You are on page 1of 13

Composite Structures 220 (2019) 460–472

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

A study of energy dissipating mechanisms in orthogonal cutting of UD-CFRP T


composites
Xiaoye Yana, Johannes Reinerb, , Mattia Baccac, Yusuf Altintasc, Reza Vazirib

a
School of Mechanical Engineering, Northwestern Polytechnical University, Xi’an 710072, China
b
Composites Research Network, Departments of Civil Engineering and Materials Engineering, The University of British Columbia, Vancouver, BC, Canada
c
Department of Mechanical Engineering, The University of British Columbia, Vancouver, BC, Canada

ARTICLE INFO ABSTRACT

Keywords: This paper presents a three-dimensional thermo-mechanical finite element (FE) model on the micro-scale for the
Micro-scale damage model orthogonal cutting simulation of unidirectional carbon fiber reinforced polymer (UD-CFRP) materials. Fiber,
Cutting mechanisms matrix and the fiber-matrix interface are modeled separately to simulate failure mechanisms associated with
Orthogonal cutting tests fiber breakage, matrix cracking and fiber-matrix debonding. Experiments on orthogonal cutting of UD-CFRP
Explicit finite element analysis
workpieces with various fiber orientations have been conducted and compared to the micro-scale FE simulations.
UD composites
The good correlation between the experimental and numerical results with respect to cutting forces, chip for-
mation and surface quality enables for an in-depth FE energy analysis to quantify dominating failure mechan-
isms in different fiber orientations. In addition, the contribution of each fiber failure mode during orthogonal
cutting is evaluated. It is found that dissipated energies associated with various failure modes and friction vary
with the fiber orientations and that the simulation provides insight into the characteristic fiber orientation-
dependent quantities observed in the experiments such as chip formation and surface morphology. Finally, a
sensitivity analysis is carried out to assess the variation of the simulated cutting response with respect to input
parameters with high uncertainty including friction coefficients or fiber strength values.

1. Introduction unidirectional carbon fiber reinforced polymer (UD-CFRP) for a better


understanding of the machining process. The term “orthogonal cutting”
Machining processes such as drilling and milling are often used to refers to the case where the cutting edge of the tool is perpendicular to
shape FRP parts with required tolerances and surface finish [1]. The the direction of relative tool-workpiece motion. Koplev et al. [7] pre-
heterogeneous and anisotropic nature of FRPs renders the mechanics of sent one of the first investigations on orthogonal cutting of UD-CFRP.
material removal different from and more complex than that of ma- They observed significant fractures due to the tool pressing on the
chining homogeneous and isotropic materials such as metals [2]. The composite material. The depth of cracks and surface quality varied
machining process of the majority of homogeneous and ductile mate- depending on whether the composite was machined parallel or per-
rials is characterized by plastic deformation with continuous chip for- pendicular to its fiber direction. Arola et al. [8] investigated the effect
mation [3], while the machining of quasi-brittle FRPs involves a com- of fiber orientation on chip formation and surface morphology during
plex combination of various modes of damage such as fiber breakage, orthogonal cutting of UD composites. Fig. 2 illustrates the definition of
matrix cracking and fiber-matrix debonding [4], leading to severe de- fiber orientation relative to the cutting tool. The study showed that
lamination and surface damage. In order to achieve the desired surface discontinuous chips developed from a series of brittle fractures. The
quality with machining processes, it is necessary to understand the removal mechanism at 0° fiber orientation was micro-buckling and
complete mechanism of material removal and predict the process states fracture in shear, while compression-induced shear failure in transverse
such as cutting forces and surface morphology [5,6]. direction occurred for fiber orientations greater than 0° . Bhatnagar
et al. [9] observed the chip formation and cutting forces in orthogonal
1.1. Experimental studies cutting of UD-CFRP and concluded that with fiber orientation < 90°,
the fiber pullout phenomenon occurred due to axial tension in the fi-
Various studies have been reported on orthogonal cutting of bers, while in the case of > 90° , the fibers mostly broke in shear.


Corresponding author.
E-mail address: Hannes.Reiner@composites.ubc.ca (J. Reiner).

https://doi.org/10.1016/j.compstruct.2019.03.090
Received 14 December 2018; Received in revised form 15 February 2019; Accepted 25 March 2019
Available online 27 March 2019
0263-8223/ © 2019 Elsevier Ltd. All rights reserved.
X. Yan, et al. Composite Structures 220 (2019) 460–472

Similarly, Wang and Zhang [10] illustrated the variation of deformation fiber-matrix debonding. DDM models damage directly at the location of
mechanism with respect to fiber orientation by analyzing the subsur- failure without physically changing the constitutive properties, al-
face damage in orthogonal cutting. It was concluded that with the or- though they lead to higher computational costs compared to CDM. Rao
ientation < 90° the fiber was under axial tension and fractured easily, et al. [17,25] implemented zero-thickness cohesive elements to simu-
resulting in comparatively better surface quality. In contrast, fibers with late the orthogonal cutting of FRP composites. It was concluded that
> 90° were pushed outwards and underwent large bending with the fiber failure is caused by a combination of crushing and bending, and
movement of the tool, leading to severe fiber-matrix debonding and sub-surface damage is due to matrix failure and interfacial debonding.
poor surface quality. Diaz and Axinte [11] monitored the cutting be- Abena et al. [26] proposed an approach for interfacial debonding in-
havior during orthogonal cutting along longitudinal and transverse fi- corporating zero-thickness cohesive elements based on a traction-se-
bers, and results showed that the fracture mechanism changed along paration law, and conducted simulations with different interface
the depth of cut. An et al. [2] used UD-CFRP samples with a wide range modeling methods. It was concluded that the chip formation me-
of varying between 0° and 180° with an increment of 10° to study the chanism could be changed if the fiber-matrix interface was modeled as
effect of fiber orientation and depth of cut on cutting forces. It was finite-thickness, and zero-thickness cohesive elements based on the
found that for < 90°, cutting forces increased linearly with , while traction-separation law were comparatively better in representing the
maintaining relatively low levels of forces for > 90° . Also, with in- interface debonding.
creasing cutting depth, cutting forces exhibited upward trends. Despite the efforts mentioned above in investigating the machining
processes in FRPs, to the best knowledge of the authors, a compre-
1.2. Finite element studies hensive model that quantitatively reveals the relation between the
cutting mechanisms and the cutting forces is lacking. The present work
There is a wide range of finite element (FE) models that simulate in aims to develop a 3D micro-scale FE model consisting of fiber, matrix
situ stress distribution, fracture modes, damage initiation and evolu- and cohesive elements at the fiber-matrix interface, and to conduct a
tion, as well as sub-surface damage in machining UD-CFRP composites. detailed energy analysis to determine the competing failure mechan-
Reiner and Vaziri [12] recently published a comprehensive overview of isms during the orthogonal cutting of UD-CFRP with different fiber
existing FE methods and implementation strategies for the structural orientations. For validation purposes, we performed orthogonal cutting
analysis of composites. Essentially, there are continuous and discrete tests and compared the measured results to the predictions of the FEA.
simulation strategies or a combination of the two.
A continuous 3D macro-scale equivalent homogeneous model 2. Experimental methodology
(EHM) was developed by Wang et al. [13] during CFRP cutting where
Hashin’s failure criteria indicate damage initiation in longitudinal To study the cutting mechanism of CFRP, we performed orthogonal
(fiber) and transverse (matrix) directions. Based on the meso-scale cutting tests to investigate the machining response in terms of cutting
model, Lasri et al. [14] investigated the effects of various failure criteria forces and chip formation. Experimental results will also be used to
(Hashin, maximum stress and Hoffman criteria) on chip formation and validate the developed finite element model.
subsurface damage. Results showed that fiber-matrix debonding and Aerospace-grade UD prepreg (Hexcel AS4 carbon fiber impregnated
fiber breakage were the first and ultimate failure mechanisms devel- with Cytec MTM45-1 epoxy matrix) with high fiber volume fraction
oped during the CFRP cutting process, respectively. In order to study (Vf = 58.72%) was used in this study. The UD-CFRP laminate was made
failure of each constituent separately, Nayak et al. [15] first proposed a up of 32 prepreg plies (all 0° direction with 0.14 mm ply thickness),
micro-scale FE model consisting of a single fiber and surrounding ma- resulting in a total thickness of 4.48 mm. The laminate was cured in an
trix. A stress distribution analysis showed that for all fiber orientations, oven with a vacuum bag according to [27]. Afterwards, the laminate
the fiber broke perpendicular to its axis due to the critical maximum was water-jet cut into workpieces with dimensions of
principal stresses acting along the longitudinal direction. 50 mm × 40 mm × 4.48 mm and desired fiber orientation angles of
Elastoplastic matrix damage was considered by Song et al. [16] and 0°, 30°, 45°, 60°, 90°, 135° and 150° for orthogonal cutting experiments.
Rao et al. [17], and Xu et al. [18] modeled matrix as a brittle material in Fig. 1 shows the experimental setup. We conducted the orthogonal
ABAQUS/Explicit. Results show that the combination of the con- cutting experiments on a vertical CNC machining center. The rotation
stitutive models are capable of predicting matrix cracking and fiber movement of the spindle shaft was constrained and cutting tool
breakage in orthogonal cutting processes. He et al. [19] developed a 3D mounted on it feeds in the horizontal direction with the turning insert
continuum macro-scale model for the simulation of orthogonal cutting
of UD-CFRP. The authors found significant differences between ex-
perimental and predicted thrust forces and sub-surface damage. The
poor predictive capabilities were attributed to fiber-matrix debonding
that was disregarded in the simulations. In order to model the pro-
gressive separation of the fiber-matrix interface, Calzada et al. [20]
adopted continuum interfacial elements and predicted the failure me-
chanisms and chip formation during 2D orthogonal cutting. Xu et al.
[21,22] implemented finite-thickness continuum elements in their 3D
micro-scale model to define the fiber-matrix interface and characterized
the machined surface profiles to reveal fiber-matrix debonding. Gao
et al. [23] and Cheng et al. [24] also proposed 3D micro-scale models
using finite-thickness continuum elements and effects of thermal-me-
chanical coupling during orthogonal cutting of UD-CFRP. Models
mentioned above use Continuum Damage Mechanics (CDM) to re-
present fiber/matrix failure and interfacial debonding by means of local
stiffness reduction.
Meanwhile, discrete damage mechanics (DDM), such as the cohe- Fig. 1. (a) Close-up view of the cutting zone and (b) schematic view of the
sive zone model (CZM), is commonly used to predict delamination or experimental setup for the orthogonal cutting tests.

