Polymer Composites - 2004 - Pontes - The Role of The Interaction Coefficient in The Prediction of The Fiber Orientation in
Polymer Composites - 2004 - Pontes - The Role of The Interaction Coefficient in The Prediction of The Fiber Orientation in
Polymer Composites - 2004 - Pontes - The Role of The Interaction Coefficient in The Prediction of The Fiber Orientation in
interaction coefficient C, in an added-in diffusion from second- to fourth-order requires the solution of a
term. This approach was successful in predicting with closure problem (5).A number of closure approxima-
reasonable accuracy the fiber orientation in plates tions have been proposed and applied with varying
and circular discs molded with heavily reinforced degree of success (5, 11). Moreover, the tensors offer
composites ( 10). the possibility of representing the statistical distribu-
In this work a study was conducted to analyze the tion of fiber orientation, and may be assigned to a
effect of the fiber interaction coefficient C, on the pre- physical meaning. The second-order tensor of fiber
dictions of the fiber orientation developing in lO?h by orientation has 3 X 3 terms (like a 3 x 3 matrix). The
weight glass fiber reinforced polycarbonate discs. The sum of the diagonal terms of the tensor is equal to the
orientation predictions were compared with experi- unity, and each of the diagonal values is allowed to
mental data obtained at three positions along the flow vary between 0 and 1. The value of each of the diago-
path of circular discs (radial direction). nal values stands for the relative orientation around
one of the co-ordinate axis. The axes are commonly
2. THEORY defined as 1-radial, 2-tangential, and 3- the out of
plane (or thickness) direction. The tensor being sym-
2.1. Definition of the Fiber Interaction Regime
metrical by definition implies that only three out-of-
The interaction regime in the composite melt is as- diagonal elements need to be computed. Therefore, five
sociated to the aspect ratio of the fibers (Lid), and to calculations are required to define the second-order
its volume fraction, (Vf). Typical long and slender tensor, without further simpliijang assumptions (only
fibers have an aspect ratio between 15 and 40. A di- two of the diagonal values are independent).
lute solution where the fibers hardly interact with The elements of the second-order tensor have a
each other is characterized by Vf < ( d / L ) 2(3).No com- physical meaning. The value of each of the diagonal
mercially important composites fall into this dilute elements of the tensor stands for the relative orienta-
regime. tion in one of the co-ordinate axes. The flow of poly-
If the volume fraction of fibers is inside the range mer composites in injection moldings is planar, lead-
( d / L ) 2< Vf < ( d / L ) ,the suspension is semi-concen- ing to planar fiber orientation states. Thus small
trated, and the spacing between fibers being smaller values of the a33tensor component are typical ob-
than L is larger than d The interaction between fibers tained. The element a l l is close to the unity if high
is frequent. The 10% by weight short-glass-fiberpoly- orientation in flow direction exists. In planar and ran-
carbonate composites used in this study fit into this dom distributions of fiber orientation the values of a ,
semi-concentrated regime. and aZ2are close to 0.5, a schematic representation
A highly concentrated solution is defined when V, > being shown in Fig. 1.
( d / L ) and the spacing between the fibers is of the In this case the second-order tensor is:
order of magnitude of the fiber diameter, d.
2.2. Tensors of Fiber Orientation
The state of fiber orientation is well described by the
ai,= [ ; ;]
0.5 0
0;
0
I
/ \
', \ /
/ \
\
/ /
t"'
\ '
Fig. 1. Random planarfiber orientation.
the fiber aspect ratio, L/d,and on the volume frac- Table 1. Molding Program.
tion. Some authors suggest that the interaction coeffi- Code Melt temp. Injectionflow rate Mold temp.
cient should be dependent on the state of orientation ("C) (cm3I s ) ("C)
(12, 13). This statement apparently contradicts the
280.10 9.7
concept of the interaction coefficient being an intrinsic
property of the suspension, and brings an additional 280.14 280 13.8
transient character to the coefficient. However, with-
280.32 32.2
out a detailed model for describing the transient inter-
actions between fibers, there is no way to predict C,, 300.14 300 13.8 100
and therefore it must be determined experimentally. 320.10 9.7
Tucker and Advani (4) suggested a n empirical expres-
sion to be used when an experimentally determined 320.14 320 13.8
interaction coefficient is not available: 320.32 32.2
(3
The measurements of fiber orientation were made
0 = arc cos - (3)
on polished cross sections cut from the molded plates.
Because the direction of penetration of the fiber in included in the FEM model. The mesh used in the
the specimens is not fully defined in a single cross C-Mold simulations is shown in Fig. 4.
section there is some uncertainty about the angle +, The information required to perform the simulation
+
indistinguishable from the angle + 180".This ambi- includes:
guity that affects only the out-of-diagonal tensor com-
The mesh of the part;
ponents is present in our data.
