Académique Documents
Professionnel Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: The finite element method (FEM) is a powerful computational technique for approximate solutions to a
Received 25 December 2009 variety of ‘‘real-world” engineering problems having complex domains subjected to general boundary
Received in revised form 10 February 2010 conditions. In this paper FEM is implemented to determine the effective thermal conductivity of partic-
Accepted 16 February 2010
ulate filled polymer composites and is validated by experimentation. A commercially available finite-ele-
Available online 12 March 2010
ment package ANSYS is used to for this numerical analysis. Three-dimensional spheres-in-cube lattice
array models are constructed to simulate the microstructure of composite materials for various filler con-
Keywords:
centrations ranging from about 6 to 36 vol.%. Composites with similar filler contents are fabricated by
Polymer composite
Pine wood dust reinforcement
hand layup technique by reinforcing micro-sized pine wood dust in epoxy resin. Guarded heat flow meter
Thermal conductivity test method is used to measure the thermal conductivity of these composites using the instrument Uni-
Finite-element analysis therm™ Model 2022 as per ASTM-E1530. This study shows that the incorporation of pine wood dust
results in reduction of conductivity of epoxy resin and thereby improves its thermal insulation capability.
With addition of 6.5 vol.% of filler, the thermal conductivity of epoxy is found to decrease by about 19.8%
and with about 36 vol.% of filler addition, a 57.3% reduction in thermal conductivity of neat epoxy is
achieved. The experimentally measured conductivity values are compared with the numerically calcu-
lated ones and also with the existing theoretical and empirical models. The values obtained using
finite-element analysis (FEA) are found to be in reasonable agreement with the experimental values.
Ó 2010 Elsevier B.V. All rights reserved.
0927-0256/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.commatsci.2010.02.025
R. Nayak et al. / Computational Materials Science 48 (2010) 576–581 577
well as numerical and analytical studies on thermal conductivity of which describe the spatial distribution of the two phases. By
some filled polymer composites [26–38]. The fillers most fre- assuming a parabolic distribution of the discontinuous phase in
quently used are aluminum particles, copper particles, brass parti- the continuous phase, Cheng and Vachon [48] obtained a solution
cles, short carbon fiber, carbon particles, graphite, aluminum to Tsao’s [47] model that did not require knowledge of additional
nitrides and magnetite particles. Progelhof et al. [26] was the first parameters. Agari and Uno [49] propose a new model for filled
to present an exhaustive overview on models and methods for pre- polymers, which takes into account parallel and series conduction
dicting the thermal conductivity of composite systems. Procter and mechanisms. According to this model, the expression that governs
Solc [39] used Nielsen model as a prediction to investigate the the thermal conductivity of the composite is:
thermal conductivity of several types of polymer composites filled
with different fillers and confirmed its applicability. Nagai and Lai
log kc ¼ ØC 2 log kf þ ð1 ØÞ logðC 1 km Þ ð3Þ
[40] found that Bruggeman model for Al2O3/epoxy system and a
modified form of Bruggeman model for AlN/epoxy system are both
good prediction theories for thermal conductivity. Griesinger et al. where C1, C2 are experimentally determined constants of order
[41] reported that thermal conductivity of low-density PE in- unity. C1 is a measure of the effect of the particles on the secondary
creased from 0.35 for an isotropic sample, to the value of 50 W/ structure of the polymer, like crystallinity and the crystal size of the
m K for a sample with an orientation ratio of 50. The thermal polymer. C2 measures the ease of the particles to form conductive
and mechanical properties of copper powder filled poly-ethylene chains. The more easily particles are gathered to form conductive
composites are found by Tavman [25] while Sofian et al. [37] inves- chains, the more thermal conductivity of the particles contributes
tigated experimentally on thermal properties such as thermal con- to change in thermal conductivity of the composite and C2 becomes
ductivity, thermal diffusivity and specific heat of metal (copper, closer to 1. Later, they modified the model to take into account the
zinc, iron, and bronze) powder filled HDPE composites in the range shape of the particles [50]. Generally, this semi-empirical model
of filler content 0–24% by volume. They observed a moderate in- seems to fit the experimental data well. However, adequate exper-
crease in thermal conductivity up to 16% of metal powder filler imental data is needed for each type of composite in order to deter-
content. Mamunya et al. [42] also reported the improvement in mine the necessary constants.