461
X. Yan, et al. Composite Structures 220 (2019) 460–472

Table 1 debonding cannot be predicted by purely continuum FE models [17],


Parameters used in the experiments. we applied a combined continuum-discrete three-dimensional micro-
Parameter Values scale FE model including fiber, matrix and cohesive elements at the
fiber-matrix interface.
UD-CFRP composites Workpiece 50 mm × 40 mm × 4.48 mm Fig. 2(a) shows the overall numerical model. The cutting tool is
dimensions
modeled according to the turning inserts used in the experiments with
Tool geometry Rake angle ( ) 10°
rake angle , relief angle and tool radius r. The tool is treated as a
Clearance angle ( ) 11°
Edge radius (r ) 15 µ m rigid body with a reference point RP to control its displacement. The
Process parameters Depth of cut (h ) 20–50 µ m global reference system is indicated by x , y and z axes, and the ma-
Fiber orientation 0°, 30°, 45°, 60°, 90°, 135° and 150° terial (local) directions by 1, 2 and 3 axes. The workpiece has a re-
( ) presentative geometric dimension of 100 µm × 16 µm × 80 µm
Cutting speed ( ) 3 m/min (length × width × height ). The fiber orientation is defined as the
angle between its longitudinal direction (direction 1) and the direc-
tion of cutting speed . The degrees of freedom at the bottom and left-
hand side of the workpiece are fully constrained. Fig. 2(b) shows a 3D
cutting the workpiece. Carbide turning inserts (ISO Code: TPMT-221) representative volume element (RVE) consisting of a fiber (diameter
were used for the experiments. The cutting edge effective length was D = 7 µ m [29]), the surrounding matrix (solid elements C3D8R with
larger than the thickness of the workpiece to comply with the as- enhanced hourglass control) and the fiber-matrix interface modeled
sumption of orthogonal cutting conditions [28]. Table 1 lists the tool with zero-thickness cohesive elements (COH3D8, 3D fully-integrated)
geometry (rake angle , relief angle and tool radius r) and the cutting having shared nodes with fiber and matrix. The workpiece has been
parameters (depth of cut h and cutting speed ), and Fig. 2 contains a meshed with a global element size of 0.8 µm × 0.8 µm × 0.8 µm , and
detailed illustration of the different parameters. We used a three-com- the cutting tool with a slightly coarser discretization.
ponent dynamometer (Kistler 9257B) together with a charge amplifier To model the mechanical interaction between the tool and the
(Kistler 5010) to obtain the cutting forces at an acquisition rate of workpiece, we apply the ABAQUS built-in “penalty contact” formula-
10 kHz. Note that we applied a relatively low cutting speed of 3 m/min. tion involving Coulomb friction. The penalty contact is used for all the
A range of cutting speeds from 1 m/min up to 10 m/min was in- elements within the workpiece to avoid inter-element penetration,
vestigated experimentally with no effect on the resulting cutting forces. which, by default, are not considered until element deletion occurs.
We repeated the experiments for each fiber orientation and for each Coefficients of friction between the tool and the workpiece µtw , and
depth of cut three times, and the results presented in this paper are between the fiber-matrix µ fm are set as 0.2 [30,31] and 0.5 [32], re-
averaged values. The maximum standard deviation in the experimen- spectively, in all fiber orientations. A sensitivity analysis at the end of
tally measured cutting forces was 4.69 N/mm corresponding to the case this study will investigate the effect of the two friction coefficients.
involving 135° fiber orientation and a depth of cut of 40 µ m for which
the mean value of the measured force was 16.42 N/mm. This reflects 3.2. Material modeling
reasonable consistency of the test data for the purposes of validating the
numerical model predictions. Fig. 5 shows the corresponding error bars A discrete-continuum model is used to analyze the microscopic
in the measured cutting force for each fiber orientation and depth of cut damage mechanisms in CFRP during orthogonal cutting. Continuum
that was tested. After the cutting experiments, we measured the damage mechanics describes the progressive softening in the fibers and
roughness of the machined surfaces in the direction of the cutting speed the matrix. The discrete cohesive zone method (CZM) is adopted to
using a Mitutoyo Surftest SV-500 profilometer according to Interna- model traction-separation behaviour at the fiber-matrix interface. Each
tional Standard ISO 4288:1996. failure mode is described in the following subsections.

3. Finite element modeling 3.2.1. Continuum damage model for carbon fiber
According to HexTow ®AS4 product data sheet [29] and experi-
3.1. Description of the FE model mental investigations in [33], carbon fiber is assumed to be a linearly
elastic and transversely isotropic material. The carbon fiber fails when
We used the commercial finite element analysis software ABAQUS/ the elemental stress exceeds the corresponding strength in each direc-
Explicit 2016 to simulate the orthogonal cutting behavior of UD-CFRP tion. This offers simplification and has proven to successfully represent
composites. Since damage such as matrix cracking or fiber-matrix the mechanism of fiber fracture observed experimentally [14,24,34].

