Beruflich Dokumente
Kultur Dokumente
Modeling of Machining
Unidirectional Composites:
Prediction of Debonding and
Fiber Damage
Chinmaya R. Dandekar
Graduate Research Assistant A multiphase finite element model using the commercial finite element package ABAQUS/
EXPLICIT is developed for simulating the orthogonal machining of unidirectional fiber
Yung C. Shin reinforced composite materials. The composite materials considered for this study are a
Professor glass fiber reinforced epoxy and a tube formed carbon fiber reinforced epoxy. The effects
of varying the fiber orientation angle and tool rake angle on the cutting force and
Center for Laser-Based Manufacturing, damage during machining are considered for the glass fiber reinforced epoxy. In the case
School of Mechanical Engineering, of carbon fiber reinforced epoxy, only the effect of fiber orientation on the measured
Purdue University, cutting force and damage during machining is considered. Two major damage phenom-
West Lafayette, IN 47907 ena are predicted: debonding at the fiber-matrix interface and fiber pullout. In the mul-
tiphase approach, the fiber and matrix are modeled as continuum elements with isotropic
properties separated by an interfacial layer, while the tool is modeled as a rigid body.
The cohesive zone modeling approach is used for the interfacial layer to simulate the
extent of debonding below the work surface. Bulk deformation and shear failure are
considered in the matrix for both the models and the glass fiber. A brittle failure criterion
is used for the carbon fiber specimen and is coded in FORTRAN as a user defined material
(VUMAT). The brittle failure of the carbon fibers is modeled using the Marigo model for
brittle failure. For validation purposes, simulation results of the multiphase approach are
compared with experimental measurements of the cutting force and damage. The model is
successful in predicting cutting forces and damage at the front and rear faces with respect
to the fiber orientation. A successful prediction of fiber pullout is also demonstrated in
this paper. DOI: 10.1115/1.2976146
Journal of Manufacturing Science and Engineering OCTOBER 2008, Vol. 130 / 051016-1
Copyright 2008 by ASME
carbon fibers. Thus the damage model incorporated in this paper
accounts for these statistical variances in strength. 2 H 2
have been limited to the prediction of the damage at the front face
of the fiber only. 2EY mw/2
D = 1 exp 7
wl
2
volume fraction. Unlike the strain rate dependency of the glass
un 2
ut 2 1/2 fiber, the carbon fiber is considered strain rate independent 18.
= + 8 The difference in the model for the carbon fiber and the glass fiber
n t
is the modulus of elasticity. Carbon fibers are orthotropic in na-
The traction separation law is then implemented through Eq. ture, while glass fibers are modeled as isotropic materials. Dam-
9, where max is the cohesive strength. In applying the traction age in the carbon fibers is then calculated using the Marigo model
separation law, the important parameters are max and the normal 5 for the carbon fiber. The two Weibull parameters needed for
separation energy Gc, or , while the shape of the traction sepa- the damage model can be obtained from literature. In the study the
ration law F is considered to be of second order relevance values of the Weibull parameters used are mw at 5.3 and wl at
15,16, 3.33 GPa 19.
27
F = 4 max1 2 + 2 9
4 Simulation Procedure
The normal separation energy Gc is the area under the traction
separation curve and is shown in Eq. 10. Knowing the cohesive A combination of 2D plane stress continuum quadrilateral
strength Tn and Tt and the Mode I GIC and Mode II GIIC CPS4R and triangular CPS3 elements is used for meshing the
fracture energies and using Eq. 10, we can then calculate the fiber and matrix, while the interface layer is modeled using the
zero-thickness 2D cohesive elements. A plane stress analysis is
maximum allowable separation distance of the cohesive ele-
used for the study contrary to the plane strain analysis convention-
ment. In Eq. 13, is the ratio of the shear to normal strength of ally used in cutting homogenous materials. The plane strain analy-
the cohesive elements, which can be calculated from the given sis is not valid for fiber reinforced composites due to the out of
material properties, plane displacements present in the machining of FRPs 2,3,6. A
16 max
9
Gc = 10 representative mesh for a fiber orientation of 90 deg is shown in
Fig. 1. The cutting tool is modeled as a rigid body. For the simu-
un lation the boundary conditions applied are as follows: the work-
Tn = F 11 piece is constrained to move in both the x and y directions at
n
the bottom side, the left-hand side, and the lower right-hand side.
ut The tool is given a constant velocity in the x direction, and the
Tt = F 12 tool movement in the y direction is constrained.