electrical and thermal conductivity of polymers filled with metal For an infinitely dilute composite of spherical particles, the ex-
powders. In a recent research Weidenfeller et al. [43] studied the act expression for the effective thermal conductivity is given as:
effect of the interconnectivity of the filler particles and its impor-
tant role in the thermal conductivity of the composites. They pre-
k kd kc
pared PP samples with different commercially available fillers by ¼1þ3 ð4Þ
kc kd þ 2kc
extrusion and injection molding using various volume fractions
of filler content to systematically vary density and thermal trans-
port properties of these composites. Surprisingly, they measured where k, kc and kd are thermal conductivities of composite, contin-
that the thermal conductivity of the PP has increased from 0.27 uous-phase (matrix), and dispersed-phase (filler), respectively, and
up to 2.5 W/m K with 30 vol.% talc in the PP matrix, while the same Ø is the volume fraction of the dispersed-phase. Eq. (4) is the well-
matrix material containing the same volume fraction of copper known Maxwell equation [51] for dilute composites.
particles had a thermal conductivity of only 1.25 W/m K despite An exhaustive review of the published literature reveals that
the fact that copper particles have a thermal conductivity approx- most of the investigations are aimed at enhancing the thermal con-
imately 40 times greater than that of talc particles. Tekce et al. [38] ductivity of the polymer rather than attempting to improve its
noticed the strong influence of the shape factor of fillers on thermal insulation capabilities. Although a large number of particulates
conductivity of the composite. While Kumlutas and Tavman [44] have been used as fillers in the past, there is no report available
carried out a numerical and experimental study on thermal con- on bio-based materials like any kind of wood dust being used for
ductivity of particle filled polymer composites, Amar et al. reported composite making. Investigation on thermal conductivity of partic-
the existence of a possible correlation between thermal conductiv- ulate filled fiber-reinforced composites is rare and the understand-
ity and wear resistance of particulate filled composites [45]. ing of the relationship between the effective thermal conductivity
Many theoretical and empirical models have been proposed to of a composite material and the micro-structural properties (vol-
predict the effective thermal conductivity of two-phase mixtures. ume fractions, distribution of particles, aggregation of particles,
Comprehensive review articles have discussed the applicability of properties of individual components, etc.) is far from satisfactory.
many of these models [46,26]. For a two-component composite, Besides, though it becomes clear that improved thermal conductiv-
the simplest alternatives would be with the materials arranged ity in polymers may be achieved either by molecular orientation or
in either parallel or series with respect to heat flow, which gives by the addition of conductive fillers, it is yet to be seen how the
the upper or lower bounds of effective thermal conductivity. For incorporation of bio-based particulates with poor heat conductiv-
the parallel conduction model: ity affects the overall conductivity of any polymer composite.
In view of the above, the present work is undertaken to investi-
kc ¼ ð1 ØÞkm þ Økf ð1Þ gate numerically and experimentally the thermal conductivity of
epoxy matrix composites filled with pine wood dust. The objec-
where kc, km, kf are the thermal conductivities of the composite, the
tives of this work include fabrication of a new class of low cost
matrix and the filler respectively and Ø is the volume fraction of
composites using pine wood dust as the reinforcing filler with an
filler.
objective to improve the insulating properties of epoxy resin.
For series conduction model:
Epoxy is already a known thermal barrier. Since this work aims
1 1Ø Ø at developing some kind of a light, cheap and insulating material,
¼ þ ð2Þ epoxy emerged as the first choice for the matrix material. Epoxy
kc km kf
is chosen primarily because it happens to be the most commonly
The correlations presented by Eqs. (1) and (2) are derived on the used polymer and because of its insulating nature (low value of
basis of the rules-of-mixture. Tsao [47] derived an equation relat- thermal conductivity, about 0.363 W/m K). It also reports the esti-
ing the two-phase solid mixture thermal conductivity to the con- mation of the equivalent thermal conductivity of particulate–poly-
ductivity of the individual components and to two parameters mer composite system using finite element method.