Fig. 2. Schematic view of (a) the overall numerical model and (b) micro-scale modeling for different fiber orientations.

462
X. Yan, et al. Composite Structures 220 (2019) 460–472

Within the adopted maximum stress criterion, damage is described by follows

Tensile loading in longitudinal direction ( 11 0) dt1f = 11


Xt
, Unew = Uold + U / f,
(2)

Compressive loading in longitudinal direction dcf1 = 11


, where
Xc
( 11 < 0) 3 old
+ new
avg avg ij ij
Tensile loading in transverse direction ( dt 2f = 22
, dt3f = 33
, U= ij ij with ij = .
22 Yt Yt i, j = 1
2 (3)
0or 33 0)
In the above equations ij (i, j = 1, 2, 3) are increments of normal and
Compressive loading in transverse direction dcf2 = 22
, dcf3 = 33
,
Yc Yc shear components of the strain tensor. ijold and ijnew (i, j = 1, 2, 3) are
( 22 < 0or 33 < 0) the elemental stresses at the start and end of an increment, respectively.
Shear loading (in-plane) dsf12 ± = 12
( 12 > 0or < 0) f is the mass density of the fiber. U represents an increment of elastic
S

Shear loading (out-of-plane) dsf23 = max{ 13


, 23
} strain energy density, as shown with the shaded area in Fig. 3(a).
S S

(1)
3.2.2. Strain-rate and temperature-dependent continuum damage model for
where dt1f , dcf1, dt f2, and
dcf2, are the damage variables
dt f3, dcf3, dsf12± dsf23 epoxy matrix
that control damage initiation. The superscript “f” in the damage The constitutive behaviour of the epoxy matrix is assumed to be
variables denotes fiber, and the subscripts “c”, “t” and “s” denote isotropic and elastoplastic. The local strain rates in the CFRP compo-
compression, tension and shear modes of loading, respectively. ii and ij sites were obtained here from the FE models by the field variable ERV
(i , j = 1, 2, 3) are the normal and shear stress components of the fiber. in ABAQUS, and the results show that these strain rates are typically in
Xt and Xc are the longitudinal strengths in tension and compression; Yt the range of 10 2 to 106 s 1. The temperature variation during orthogonal
and Yc are the transverse strengths in tension and compression; and S is cutting of UD-CFRP is larger than 55 °C according to [23,35], which
the shear strength. Table 2 lists the corresponding input values. affects the mechanical properties of the matrix [36]. Therefore, the
Fig. 3(a) shows a typical stress-strain curve of a single fiber element effects of strain rate and temperature are also considered for the epoxy
under longitudinal uniaxial tension and compression. When the ele- matrix.
mental stress 11 reaches the tensile strength Xt , the value of the damage Fig. 3(b) shows the stress-strain curve under uniaxial tension and
variable dt1f in Eq. (1) becomes unity representing a fully damaged state. compression. The linear elastic phase is defined by the elastic modulus
At this point 11 reduces to zero instantaneously, and the element is Em . When the elemental stress reaches the yield strength ym0 , the built-
deleted. The same behavior applies to all other stress components in the in Johnson-Cook constitutive model is utilized to describe the yielding
fiber. It is more physically realistic to model compressive damage as a behavior of the matrix. Johnson-Cook parameters are determined from
gradual process, but this prolongs computational time. Therefore, in an empirical curve fit with data from experimental work [36,37] and
this study the compression failure is assumed to be more sudden. listed in Table 2. The strain rate in [36,37] is in the range of 683 s 1 to
The dissipation energy due to fiber damage corresponds to the 1890 s 1, and the temperature varies from 20 °C to 130 °C. Fig. 3(b)
shaded area in Fig. 3(a) and is calculated within the user subroutine shows the stress-strain responses with two levels of temperature (20 °C
VUMAT from the elastic strain energy of the fiber element under the and 80 °C) and strain rates (683 s 1 and 1513 s 1). It is noted that ma-
condition that any of the damage modes defined in Eq. (1) is initiated. terial properties of matrix are based on a similar epoxy LY556, since
The specific elastic strain energy Unew is calculated incrementally as those of MTM45-1 are not available in the literature.
In the plastic regime, the elastic modulus remains undamaged.
Damage initiates when the equivalent plastic strain is greater than fpl ;
Table 2 after that point, Em degrades linearly until damage saturates at an
Mechanical properties of the constituents of the UD-CFRP laminates. equivalent plastic strain of fm , when the matrix completely loses its
load-bearing capacity. At this instant, the fracture energy density cor-
Material Property Values
responding to the shaded area in Fig. 3(b), gcm , multiplied by the
AS4 Fiber Elastic constants E11 = 231 GPa [29], E22 = 15 GPa [42], characteristic length of the element, l (calculated by ABAQUS/Explicit),
12 = 0.2 [42], G12 = 15 GPa [42], is equal to the fracture energy Gcm (i.e., energy per unit area of advan-
G23 = 7 GPa [42] cing crack), which is listed in Table 2. Matrix fracture energy Gcm and
Tensile strength Xt = 4620 MPa [29], Yt = 1500 MPa ∗
dissipated plastic work are obtained by the history variables ALLDMD
Compressive Xc = 3960 MPa [42], Yc = 3340 MPa [33]
strength
and ALLPD, respectively.
Shear strength S = 1500 MPa ∗∗ [43] As for the effect of strain rate on the elastic modulus and the ulti-
MTM45-1 Elastic constants Em = 3.35 GPa [42], m = 0.35 [42] mate strength of CFRP, Hsiao et al. [38] conducted experiments with
Epoxy UD-CFRP specimens under high strain rate loading. Results showed that
Yield strength
the transverse strength and the elastic modulus of UD-CFRP are gov-
pl
m
y0 = 120 MPa [42], f = 5% [42]
Johnson-Cook A = 120 MPa [42], B = 654.18 MPa [36], erned by the matrix and show a nearly twofold increase over the quasi-
Param∗∗∗ n = 0.772 [36], c = 0.124 [36], m = 0.304 static values. Daniel and Liber [39] concluded from their experimental
[37]
Fracture energy
study that the material properties in the longitudinal (fiber) direction
Gcm = 0.1 N/mm[44]
Interface Normal strength tn = 50 MPa [40,45] are dominated by fibers and do not vary from the static values.
Shear strength ts = 75 MPa [40,45] Therefore, only the effect of strain rate on the properties of the matrix is
Elastic stiffness K = 100, 000 N/mm3 [40] considered here. ym0 and Em used in the FE models are doubled for
Fracture energy G IC = 0.002 N/mm [45,46] simplicity compared to the corresponding properties in Table 2, while
G IIC = G III
C
= 0.006 N/mm [45,46] no scaling is applied to the fiber properties.
Mixed-mode = 1.45 [45]
parameter
3.2.3. Discrete damage model for fiber-matrix interface

Assumed. The interface between the fiber and the matrix has been modeled
∗∗
The shear strength of different FRPs is reported to be 20–40% of the with zero-thickness cohesive elements (COH3D8, 3D fully-integrated),
tensile strength, and 30% is used here. the behavior of which is defined in terms of a traction-separation law.
∗∗∗
The Johnson-Cook parameters are calibrated based on LY556 epoxy. Fig. 4 shows the constitutive relation under pure tension (Mode I) or

463
X. Yan, et al. Composite Structures 220 (2019) 460–472

Fig. 3. Assumed stress-strain response of: (a) carbon fiber under uniaxial tension and compression in the longitudinal direction and (b) epoxy matrix under uniaxial
tension and compression at various temperature and strain rates.