t
GIIC
= 13 Table 2 Mechanical properties of carbon fiber and epoxy
GIC
Carbon fiber
3 Material Properties Diameter, m 14
Modulus of elasticity E11, GPa 240
The material properties specified for the glass fiber reinforced Modulus of elasticity E22, GPa 18
composite are provided in Table 1. The fibers are considered as a Tensile strength, MPa 3600
fiber bundle to represent the fiber volume fraction of 60% used in Shear strength, MPa 38
the study. One of the reasons for this is to simplify the numerical
problems associated with concurrent single fiber and matrix Matrix, epoxy
phases due to the considerable difference in the moduli and Modulus of elasticity, GPa 3.1
strength between the fiber and matrix. The matrix material has Tensile strength, MPa 70
been considered to be strain rate and temperature independent Fiber-matrix interface
since the temperatures obtained during machining of carbon fiber Normal strength, MPa 160
reinforced composites are low. The glass fiber is found to be strain Shear strength, MPa 34
rate dependent and follows the constitutive equation Eqs. 14 Strain energy release rate, J / m2 50
and 15 proposed by Wang and Xia 17,
Journal of Manufacturing Science and Engineering OCTOBER 2008, Vol. 130 / 051016-3
Material Rake angle deg Nose radius m Depth of cut mm Cutting speed m/s Fiber orientation deg
Journal of Manufacturing Science and Engineering OCTOBER 2008, Vol. 130 / 051016-5
Data points
Force data N/mm Fiber orientation deg 1 2 3 4 5 6 Average
measurement of the cutting force. Figure 5 shows a representative terior cross-section of the samples. For each fiber orientation, five
graph of the measured cutting force for a specimen with a 90 deg measurements each from three different samples were taken. The
fiber orientation. The solid line in the figure indicates the eight reported value is the average value of these 15 measurements for
point moving average, which, as mentioned previously, is used in each fiber orientation. A variation of 15% is observed in the ex-
reporting the cutting force values. The final reported value of the perimental measurements. The images indicate the extent of deb-
cutting force is the average of the steady state values for six ex- onding between the fiber and the matrix. Other damage phenom-
periments conducted under the same cutting condition, as given in ena are also observed, primarily microcracking of fibers,
Table 5. The width of cut was 2.15 mm, and the length of each cut debonding, and fiber pullout, as shown in Figs. 79. The average
was 6 mm. The average values of the main cutting and thrust force measured values for debonding are 78 m, 56 m, and 47 m,
for machining 45 deg fiber oriented samples were 24 N/mm and respectively, for the 45 deg, 90 deg, and 120 deg fiber orientation
47.8 N/mm, respectively. For the 90 deg fiber orientation samples, samples. The average measured values for the fiber pullout are
the average values for the main cutting force and the thrust force 58 m, 40 m, and 30 m for the 45 deg, 90 deg, and 120 deg
per width were 47 N/mm and 19.7 N/mm respectively, while they fiber orientation specimens, respectively. It is clear from Figs. 79
were 41.4 N/mm and 22 N/mm, respectively, in the case of the
120 deg fiber orientation.
Figure 6 shows the measured cutting forces per unit width as a
function of the fiber orientations. Contrary to the expectation, the
average cutting force was lower than the average thrust force for
the fiber orientation of 45 deg, while the cutting force was much
larger than the average thrust force for the 90 deg and 120 deg
fiber orientations. The average cutting force increased with in-
creasing fiber orientation, but above the 90 deg fiber orientation
the cutting force decreased. In the case of the thrust force, the
maximum value was observed during machining of the 45 deg
fiber orientation samples, but as the fiber orientation angle in-
creased the thrust force decreased drastically. A further increase in
the fiber orientation angle resulted in a slight increase in the thrust
force. The above phenomenon wherein the cutting force is smaller
than the thrust force for the 45 deg fiber orientation and the
gradual increase and subsequent decrease in the cutting forces
were previously observed for edge trimming of graphite epoxy
unidirectional laminates by Arola et al. 3 and by Wang et al.
26.
To check for the prediction of damage during the machining of Fig. 7 Damage observations in the machining of 45 deg fiber
CFRP polymers, SEM images were obtained by studying the in- orientation CFRP specimens at a cutting speed of 1 m/s
Fig. 10 Simulated von Mises stress distribution during the machining of a 90 deg fiber orientation with a 5 deg
rake angle tool
Journal of Manufacturing Science and Engineering OCTOBER 2008, Vol. 130 / 051016-7
33 N/mm, respectively. Finally for the 135 deg fiber orientation, Rao et al. 7 showed encouraging results in their prediction of
the simulated values for the main and thrust forces are 40.5 N/mm both the cutting and thrust force results. The model presented here
and 30.2 N/mm, while the experimental results are 38 N/mm and predicts the main cutting force and the thrust force along with the
36 N/mm, respectively. The simulated force results compare well observable trend very well. The cutting and thrust forces margin-
with the experimental data, as shown in Fig. 14, where the cutting ally decreased with increasing rake angle, as shown in Fig. 15.