The in-plane orientation is determined by the angles The material properties:
defined by the major axis of the ellipse and the prese-
lected reference axes (Fig. 3 ) . These angles, 8 and +, The process conditions:
can be determined by digitizing the coordinates of the The parameter data.
endpoints of the major and minor axes of the ellipses,
The processing conditions were defined according to
either manually or by image analysis tools.
the set up for the experimental tests. The required
The relevant components of the unit vector p for the
material properties were obtained in the material
calculation of the orientation tensor are shown in Eqs
database of C-Mold, except for the number average
4 to 6 for the specimens cut in the 1-3 sectioning
fiber length, diameter and fiber volume fraction. These
plane.
data were determined experimentally after pirolysis of
p 1 = sin 8 cos + (4) polycarbonate granules, as 238 km, 14 pm and 5.8%,
respectively.
p2 = cos 0 (51 In order to increase the accuracy of the numerical
solution, the default number of layers across the
p 3 = sin 8 sin + (6) thickness was increased from the C-Mold standard
number of 12 to 20.
The elements of the second-order tensors that de- There are two main parameters to be adjusted when
scribe the fiber orientation can be calculated from the fiber orientation is simulated with C-Mold: the inter-
values of the vectors piusing Eq 7. action coefficient, C,,and the inlet boundary condi-
tion. In this study the boundary condition (at the
sprue entrance) was defined considering that the
fibers are aligned in the flow direction at the skin, and
randomly aligned at the core. This parameter is less
important when compared to C,,since at a short dis-
The parameter F, is a weighting function that corrects tance after the gate entrance the effect of this parame-
the bias resulting from the lower probability of a fiber ter on the simulation results vanishes. The inlet
lying parallel to the section plane appear in the image boundary condition is very important for large gates
(14). in which the orientation at the entrance dominates
the flow pattern.
3.4. C-Mold Simulations
3.5. Analysis of the Influence of C,
For the simulation of the fiber orientation a model
with 707 elements was created in the C-Mold pack- The effect of varying the interaction coefficient, C,
age (15).The direct sprue and the impression were was determined quantitatively considering the average
Fg.3. Possible forms of fher sections in a polishing surface, 4 is the out of plane orientatwn angle.
deviation between the experimental and the predicted the all tensor component of fiber orientation higher
values for the a l l tensor component. For each set of than 0.6. At the core, the fiber alignment is high in
processing conditions the calculation uses data from the across flow direction both in the simulations pre-
three locations along the flow path (20,35and 50 mm dictions and in the experimental data, as it is sug-
distance from the gate) and twelve locations across gested by the large value of aZ2(small value of a , ,).
the thickness: The experimental results suggest a less defined lay-
ering through-thickness. At most of the data points
Deviation =
i2
I= I J- 1
lall,re.,,
(i.j)- (i.j)
alle,,,> 1 (8)
the a l l tensor component is lower than 0.5. These re-
sults c o n f m that the disc geometry, having a diver-
3 x 12 gent flow, induces higher fiber orientation in the
across flow direction.
The subscripts exp and pred refer to experimental and The simulations with C-Mold seem to predict quali-
predicted results, respectively. The indices i and j refer tatively well the fiber orientation close or far from the
to the location along the flow path and to the layers gate. At intermediate locations (in this case, 35 mm
across the thickness, respectively. from the gate) the degree of fiber orientation in the
flow direction is overestimated regardless the molding
4. RESULTS AND DISCUSSION conditions as it is summarized in the Fig. 6.
The results of the experimental fiber orientation 4.2. Effect of the Melt Temperature
characterization are presented in terms of the a l l sec-
ond order tensor component. This tensor component Increasing the melt temperature (Fig. 7) leads to
characterizes the fraction of fibers aligned in the flow lower levels of fiber orientation in the flow direction.
direction. The C-Mold software predicts a predomi- This is in accordance with the expectations, since the
nantly planar fiber orientation pattern with very small flow-induced stresses acting in the melt are propor-
values for the out-of-plane component of the tensor of tional to the viscosity. Lower melt temperature in-
fiber orientation (a33= 0.005). Thus, the transverse- creases and thus the alignment of the fibers. Experi-
flow-direction fiber orientation at any point in the re- mental data shows that molding with higher melt
sults may be seen as the difference between a,, and
the unity la,, = l-all). through-thickness fiber orientation (all -, -
temperatures results in close to random in-plane
0.5).
The relatively better agreement between predictions
and experimental data that are observed at the lower
4.1. Variation Along the Flow Path
melt temperature conditions (280.10, 280.14 and
Both experimental data and predictions correspond- 280.32) may be attributable to the stronger stress
ing to the processing condition code 280.10 considering field built up in the melt during the flow. Moreover, it
different C,are shown in Q. 5, at three locations along seems that in C-Mold the effect of shear stresses are
the flow path. The C-Mold predictions show a skin with more effectively predicted than the effect of exten-
fiber orientation predominant in the flow direction, with sional stresses.