pure shear loading (Mode II). The initial response is assumed to be Mixed-mode progressive damage is defined based on the interaction
linear with the slope representing the elastic stiffness K. A high value of of fracture energies using the Benzeggagh-Kenane (BK) fracture cri-
K = 105 N/mm3 [40] is applied in all three directions in order to prevent terion [41] given by
penetration of the interfaces. The shaded areas under the traction-se-
paration curve in Fig. 4 refer to GIC , GIIC and GIII
C
, fracture energies in the GIC + GIIC GIC
GS
= GC
normal (Mode I) and the two local shear directions (Mode II and III), GT (5)
respectively. The energy dissipation due to interfacial damage is ob-
tained by the history variables ALLDMD in ABAQUS. where GS = G II + G III and GT = G I + G II + G III . is a mixed-mode
Cohesive damage is assumed to initiate when parameter. Table 2 and Table 3 show the material properties of all
constituents used in the micro-scale FE model.
2 2 2
tn ts tt
+ + =1
tn0 ts0 tt0 (4) 4. Results and discussion

where tn0, ts0 and tt0 denote the normal strength and the two shear It has been reported [2,9] that cutting forces are influenced by
strengths, respectively. tn, ts and tt are the nominal traction vectors. The different failure mechanisms with respect to the fiber orientation .
use of Macaulay bracket in ts indicates that a pure normal compressive However, it is not possible to directly quantify these failure mechanisms
stress does not initiate damage, as shown in Fig. 4(a). experimentally. In this section, simulated cutting forces are compared

Fig. 4. The traction-separation response of a zero-thickness cohesive element under (a) normal and (b) shear loading.

464
X. Yan, et al. Composite Structures 220 (2019) 460–472

Table 3
Thermal properties of UD-CFRP laminates.
Material Property Values

AS4 Fiber Longitudinal thermal conductivity l = 9.37 mJ/(mm·s·°C) [23]


Transverse thermal conductivity t = 4 mJ/(mm·s·°C) [23]
Longitudinal thermal expansion l = 0.7 × 10 6 /°C [42]
Transverse thermal expansion t = 12 × 10 6/ /°C [42]
Specific heat Cf = 7.94 × 10 8 mJ/(t·°C) [42]

MTM45-1 Epoxy Thermal conductivity m = 0.36 mJ/(mm·s·°C) [47]


Thermal expansion m = 5.8 × 10 5 /°C [42]
Specific heat Cm = 1.31 × 109 mJ/(t·°C) [47]

to experimental results to validate the proposed numerical model. We Fig. 6(a) shows the predicted chip formation at = 0°. The cracks
then report a thorough energy analysis to provide further insight into propagate along the fiber-matrix interface and the fibers bend as the
the failure mechanisms and their contribution towards energy dissipa- tool penetrates into the workpiece. After fiber fracture, the chip forms
tion and their influence on the cutting forces. and slides along the tool rake surface. This process has been captured by
an imaging system in [11] and is referred to as peel-up. At the fiber
4.1. Cutting forces orientation of = 45° in Fig. 6(b), fiber breakage occurs due to a
crushing-dominated failure by tool indentation, and the formed chip
Fig. 5(a) and (b) show a comparison between the simulated and releases from the machined surface along the longitudinal fiber direc-
measured cutting forces Fc versus the fiber orientation for two depths tion. This mechanism was also observed experimentally in graphite/
of cut, h = 40 µ m and 50 µ m, respectively. All the experiments have epoxy composites [8]. For the case when = 90° shown in Fig. 6(c), the
been conducted at the same constant cutting speed (v = 3 m/min) and fibers bend along their axes because of the excessive transverse com-
tool geometry ( = 10°, = 11°) with three repeats for each fiber or- pression in the contact region. Chips are in the form of small dust
ientation and for each depth of cut h. The following analysis shows the [8,11]. The case at = 135° is shown in Fig. 6(d). The tool rake face
average values of the three repeated experiments. Thermal strains are contacts with the workpiece, and long discontinuous chips are formed
introduced by the thermo-mechanical FE model considering the gen- due to fiber fracture in bending. Similar results were reported in ex-
erated heat during cutting can affect the stress-strain relationship of the perimental studies [10].
composite and thus the final cutting forces as discussed later. It can be To evaluate the surface quality of the machined surfaces through the
seen from Fig. 5 that the proposed model is able to capture the overall cutting distance, a simple surface texture parameter Rt is calculated
trend of the cutting forces for both depths of cut. It can be concluded from the roughness measurements at fiber orientations of 0°, 45°, 90°
from both the measured and predicted results that for fiber orientation and 135°. Rt is defined as the sum of the largest profile peak height and
in the range 0° < < 90° , the cutting force increases with the rise of the largest profile valley height within the evaluation length, and thus
the fiber orientation and reaches the peak point at = 90°, while in the can describe the maximum sub-surface damage of the CFRP workpieces.
range 90° < < 180°, the cutting force shows a decreasing trend until it Experimental measurements here show that at = 0°, the CFRP
reaches the minimum value at = 150°. The results are supported by workpiece has the lowest surface roughness, 3.9 µm Rt, = 0° 5.7 µm .
previous findings [2,17]. The surface roughness at = 45° and = 90° are slightly higher in the
range 7.2 µm Rt, = 45° 10.3 µm and 10.0 µm Rt, = 90° 20.5 µm , re-
spectively. When = 135°, machining results in poor surface quality,
4.2. Chip formation and surface quality and the value of Rt is greater than 50.0 µ m. The maximum depths of
subsurface damage have also been measured from the simulations with
The mechanism of chip formation changes for different fiber or- respect to fiber orientation with a depth of cut h = 50 µ m. It can be
ientations. This has been analyzed in [8,11] by using an imaging observed from Fig. 7 that the trend of simulated results agrees well with
system, which provides a method to observe the material removal the experimental measurements.
mechanism. The numerical simulation of chip formation in Fig. 6 agrees It has been reported [9,17] that failure mechanisms for different
qualitatively with experimental observations and is discussed in more fiber orientations vary, and so do the cutting forces, chip formation
detail as follows.

Fig. 5. Comparison between simulated and experimental cutting forces with depth of cut (a) h = 40 µ m, (b) h = 50 µ m. Error bars indicate the variation of the
measured forces in the experiments.

465
X. Yan, et al. Composite Structures 220 (2019) 460–472

Fig. 6. Predicted chip formation at selected times during the cutting process for fiber orientations (a) = 0°, (b) = 45°, (c) = 90°, (d) = 135°.

