter holes in CFRP. Int J Mach Tool Manuf 2009;
49(1213):100814.
[10] Lazar MB, Xirouchakis P. Mechanical load distribution along the main cutting
edges in drilling. J Mater Process Technol 2013;213(2):24560.
[11] Bhatnagar N, Ramakrishnan N, Naik NK, Komanduri R. On the machining of
ber-reinforced plastic (FRP) composite laminates. Int J Mach Tool Manuf
1995;35(5):70116.
[12] Davim JP, Reis P, Antonio CC. Experimental study of drilling glass ber
reinforced plastics (GFRP) manufactured by hand lay-up. Compos Sci Technol
2004;64(2):28997.
[13] Zitoune R, Collombet F, Lachaud F, Piquet R, Pasquet P. Experiment
calculation comparison of the cutting conditions representative of the long
ber composite drilling phase. Compos Sci Technol 2005;65(34):45566.
[14] Sardinas RQ, Reis P, Davim JP. Multi-objective optimization of cutting
parameters for drilling laminate composite materials by using genetic
algorithms. Compos Sci Technol 2006;66(15):30838.
[15] Iliescu D, Gehin D, Iordanoff I, Girot F, Gutierrez ME. A discrete element
method for the simulation of CFRP cutting. Compos Sci Technol 2010;70(1):
7380.
111