578 R. Nayak et al. / Computational Materials Science 48 (2010) 576–581
defined in terms of the values of the field variables at specific typical three-dimensional sphere-in-cube model with a body
points, referred to as nodes. Nodes are usually located along the centered cubic (BCC) structural arrangement for the composite
element boundaries, and they connect adjacent elements. The abil- with pine wood dust concentration of 35.9 vol.% is illustrated
ity to discretize the irregular domains with finite elements makes in Fig. 2. The temperature profiles obtained from FEM analysis
the method a valuable and practical analysis tool for the solution of for the composites with particulate concentrations of 6.5, 11.3,
boundary, initial, and eigenvalue problems arising in various engi- 26.8 and 35.9 vol.% are presented in Fig. 3a–d respectively.
neering disciplines. The values of effective thermal conductivities of the particulate
The FEM is a numerical procedure that can be used to obtain filled epoxy composites with varied proportions of pine wood dust
solutions to a large class of engineering problems involving stress obtained using Maxwell’s correlation, rules-of-mixture model and
analysis, heat transfer, fluid flow etc. ANSYS is general-purpose fi- those obtained from FEM analysis are presented in Table 2. It pre-
nite-element modeling package for numerically solving a wide sents a comparison among the results obtained using these models
variety of mechanical problems that include static/dynamic, struc- with regard to the values of effective conductivity obtained
tural analysis (both linear and nonlinear), heat transfer, and fluid experimentally.
problems, as well as acoustic and electromagnetic problems. Fig. 4 presents comparative picture of the thermal conductiv-
ity values obtained from different methods. It is noticed that the
results obtained from the finite-element analysis using ANSYS
4. Results and discussion
are closer to the measured values of effective thermal conduc-
tivity for composites of different filler content. On comparison,
4.1. Numerical analysis
it is found that while the errors associated with the FEM values
with respect to the experimental ones lie in the range of 3–13%,
Using the finite-element program ANSYS, thermal analysis is
the same for results from rules-of-mixture and Maxwell’s corre-
carried out for the conductive heat transfer through the composite
lation lie in the ranges of 2–44% and 4–28% respectively. The
body. In order to make a thermal analysis, three-dimensional phys-
percentage errors associated with each method for individual
ical models with spheres-in-a-cube lattice array have been used to
composites are given as Table 3. It is further noted that while
simulate the microstructure of composite materials for four differ-
the FEM and Maxwell’s model overestimate the value of thermal
ent filler concentrations. Furthermore, the effective thermal con-
conductivity the rules-of-mixture model underestimates the
ductivities of these epoxy composites filled with pine wood dust
up to about 36% by volume is numerically determined using
ANSYS.
Fig. 3. Temperature profiles for composite with particle concentration of: (a) 6.5 vol.%, (b) 11.3 vol.%, (c) 26.8 vol.%, (d) 35.9 vol.%.
Table 2
Thermal conductivity values for composites obtained from different methods.
Sample Particulate content (vol.%) Effective thermal conductivity of the composites (W/m K)
Rule of mixture model Maxwell’s model FEM model Experimental value
1 6.5 0.2831 0.336 0.300 0.291
2 11.3 0.2436 0.317 0.315 0.273
3 26.8 0.1678 0.254 0.270 0.242
4 35.9 0.1317 0.217 0.170 0.155
Table 3
Percentage errors with respected to the measured value.
Model Particulate content (vol.%) Percentage errors with respected to the measured value
Rule of mixture model (%) Maxwell’s model (%) FEM model (%)
1 6.5 2.79 13.3929 3.0000
2 11.3 12.06 13.8801 13.3333
3 26.8 44.21 4.7244 10.3704
4 35.9 17.69 28.5714 8.8235