during cutting of an elastoplastic material. Inspired by these studies, we


analyze the energy dissipating mechanisms during orthogonal cutting
of UD-CFRP. Fig. 8 shows the dissipated mechanisms at depth of cut
h = 50 µ m for different fiber orientations . Fracture energy (fiber da-
mage, matrix damage and fiber-matrix interfacial damage), plastic
dissipation energy of the matrix, as well as frictional energy between
the tool and the workpiece Etw and between fiber and matrix Efm are
shown in the bar charts. Other forms of energy such as kinetic energy,
viscous dissipation energy and other artificial energy introduced in
ABAQUS prove to be either negligible or unrelated to fiber orientation,
and thus are not discussed here.
Percentages in Fig. 8(a) refer to the ratio of dissipated energies due
to individual mechanisms to the overall dissipated energy for cutting at
a given fiber orientation. Fig. 8(b) compares the total dissipated en-
Fig. 7. Comparison between simulated maximum sub-surface damage depth
ergies between different fiber orientations. Energy dissipated by fiber
and measured roughness Rt at fiber orientations = 0°, 45°, 90° and 135°. damage is calculated using Eq. 2 and implemented in the VUMAT. All
other components of energy are computed and reported by built-in
features of ABAQUS.
and surface quality. However, it is not possible to quantify these failure
It can be observed from Fig. 8(b) that the value of the overall dis-
mechanisms in the experiments. By comparing the simulated results
sipated energy has the same trend as the predicted forces shown in
with experimental data with regards to cutting forces in Fig. 5 and the
Fig. 5(b) where the maximum value is at = 90° and the minimum at
surface quality in Fig. 7, it can be concluded that the FE model im-
= 135°. This can be explained by the external work Wtool Fc . There-
plemented in this study is sufficiently accurate to provide further in-
fore, the more energy dissipated through the cutting process, the larger
sight into the dominating failure mechanisms, which will be discussed
the cutting forces Fc . When < 90° , fiber damage dissipates more en-
in the following sections.
ergy compared to interfacial damage, matrix damage and plastic dis-
sipation while matrix failure is the dominating energy when = 135° as
4.3. Energy analysis evident by comparing the percentages of each mechanism in Fig. 8(a).
This is due to the fibers failing in shear and, since their resistance to
Reiner et al. [48] conducted a complete energy analysis in the nu- shear is lower than that to normal stresses, the energy dissipation is
merical simulation of drop-weight impact tests in order to quantify minimal. The percentage of each individual dissipated energy at
energy dissipating mechanisms in hybrid titanium-composite materials. = 45° is quantitatively similar to that at = 90°. When 90°, the
Similarly, McGregor et al. [49] investigated the energy absorbing percentage of fiber damage energy and matrix damage energy in Fig. 8
qualities of braided composites under axial crushing. Williams and (a) increase when grows. The fracture energy for a single element is
Patel [50] provided a model for the energy balance involving external constant, this implies that element deletion of the fiber and the matrix
work, fracture energy, frictional energy and plastic dissipation energy occurs more often for the cases with larger , as supported by

466
X. Yan, et al. Composite Structures 220 (2019) 460–472

Fig. 8. Energy dissipated through material damage, friction and plastic deformation: (a) dissipated energy divided by the corresponding total dissipated energy, (b)
dissipated energy divided by the total dissipated energy at 90°.

experimental observations. Indeed at = 0° long chips are formed from fiber damage is more pronounced in the cases of = 45° and = 90°.
peel-up while chips are powder dust at = 90° [8,11]. As shown in Fig. 8(a), for all fiber orientations, the percentage of
Temperature distribution of CFRP in the cutting region is obtained frictional energy is one of the highest, which means that the effect of
by analyzing the field output variable TEMP in ABAQUS for the fiber friction cannot be neglected. However, to the extent of the authors’
and matrix material. Fig. 9 shows the maximum cutting temperature at knowledge, there is no data available to accurately describe the contact
the four fiber orientations with depth of cut h = 50 µ m predicted by the properties among the tool, the pure fiber, the pure matrix and the fiber-
thermo-mechanical model and compared to similar experimental [35] matrix interface. Some researchers consider a constant value between
and numerical results [23]. Although the composite materials, the the tool and CFRP such as 0.2 [30,31], 0.3 [17,24] and 0.5 [1]. Others
depths of cut and the cutting speeds in these studies are different from consider the friction coefficient with respect to fiber orientations in the
each other, the results can provide a rough estimation of the varying range of 0.3–0.9 [15,23]. Based on these studies, friction coefficients
range of temperature. It can be seen that the temperature variation for between the tool and the workpiece µtw , and between the fiber and
these fiber orientations is in the range of 60 °C–100 °C. The temperature matrix µ fm are set initially as 0.2 [30,31] and 0.5 [32]. Fig. 8(a) and (b)
at = 135° is the highest, while temperatures at = 90° and = 45° are show that the percentage of friction energy between the tool and the
comparatively low. Heat generates due to plastic deformation of the workpiece is smaller than the one between the fiber and matrix. At
matrix, sliding friction of fiber-matrix, and friction between the cutting = 90°, friction between the fiber and matrix dissipates more energy
tool and the CFRP. The dissipative energy mechanisms in Fig. 8(a) can compared to the others, while the case of = 135° dissipates the least.
explain the variation in temperature for the different fiber orientations. In the next section, a sensitivity analysis is conducted with a range of
The heat generating mechanisms are dominant at = 135° whereas reasonable µtw and µ fm values to study the effect of friction coefficients
on the frictional dissipation energy and cutting forces.

4.4. Fiber damage modes

Since fiber fracture is typically a dominant failure mode in CFRP


materials, an in-depth analysis is needed to study the nature of damage:
tension or compression in longitudinal, transverse or shear direction. By
analyzing the chip formation for less than 90° experimentally, Arola
et al. [8] concluded that fibers fracture due to compression in the
transverse direction, while Bhatnagar [9] attributed fiber fracture to the
axial tension mode. In order to determine the type of damage in each
fiber orientation, an FE-based method is introduced here to identify the
damage modes by analyzing the damage variables
dt1f , dcf1, dt f2, dcf2, dt f3, dcf3, dsf12± and dsf23 for the different modes of loading
introduced in Eq. (1).
Elements where a damage variable equals one (d = 1) are con-
sidered as damaged in that mode of loading. A post-processing analysis
method implemented in Python counts the number of damaged ele-
ments associated with fiber damage. Fig. 10(a)–(d) show the number of
Fig. 9. Comparison of predictions and measured cutting temperature during damaged elements normalized with respect to the corresponding total
orthogonal cutting at different fiber orientations. number of damaged elements for all fiber orientations = 0°, 45°, 90°

467
X. Yan, et al. Composite Structures 220 (2019) 460–472

Fig. 10. (a)–(d) Contribution of fiber damage modes for damage value equal to 1.0, and (e)–(h) longitudinal stress distribution 11 at four specific fiber orientations
= 0°, 45°, 90° and 135°.

and 135° . It can be observed that compressive damage dcf1 in the local are dominating at both = 45° and = 90°. The dominating types for
direction 1 (referring to Fig. 2(a) for the definition of local reference the case at 135° are compressive damage dcf1 in direction 1 and in-plane
system) is predominant for the case of = 0°, while a combination of shear damage dsf12 ± .
compression induced damage dcf1 in direction 1 (fiber direction) and dcf2 The damage mechanisms for all fiber orientations have an influence
direction 2 (transverse direction) as well as in-plane shear damage dsf12 ± on the surface quality. For = 0°, fibers bend under the compressive

468
X. Yan, et al. Composite Structures 220 (2019) 460–472

θ=0° θ=90° θ=0° θ=90°


θ=45° θ=135° θ=45° θ=135°

Cutting force Fc (N/mm)

Cutting force Fc (N/mm)


60 60

55 55

50 50

30 30

25 25

20 20

15 15
0.2 0.3 0.4 0.5 0.6 0.5 0.6 0.7 0.8 0.9
Friction coefficient between tool-FRP µ Friction coefficient between fiber-matrix µ
(a) tw (b) fm

60% 60%
θ=0° θ=90° θ=0° θ=90°
Friction energy normalized by the

Friction energy normalized by the


θ=45° θ=135° θ=45° θ=135°
corresponding total energy

corresponding total energy


50% 50%

40% 40%

30% 30%

20% 20%

10% 10%

0% 0%
0.2 0.3 0.4 0.5 0.6 0.5 0.6 0.7 0.8 0.9
Friction coefficient between tool-FRP µ Friction coefficient between fiber-matrix µ
tw fm
(c) (d)
Fig. 11. Effect of friction coefficients ( µtw and µfm ) on: (a), (b) cutting forces Fc , and (c), (d) friction energy.