force decreases and the thrust force increases with decreasing fi-
ber orientation. Overall the maximum discrepancy between the 7.2 Simulation Results for Machining of Carbon Fiber Re-
simulated and experimental results for the cutting force is 611%, inforced Epoxy. The finite element models for the 90 deg and
while the thrust forces differ by 1216%. In comparison the FEM 120 deg fiber orientations are shown in Figs. 16 and 17. Once
model by Nayak and Bhatnagar 6,10 predicted the cutting force again apparent in the figures is the extent of the debonding zone
with a discrepancy of 10%. However their thrust force results represented by the deleted element along the fiber-matrix interface
differed by almost 38%, showing poor correlation. On the other at the front and back sides of the fiber bundle. The novel result for
hand, Arola et al. 3 failed to predict the trend observed for the this composite involves the deleted elements in the fiber below the
thrust force during the cutting of GFRP. A more recent study by cutting plane, representing the fiber damage. As is the case with
the glass fiber case, the fiber failure mechanism shows consider-
able compression induced failure on the front side of the fiber,
whereas the rear of the fiber shows a tensile rupture of the fiber.
Fiber orientation was varied for the tool with a 5 deg rake angle.
The maximum debonding depth decreased with increasing fiber
orientation: the simulated values of the damage depth for 45 deg,
90 deg, and 120 deg fiber orientations are 96 m, 75.9 m, and
56 m, respectively. This is very close to the experimental obser-
vations Fig. 18.
There was a very close agreement between the predicted and
measured values of fiber damage by experimental measurements
and simulated measurements for 90 deg and 120 deg fiber orien-
tation samples. In the case of the 90 deg fiber orientation sample,
the experimental damage depth was measured at 40 m, while
the simulated value of damage depth in the fiber is 50.6 m. In
the case of 120 deg fiber orientation samples, the experimental
damage depth was 30 m, while the simulated damage depth is
29 m. The simulation also corroborated the experimental data in
Fig. 12 Internal damage during orthogonal machining of that the damage observed during the machining of these compos-
GFRP during machining with a 5 deg rake angle tool, in com- ites decreased with an increase in the fiber orientation.
parison with experimental measurements 10 The comparison of simulated cutting and thrust forces with the
experimental data is plotted in Fig. 19. The simulated force results 8 Conclusions
compare well with the experimental data where the cutting force
Multiphase orthogonal machining simulations were conducted
increases gradually and then decreases once again with decreasing
for two unidirectional polymeric composites: a glass fiber rein-
fiber orientation, while the thrust force increases with decreasing
forced epoxy and a carbon fiber reinforced epoxy. The simulta-
fiber orientation. This reduction in the cutting forces resulted in
neous use of the Marigo brittle failure model and the cohesive
less damage as observed during the machining of higher fiber
zone model to describe material behavior was successful in pre-
orientation samples. Overall the maximum difference between the
dicting damage during machining of composites. Both machining
simulated and experimental results for the cutting force is 810%,
models compared well with the experimental data in terms of
while the thrust force differs by 1415%. The trends observed
cutting force and damage prediction. Debonding depth and fiber
both experimentally and through simulated results match ex-
damage decreased with increasing fiber orientation. The model
tremely well.
presented in this study is able to predict debonding at the front and
rear faces of the fiber, fiber damage, and cutting forces during
Fig. 14 Force per unit width as a function of fiber orientation Fig. 15 Force per unit width as a function of rake angle during
during the machining of GFRP with a 5 deg rake angle tool, in the machining of a 90 deg fiber oriented GFRP with a 5 deg rake
comparison with experimental measurements 10 angle tool
Journal of Manufacturing Science and Engineering OCTOBER 2008, Vol. 130 / 051016-9
Fig. 17 Simulated von Mises stress distribution during the machining of 120 deg fiber orientation carbon fiber
specimens with a 5 deg rake angle tool
References
1 Konig, W., Wulf, Ch., Gra, P., and Willerscheid, H., 1985, Machining of
Fiber Reinforced Plastics, CIRP Ann., 342, pp. 537547.
2 Arola, D., and Ramulu, M., 1997, Orthogonal Cutting of Fiber-Reinforced
Composites: A Finite Element Analysis, Int. J. Mech. Sci., 39, pp. 597613.