4
Flg. 5. Prediction and experimentalfber orientation in theflow direction at 280°C melt temperature and 9.7 cm3/sfzow rate, at three
distancesfrom the gate: a) 20 mm, b] 35 mm and c) 50 nun
4.3. Effect of the Flow Rate and 280.32). This is understandable because the
At the lowest melt temperature (Fig. 8) a clear ten- shear flow is better modeled and probably overesti-
dency of the fibers to get more aligned in the across mated by the C-Mold software. As a general observa-
flow direction is observed (all < 0.5). A slower flow tion it can be said that the effect of C , on the predic-
rate seems to favor the fiber orientation in the flow di- tions is qualitatively important, but it is not possible
rection. The fiber orientation in the flow direction is to find a value of C, that might lead to general qualita-
overpredicted close to the wall and at the core of the tive and quantitative close prediction of the fiber ori-
moldings. entation in the moldings.
0.4 I
20 mm I n
- 0-
0 0
C ] = 0.1 0
0.3
. . . . . . . . . . . . . . .
0
0
0.2 0 -
B 0
0- ~~ . . . . . .
A A
A
n ,
0.1 I I l
................................................
-
0
..........................
0.8 C-MoldCI4.01 - D C M . Y C I = O .
0.6
-
I
0 ' I
-0.5 -0.25 0 0.25 0.5
Fnrtiomldwtance to the midplane
(c)
A Epnclcd.l -O-CM.UCI=O.l
0.8 -
CMbUCI=Wl -C)-C-MoUCI=O
-
"
0.4 -
-
8
(a) (b)
0.8
C-MddCI=O.Ol -O-C-M&K.i=O.W
0.6
-
0.4 Fig. 8. Prediction and experimental f h e r orientation
in theflow direction at 280°C melt temperature and
0.2
at 20 mm from the gate, at differentflow rates:
a) 9.7 cm3/s, b) 13.8 cm3/s and c) 32.2 cm3/s.
0
-0.5 -0.25 0 0.25 0.5
Fractional distance to the midplane
(c)
l?g.9. Deviation between C-Mold predictions using duerent C, and experimental results.
glass fiber polycarbonate seems to be better predicted Department of Polymer Engineering of the Universi-
when a large value of C, is used. When the value of dade do Minho, Portugal. The authors also acknowl-
this interaction coefficient is decided from using the edge the assistance provided by the C-MOLD Europe
empirical expression suggested by Tucker (4). poor office at Enschede, The Netherlands.
predictions result for lightly reinforced materials in
this molding geometry. REFERENCES
The predictions close to the gate or at the end of the 1. N. M. Neves, G. Isdell. A. S. Pouzada, and P. C. Powell.
flow path are much closer to the experimental results Polym Comp., 19,5 (1998).
than at intermediate positions in the flow path. This 2. G. B. Jeffery, Roc. Roy. Soc., A102 (1922).
unexpected result suggests that further investigation 3. F. Folgar and C. L. Tucker 111, J . ofReinf. Plast. and
Cornp., 9 ( 1984).
of the specific flow conditions and fiber orientation in 4. C. L. Tucker 111 and S. G. Advani, Processing of Short-
this region is necessary. Fiber Systems, In: s. G. Advani, (Ed.), Flow and Rheol-
For this material and molding geometry, tuning the ogy in Polymer Composites Manufacturing, Comp. Mater.
interaction coefficient does not lend to an overall bet- Series, Amsterdam (1994).
ter fit of predictions to experimental data. 5. S. G. Advani and C. L. Tucker 111, J. ofRheology, 31,8
( 1987).
It was shown that the improvement of the predic- 6. X. J. Fan. N. Phan-Thien, and R. Zheng. J . Non-Newtonian
tion of the fiber orientation requires the definition of Fluid Mech., 74 (1998).
an adequate interaction coefficient for each case. Un- 7. M. C . Altan, S. Subbiah, S. 1. Guceri. and R. B. Pipes,
fortunately there is not yet a clear criterion for its Polym Ehg. Sci, 30,14 (1990).
8. K. K. Kabanemi, J. F. Hetu, and A. Garcia-Rejon, Intern.
specification. Using the direct numerical simulation of Polym ROC., 12.2 (1997).
fiber-fiber interaction proposed by Fan et al. (6) may 9. M. L. Becraft and A. B. Metzner, J . Rheology. 36 (1992).
overcome the uncertainty in choosing of C,, but it is 10. R. S . Bay and C. L. Tucker 111. Polym Cornp., 13,4
not yet a commercially available simulation program. ( 1992).
Nevertheless we may conclude that, for the case of 11. J. S. Cintra and C. L. Tucker 111. J. Rheology, 39 (1995).
12. M. R. Kamal and A. T. Mutel, Polym Comp., 10 (1989).
geometries in which extensional stresses are high 13. S. Ranganathan and S. S. G. Advani, J. Rheology, 35
(such as center gated parts), values of C, much higher (199 1).
than those obtained with t h e Tucker and Advani 14. R. S. Bay and C. L. Tucker 111, Polym Eng. Sci., 32.4
equation should be used, for getting patterns closer to (1992).
15. C-Mold version 96.10, Advanced CAE Technology, Inc.,
the random in-plane state. Ithaca, N.Y.
8. ACXNOWLEDGMENTS
The authors are indebted to IMAT-Institute of Mate-
rials, and the Center for Polymer Engineering of the