force exerted by the tool in the direction of cutting speed . After the It can be concluded that fiber orientation plays a crucial role in chip
fiber fractures along the plane orthogonal to its axis, it is in contact with formation and surface quality. The former is mainly associated with
the tool rake surface and in-plane shear damage occurs in this region, as fiber damage modes, while the latter additionally depends on the di-
depicted in Fig. 10(e). The machined CFRP in this case has the best rection of propagation of fiber-matrix debonding. For the case of = 0°,
machined surface quality, as shown in Fig. 7. Since fiber-matrix de- fiber-matrix debonding propagates parallel to the cutting speed, and
bonding propagates parallel to , it can be concluded that debonding hence improves the surface quality. When 90°, debonding is less
contributes to improving the machined surface quality, while com- important and the angle between the dominating fiber damage modes
pressive damage in direction 1 and the in-plane shear damage play a and the cutting speed is 90°, so the surface quality is relatively good.
role in chip breaking. For the case at = 45°, after crack initiation the While for the case when > 90°, interface debonding makes an acute
fibers contacting with the tool rake surface are damaged due to com- angle with the cutting speed, resulting in severe sub-surface damage.
pression in direction 1 or in-plane shear, while the fibers fracture under
the cutting plane due to compression in direction 2. The latter is the
5. FE sensitivity analysis
primary fiber damage mechanism which affects the surface quality.
This results in inferior surface quality compared to the case at = 0° as
5.1. Friction coefficients
the fiber fractures perpendicularly to its axis. Fiber-matrix debonding is
secondary. This is due to the obtuse angle between the direction of
As shown in the energy analysis in Fig. 8(a), the amount of dis-
propagation of the debonding and the cutting speed. Note that the
sipated energy due to friction in all fiber orientations is around a third
conclusions for compression in direction 2 or in-plane shear in Fig. 10
of the total dissipated energy. Due to the uncertainty of the input data,
are drawn from plotting the stress distribution of 22 and 12 , respec-
the effect of friction coefficients between the tool and the workpiece µtw
tively, but have been added here for comparison.
and the fiber-matrix interface µ fm on the frictional energy and cutting
Although compressive damage in direction 1 and 2, and in-plane
forces Fc needs to be studied further. Fig. 11 shows the results of the
shear damage are the main fiber damage modes both at 45° and 90° , the
sensitivity analysis.
dominating damage mechanisms at 90° is compressive damage in the
We studied the friction coefficient for the tool-workpiece interface
transverse direction. When = 90°, the interfacial debonding and the
in the range 0.2 µtw 0.5 with constant friction coefficient at the
dominating damage mode are perpendicular and parallel to the cutting
fiber-matrix interface µ fm = 0.5 in Fig. 11(a) and (c). Similarly,
speed, respectively. Fibers bend under the transverse compression and
µtw = 0.2 was kept constant while varying the fiber-matrix friction
break below the cutting plane, which determines the surface roughness, coefficient 0.5 µfm 0.8 in Fig. 11(b) and (d). For the case of = 0°,
since the absolute value of energy dissipated by interface damage is the when µtw varies from 0.2 to 0.5, the cutting force in Fig. 11(a) and the
lowest when = 90°. Micrographs of damage beneath the machined friction energy in Fig. 11(c) increase by more than 20%. In the = 45°
surface at = 90° shown in [51] support this argument. At fiber or- samples, both the friction energy and the cutting force decrease slightly
ientation of 135° in Fig. 10(d), interface debonding forms an acute angle when 0.4 µtw 0.5. This suggests that when µtw is high enough, the
with the cutting speed, generating a poor machined surface. Fiber chip tends to stick to the tool rake face until the chip forms, then slides
bending which causes the primary damage modes in compression along away from it along the fiber orientation. When = 135°, both cutting
the direction 1, increases debonding. Therefore, more energy is dis- force and friction energy increase and then decrease with a maximum at
sipated by debonding compared to the cases at = 45° and = 90°, as µtw = 0.3.
shown in Fig. 8(b). Fig. 11(c) and Fig. 11(d) illustrate the effect of µ fm on the cutting

469
X. Yan, et al. Composite Structures 220 (2019) 460–472

Contribution to fiber damage


Contribution to fiber damage
ShrDmg, S=2.0GPa ComprDmg, Xc=4.6GPa

Cutting force Fc (N/mm)

Cutting force Fc (N/mm)


ShrDmg, S=1.5GPa ComprDmg, Xc=3.9GPa
Force Fc Force Fc
ShrDmg, S=1.0GPa 70 ComprDmg, Xc=3.5GPa 70
ShrDmg, S=0.5GPa ComprDmg, Xc=3.0GPa
100% 60 100% 60

80% 50 80% 50

60% 40 60% 40

40% 30 40% 30

20% 20 20% 20

0% 10 0% 10
0 45 90 135 0 45 90 135
Fiber orientation θ (°) Fiber orientation θ (°)
(a) Shear strength (b) Longitudinal compressive strength
Contribution to fiber damage

TensDmg, Yt=1.5GPa

Cutting force Fc (N/mm)


TensDmg, Yt=1.0GPa Force Fc
TensDmg, Yt=0.5GPa 70

100% 60

80% 50

60% 40

40% 30

20% 20

0% 10
0 45 90 135
Fiber orientation θ (°)
(c) Transverse tensile strength
Fig. 12. Contribution to fiber damage and cutting force made by (a) shear strength S, (b) longitudinal compressive strength X c , (c) transverse tensile strength Yt .

forces and friction energies for the four fiber orientations, respectively. 2.0 GPa), longitudinal compressive strength Xc takes the values
The variation of cutting force and friction energy at = 0° is com- 4.62 GPa, 3.96 GPa, 3.50 GPa and 3.00 GPa and the transverse tensile
paratively small. For this case, long chips are peeled up and removed by strength Yt is 1.5 GPa, 1.0 GPa and 0.5 GPa. Fig. 12 shows the results of
sliding along the tool rake face. In this way, the tool and the workpiece the sensitivity analysis on the three fiber strengths for the four specific
interact with each other and dissipate friction energy. Hence, the effect fiber orientations = 0°, 45°, 90°, 135°. Friction coefficients between
of µtw is more significant on the variation of friction energy compared the tool and the workpiece µtw , and between the fiber-matrix µ fm are
with µ fm . The friction coefficients µ fm and µtw have the opposite effect defined as 0.2 and 0.5, respectively. The bar graphs depict the per-
on friction energy and cutting forces for the case of = 90°. When µtw centage of the damaged elements corresponding to the studied fiber
changes from 0.2 to 0.5, friction energy and cutting forces increase strength with respect to the maximum number of damaged elements at
slightly by about 10%. In contrast, as shown in Figs. 11(b) and (d), the = 90°. Hence, the height of a bar graph can be assumed as the con-
cutting force and friction energy slightly decrease with increasing µ fm . tribution of a specific fiber damage mode to the overall fiber damage, as
Therefore, it can be concluded that as the tool is cutting perpendicular outlined in Section 4.4 and in Fig. 10. The resulting cutting forces Fc are
to the fibers at = 90°, the increase of µ fm can suppress the tendency of also included in Fig. 12.
sliding at the fiber-matrix interface. At = 135° , the trend of cutting Sensitivity analysis of fiber shear strength S Fig. 12(a) shows the sen-
forces and friction energy with respect to friction coefficient µ fm is the sitivity of the fiber shear strength S. By reducing S from 2.0 GPa to
same as the case with µ fm where both cutting force and friction energy 0.5 GPa, fiber shear damage gradually becomes the dominating fiber
increase and then decrease with a maximum at µ fm = 0.3. Peak points of damage mode for all these fiber orientations. Meanwhile, the shear
friction energy and cutting forces correspond to µ fm = 0.6 which can be strength S affects the cutting forces. When S decreases, the cutting
attributed to the fact that the CFRP is less likely to slide at the fiber- forces follow a similar trend except for the case at = 0°. A significant
matrix interface when µ fm > 0.6 and hence friction dissipates less en- drop in the cutting forces can be observed for S 1.0 GPa and equals
ergy. to 45° or 90°. This drop coincides with the fact that shear damage be-
In summary, the friction coefficients µtw and µ fm have an influence comes the dominating fiber damage mode. At higher shear strength
on the cutting responses such as cutting forces, and the effect of friction values, in-plane shear damage is minor as can be seen in the bar graphs
coefficients differs depending on the fiber orientation. in Fig. 12(a). As S reduces, the percentage of in-plane shear damage
increases and hence the cutting forces decrease.
5.2. Fiber strengths Sensitivity analysis of fiber longitudinal compressive strength Xc
Fig. 12(b) shows the effect of Xc on the contribution of longitudinal
Fiber strengths Xc , Yt and S listed in Table 2 are assumed based on compression damage as well as the cutting forces. It can be seen that the
different materials, since these are difficult to obtain through experi- cutting forces do not change significantly by changing Xc . While Xc
mental work. In order to investigate the effect of the fiber strength on affects the contributions of fiber damage for = 45° and 90° , the other
the cutting forces, we studied the sensitivity of the three strength two fiber orientations are not affected by varying Xc . Compared to the
parameters. Shear strength S varies by four levels (0.5, 1.0, 1.5 and studied shear strength S, the longitudinal fiber compressive strength Xc