3 Arola, D., Sultan, M. B., and Ramulu, M., 2002, Finite Element Modeling of
Fig. 19 Comparison of experimental and simulated cutting Edge Trimming Fiber Reinforced Plastics, ASME J. Manuf. Sci. Eng.,
forces per unit width for CFRP specimens machined at a cut- 1241, pp. 3241.
ting speed of 1 m/s with varying fiber orientations 4 Dandekar, C., Orady, E. A., and Mallick, P. K., 2007, Drilling Characteristics
of an E-Glass Fabric Reinforced Polypropylene Composite and an Aluminum
Alloy: A Comparative Study, ASME J. Manuf. Sci. Eng., 129, pp. 1080
1087.
machining of composites with a few limitations on the considered 5 Nayak, D., Singh, I., Bhatnagar, N., and Mahajan, P., 2004, An Analysis of
fiber orientation angles. One limitation is that the model does not Machining Induced Damages in FRP Composites: A Micromechanics Finite
Element Approach, AIP Conf. Proc., 712, pp. 327331.
predict damage in the matrix in the form of cracking and matrix 6 Nayak, D., and Bhatnagar, N., 2005, Machining Studies of UD-FRP Com-
redistribution. Since the aim of the study was to consider only posites. Part II: Finite Element Analysis, Mach. Sci. Technol., 9, pp. 503
fiber damage along with debonding, in the future matrix damage 528.
and redistribution of fibers can be considered. 7 Rao, G. V. G., Mahajan, P., and Bhatnagar, N., 2007, Micro-Mechanical
Modeling of Machining of FRP Composites-Cutting Force Analysis, Compos.
In the case of glass fiber reinforced epoxy composites, the cut- Sci. Technol., 67, pp. 579593.
ting force decreased with increasing fiber orientation and tool rake 8 Rao, G. V. G., Mahajan, P., and Bhatnagar, N., 2007, Machining of UD-
angle. The thrust force increased with increasing fiber orientation GFRP Compsites Chip Formation Mechanism, Compos. Sci. Technol., 67,
but remained fairly constant with increasing tool rake angle. In- pp. 22712281.
9 Hocheng, H., and Dharan, C. K. H., 1990, Delamination During Drilling in
creasing the tool rake angle resulted in no debonding for the glass Composite Laminates, ASME J. Eng. Ind., 1123, pp. 236239.
fiber reinforced epoxy simulations. For the carbon fiber reinforced 10 Nayak, D., and Bhatnagar, N., 2005, Machining Studies of Uni-Directional
epoxy composites, the average cutting force increased with in- Glass Fiber Plastic UD-GFRP Composites. Part I: Effect of Geometrical and
creasing fiber orientation until 90 deg, after which the cutting Process Parameters, Mach. Sci. Technol., 9, pp. 481501.
11 Lemaitre, J., and Desmorat, R., 2005, Engineering Damage Mechanics: Duc-
force decreased. A large value of the thrust force was observed tile, Creep, Fatigue and Brittle Failures, Springer-Verlag, Berlin.
during machining of the 45 deg fiber orientation samples, but the 12 Hibbit, Karlesson & Sorenson, Inc., 2004, ABAQUS Users Manual, Version 6.5,
thrust force decreased drastically with higher fiber orientation Pawtucket, RI.
angles. The brittle failure model incorporated as a user defined 13 Tian, Y., and Shin, Y. C., 2007, Multiscale Finite Element Modeling of Sili-
con Nitride Ceramics undergoing Laser-Assisted Machining, ASME J.
material model VUMAT was successful in simulating the fiber Manuf. Sci. Eng., 1292, pp. 287295.
damage observed in machining carbon fiber reinforced epoxy 14 Tvergaard, V., 1990, Effect of Fiber Debonding in a Whisker-Reinforced
composites for 90 deg and 120 deg fiber orientations. Metal, Mater. Sci. Eng., A, 125, pp. 203213.
15 Tvergaard, V., and Hutchinson, J. W., 1992, The Relation Between Crack
Growth Resistance and Fracture Process Parameters in Elastic-Plastic Solids,
Acknowledgment J. Mech. Phys. Solids, 40, pp. 13771397.
16 Hutchinson, J. W., and Evans, A. G., 2000, Mechanics of Materials: Top-
The authors are grateful to Dr. R. B. Pipes and J. Coulter for Down Approaches to Fracture, Acta Mater., 48, pp. 125135.
their assistance in providing and preparing the carbon fiber rein- 17 Wang, Y., and Xia, Y., 2000, Dynamic Tensile Properties of E-Glass, Kevlar
49 and Polyvinyl Alcohol Fiber Bundles, J. Mater. Sci. Lett., 19, pp. 583
forced epoxy tube formed specimens. The authors would like to 586.
gratefully acknowledge that this research was partially supported 18 Zhou, Y., Wang, Y., Jellani, S., and Xia, Y., 2007, Experimental Study on
by the National Science Foundation Grant No. IIP0538786. Tensile Behavior of Carbon Fiber and Carbon Fiber Reinforced Aluminum at
Journal of Manufacturing Science and Engineering OCTOBER 2008, Vol. 130 / 051016-11