470
X. Yan, et al. Composite Structures 220 (2019) 460–472

has no significant effect on the overall cutting response. Acknowledgements


Sensitivity analysis of fiber transverse tensile strength Yt Fig. 12(c)
analyzes the transverse fiber tensile strength Yt . At = 0°, the con- This research is supported by the Natural Sciences and Engineering
tribution of transverse tensile damage doubles as Yt decreases, which Research Council of Canada CANRIMT2 Strategic Research Network
results in decreasing cutting forces. For other fiber orientations, Yt has Grant (NETGP 479639-15) in machining and several industrial mem-
no obvious influence on the fiber damage mode and the cutting forces. bers of NSERC CANRIMT. The principal author has been sponsored by
In summary, we showed that the fiber orientation-related cutting National Natural Science Foundation of China (Grant No. 51705426)
forces can be dependent on the investigated fiber strength values which and the Key Research and Development Program of Shaanxi Province
is closely related to the varying dominant fiber damage mode. When the (No. 2017GY-101).
shear strength S decreases, the dominating fiber damage mode changes
and results in a significant reduction in cutting forces. Xc slightly affects Appendix A. Supplementary data
the contributions of fiber damage modes for = 45° and 90° , while the
cutting forces remain constant. Yt has no obvious effect for fiber or- Supplementary data associated with this article can be found, in the
ientations > 0° . online version, athttps://doi.org/10.1016/j.compstruct.2019.03.090.

References
6. Conclusion
[1] Santiuste C, Soldani X, Miguélez M. Machining fem model of long fiber composites
The aim of this paper is to investigate the underlying (dominating) for aeronautical components. Compos Struct 2010;92(3):691–8.
[2] An Q, Ming W, Cai X, Chen M. Study on the cutting mechanics characteristics of
failure and damage mechanisms during orthogonal cutting of uni- high-strength ud-cfrp laminates based on orthogonal cutting method. Compos Struct
directional CFRP (UD-CFRP) composites. A 3D micro-scale finite ele- 2015;131:374–83.
ment model is proposed with numerical predictions from this model [3] Altintas Y. Manufacturing automation: metal cutting mechanics, machine tool vi-
brations, and CNC design. New York: Cambridge University Press; 2012.
matching well with experimentally measured cutting forces for fiber [4] Qi Z, Zhang K, Li Y, Liu S, Cheng H. Critical thrust force predicting modeling for
orientations ranging from 0° to 180°. The good correlation allows for a delamination-free drilling of metal-frp stacks. Compos Struct 2014;107:604–9.
detailed FE analysis, and conclusions are drawn from this analysis as [5] Merino-Pérez J, Royer R, Merson E, Lockwood A, Ayvar-Soberanis S, Marshall M.
Influence of workpiece constituents and cutting speed on the cutting forces devel-
follows.
oped in the conventional drilling of cfrp composites. Compos Struct
2016;140:621–9.
• A detailed energy analysis enables quantification of the different [6] Kahwash F, Shyha I, Merson E, Maheri A. Meshfree formulation for modelling of
orthogonal cutting of composites. Compos Struct 2017;166:193–201.
energy-absorbing mechanisms during orthogonal cutting. The
[7] Koplev A, Lystrup A, Vorm T. The cutting process, chips, and cutting forces in
overall dissipated energy varies with fiber orientation, with trends machining cfrp. Composites 1983;14(4):371–6.
that follow the variation of cutting forces. The percentage of con- [8] Arola D, Ramulu M, Wang D. Chip formation in orthogonal trimming of graphite/
tribution of each mechanism to the overall energy dissipation also epoxy composite. Compos Part A-Appl S 1996;27(2):121–33.
[9] Bhatnagar N, Ramakrishnan N, Naik N, Komanduri R. On the machining of fiber
changes at different fiber orientation. Fiber damage dominates when reinforced plastic (frp) composite laminates. Int J Mach Tool Manuf
fiber orientation is less than 90° while plastic dissipation in the 1995;35(5):701–16.
matrix material dominates when fiber orientation is greater than [10] Wang X, Zhang L. An experimental investigation into the orthogonal cutting of
unidirectional fibre reinforced plastics. Int J Mach Tool Manuf
90° .

2003;43(10):1015–22.
More insight into the fiber orientation-dependent surface mor- [11] Diaz O, Axinte D. Towards understanding the cutting and fracture mechanism in
phology is provided. Fiber damage modes are identified and found ceramic matrix composites. Int J Mach Tool Manuf 2017;118:12–25.
[12] Reiner J, Vaziri R. Structural analysis of composites with finite element codes: an
to be different for different fiber orientations. Experimental ob- overview of commonly used computational methods. In: Beaumont P, Zweben C,
servations such as peel-up, crushing and bending failure can be at- editors. Comprehensive composite materials II. Oxford: Academic Press; 2018. p.
tributed to the fact that the fibers in contact with the tool surface are 61–84. vol. 8; chap. 8.4.
[13] Wang D, He X, Xu Z, Jiao W, Yang F, Jiang L, et al. Study on damage evaluation and
experiencing various failure modes with respect to fiber orienta- machinability of ud-cfrp for the orthogonal cutting operation using scanning
tions. Surface roughness is closely associated with fiber damage and acoustic microscopy and the finite element method. Materials 2017;10(2):204.
the direction of propagation of fiber-matrix debonding. [14] Lasri L, Nouari M, Mansori M. Modelling of chip separation in machining uni-

• Friction coefficient and some fiber strength values are difficult to directional frp composites by stiffness degradation concept. Compos Sci Technol
2009;69(5):684–92.
obtain experimentally, thus a detailed sensitivity analysis is con- [15] Nayak D, Bhatnagar N, Mahajan P. Machining studies of ud-frp composites part 2:
ducted to study their effect. It is found that increasing the friction finite element analysis. Mach Sci Technol 2005;9(4):503–28.
coefficient leads to higher levels of frictional energy and cutting [16] Song D, Li Y, Zhang K, Cheng H, Liu P, Hu J. Micromechanical analysis for mi-
croscopic damage initiation in fiber/epoxy composite during interference-fit pin
forces, however the magnitude differs depending on the fiber or- installation. Mater Des 2016;89:36–49.
ientation. In the case of = 90°, the cutting force and frictional [17] Rao G, Mahajan P, Bhatnagar N. Micro-mechanical modeling of machining of frp
energy decrease when the friction coefficient between the fiber- composites-cutting force analysis. Compos Sci Technol 2007;67(3):579–93.
[18] Xu W, Zhang L, Wu Y. Effect of tool vibration on chip formation and cutting forces
matrix increases. When fiber shear strength or fiber longitudinal in the machining of fiber-reinforced polymer composites. Mach Sci Technol
compressive strength is reduced, the corresponding fiber damage 2016;20(2):312–29.
mode gradually becomes the dominating mechanism. Consequently, [19] He Y, Davim J, Xue H. 3d progressive damage based macro-mechanical fe simu-
lation of machining unidirectional frp composite. Chin J Mech Eng 2018;31(1):51.
these fiber strength values affect the overall energy dissipated in [20] Calzada K, Kapoor S, DeVor R, Samuel J, Srivastava A. Modeling and interpretation
fiber damage and cutting forces. Transverse tensile strength has of fiber orientation-based failure mechanisms in machining of carbon fiber-re-
comparatively less influence on the fiber damage modes and cutting inforced polymer composites. J Manuf Process 2012;14(2):141–9.
[21] Xu W, Zhang L, Wu Y. Elliptic vibration-assisted cutting of fibre-reinforced polymer
responses. composites: understanding the material removal mechanisms. Compos Sci Technol
2014;92:103–11.
This combined experimental and numerical study provides further [22] Xu W, Zhang L. A new approach to characterising the surface integrity of fibre-
reinforced polymer composites during cutting. Compos Part A-Appl Sci
insight into the fundamental principles of orthogonal cutting of CFRP
2017;103:272–82.
composites. The techniques of energy analysis and sensitivity analysis [23] Gao C, Xiao J, Xu J, Ke Y. Factor analysis of machining parameters of fiber-re-
used in this study can be directly applied to complex scenarios such as inforced polymer composites based on finite element simulation with experimental
milling or drilling of CFRP panels to study the dominating failure me- investigation. Int J Adv Manuf Technol 2016;83(5–8):1113–25.
[24] Cheng H, Gao J, Kafka O, Zhang K, Luo B, Liu W. A micro-scale cutting model for ud
chanisms and their interaction during the machining process of com- cfrp composites with thermo-mechanical coupling. Compos Sci Technol
posite materials.

471
X. Yan, et al. Composite Structures 220 (2019) 460–472

2017;153:18–31. 1999;33(17):1620–42.
[25] Rao G, Mahajan P, Bhatnagar N. Machining of ud-gfrp composites chip formation [39] Daniel I, Liber T. Testing of fiber composites at high strain rates. Proceedings of the
mechanism. Compos Sci Technol 2007;67(11–12):2271–81. 2nd International Conference on Composite Materials. 1978. p. 1003–18. Toronto,
[26] Abena A, Soo SL, Essa K. Modelling the orthogonal cutting of ud-cfrp composites: Canada.
development of a novel cohesive zone model. Compos Struct 2017;168:65–83. [40] Vaughan TJ, McCarthy CT. Micromechanical modelling of the transverse damage
[27] Cytec mtm45-1 epoxy matrix datasheet. Cytec Solvay Group 2012. behaviour in fibre reinforced composites. Compos Sci Technol 2011;71(3):388–96.
[28] Merchant M. Mechanics of the metal cutting process. i. Orthogonal cutting and a [41] Benzeggagh M, Kenane M. Measurement of mixed-mode delamination fracture
type 2 chip. J Appl Phys 1945;16(5):267–75. toughness of unidirectional glass/epoxy composites with mixed-mode bending ap-
[29] Hextow as4 carbon fiber datasheet. Hexcel Corp 2014. paratus. Compos Sci Technol 1996;56(4):439–49.
[30] Matsunaga S, Matsubara T, Wang W, Takao Y. Effects of reciprocation number on [42] Kaddour A, Hinton M, Smith P, Li S. Mechanical properties and details of composite
the friction behaviors of carbon/epoxy for various fiber orientations and high laminates for the test cases used in the third world-wide failure exercise. J Compos
contact pressures. Proceedings of the 13th International Conference on Composite Mater 2013;47(20–21):2427–42.
Materials. 2001. Beijing, China. [43] Weeton J, Thomas K, Peters D. Engineers’ guide to composite materials. American
[31] Klinkova O, Rech J, Drapier S, Bergheau J. Characterization of friction properties at Society of Metals; 1987.
the workmaterial/cutting tool interface during the machining of randomly struc- [44] Melro A, Camanho P, Pires FA, Pinho S. Micromechanical analysis of polymer
tured carbon fibers reinforced polymer with carbide tools under dry conditions. composites reinforced by unidirectional fibres: Part ii-micromechanical analyses.
Tribol Int 2011;44(12):2050–8. Int J Solids Struct 2013;50(11):1906–15.
[32] Schön J. Coefficient of friction of composite delamination surfaces. Wear [45] Camanho P, Arteiro A. Analysis models for polymer composites across different
2000;237(1):77–89. length scales. The structural integrity of carbon fiber composites. 2017. p. 199–279.
[33] Kawabata S. Measurement of the transverse mechanical properties of high-perfor- [46] Varna J, Berglund LA, Ericson ML. Transverse single-fibre test for interfacial de-
mance fibres. J Text I 1990;81(4):432–47. bonding in composites: 2. Modelling. Compos Part A-Appl Sci 1997;28(4):309–15.
[34] Belingardi G, Mehdipour H, Mangino E, Martorana B. Progressive damage analysis [47] Rabearison N, Jochum C, Grandidier J. A fem coupling model for properties pre-
of a rate-dependent hybrid composite beam. Compos Struct 2016;154:433–42. diction during the curing of an epoxy matrix. Comp Mater Sci 2011;45(3):715–24.
[35] Zitoune R, Collombet F, Lachaud F, Piquet R, Pasquet P. Experiment-calculation [48] Reiner J, Torres J, Veidt M, Heitzmann M. Experimental and numerical analysis of
comparison of the cutting conditions representative of the long fiber composite drop-weight low-velocity impact tests on hybrid titanium composite laminates. J
drilling phase. Compos Sci Technol 2005;65(3–4):455–66. Compos Mater 2016;50(26):3605–17.
[36] Naik N, Shankar P, Kavala V, Ravikumar G, Pothnis J, Arya H. High strain rate [49] McGregor C, Vaziri R, Xiao X. Finite element modelling of the progressive crushing
mechanical behavior of epoxy under compressive loading: experimental and mod- of braided composite tubes under axial impact. Int J Impact Eng
eling studies. Mater Sci Eng A-Struct 2011;528(3):846–54. 2010;37(6):662–72.
[37] Fiedler B, Hobbiebrunken T, Hojo M, Schulte K. Influence of stress state and tem- [50] Williams J, Patel Y. Fundamentals of cutting. Interface Focus 2016;6:3.
perature on the strength of epoxy resins. Proceedings of the 11th International [51] Voß R, MHenerichs M, Kuster F, Wegener K. Chip root analysis after machining
Conference on Fracture (ICF 11). 2005. p. 2271–5. Turin, Italy. carbon fiber reinforced plastics (cfrp) at different fiber orientations. Proc CIRP
[38] Hsiao H, Daniel I, Cordes R. Strain rate effects on the transverse compressive and 2014;14:217–22.
shear behavior of unidirectional composites. J Compos Mater

472

You might also like