Journal of Thermoplastic

Composite Materials
http://jtc.sagepub.com/

Micro- and macro-impregnation of fabrics using thermoplastic matrices

Riccardo Gennaro, Antonio Greco and Alfonso Maffezzoli
Journal of Thermoplastic Composite Materials published online 14 November 2011
DOI: 10.1177/0892705711425849

The online version of this article can be found at:

http://jtc.sagepub.com/content/early/2011/11/13/0892705711425849
A more recent version of this article was published on - Apr 24, 2013

Published by:

http://www.sagepublications.com

Additional services and information for Journal of Thermoplastic Composite Materials can be found
at:

Email Alerts: http://jtc.sagepub.com/cgi/alerts

Subscriptions: http://jtc.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Version of Record - Apr 24, 2013

>> OnlineFirst Version of Record - Nov 14, 2011

What is This?

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

 Journal of Thermoplastic Composite
Materials
Micro- and 1–17
ª The Author(s) 2011
macro-impregnation Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav

of fabrics using DOI: 10.1177/0892705711425849

jtc.sagepub.com

thermoplastic matrices
Riccardo Gennaro1, Antonio Greco1 and
Alfonso Maffezzoli1

Abstract
In this work, a method developed for the measurement of the transversal permeability of
fibrous reinforcement is presented. The permeability of a reinforcement is defined by the
Darcy equation and can be obtained once the pressure drop through the reinforcement
and the viscosity and average velocity of the fluid are known. The method used in this work
is based on a proper modification of a capillary rheometer, obtained by substituting the
capillary with a tool, capable of sustaining the reinforcement during reinforcement
impregnation and through thickness flow. The developed device was used to measure the
pressure built during the flow at different velocities of the rheometer piston. The
impregnation tests were performed at different temperatures using a high-viscosity matrix
characterized by a Newtonian behaviour. At each temperature, pressure versus velocity
plots showed two distinct zones, each characterized by a different slope. The slope
observed at low pressures was higher than the slope observed at pressures, suggesting
an increase in the permeability with increasing pressure or velocity. The double slope was
attributed to the existence of two different impregnation mechanisms, the first one being
characteristic of the flow of the matrix around the reinforcement bundles and the second
is the characteristic of the flow of the matrix inside each bundle. Dimensionless analysis
models and scanning electron micrographs were used to support that the slope of the first
portion of the plot is due to inter-bundle flow, whereas the slope of the second portion is
due to global flow including both inter- and intra-bundle flow.

Keywords
thermoplastic matrix composite, impregnation, Newtonian flow, transversal permeability

1
Department of Innovation Engineering, University of Salento, Via per Arnesano, Lecce, Italy

Corresponding author:
Antonio Greco, Department of Innovation Engineering, University of Salento, Via per Arnesano, Lecce 73100, Italy
Email: antonio.greco@unisalento.it

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

2 Journal of Thermoplastic Composite Materials

Introduction
Transversal permeability of textiles is a key characteristic for composite manufacturing
and in particular for the resin infusion processes where the through-thickness impregna-
tion is promoted by resin distribution nets. Permeability has posed a major challenge to
researchers and engineers in a wide range of industrial and academic disciplines. These
include, but are not limited to, resin transfer moulding,1–6 pultrusion and autoclave
processing,7 biomedical engineering,8,9 subsurface flow of oil and groundwater,10,11
filter12,13 and fuel cell simulation.14,15
The permeability of a porous medium can be defined as a tensorial quantity, which
relates the local velocity vector of the fluid flow with the associated pressure gradient. In
a monodimensional flow perpendicular to the plane of a composite material reinforce-
ment, as occurring in compression moulding or vacuum moulding of thermoplastic
matrix composites, the permeability K of the medium can be calculated using the Darcy
law,16 strictly valid for Newtonian fluids17:

DP
v¼k ð1Þ
ZL
In which Z is the Newtonian viscosity of the fluid, DP is the pressure drop across
fabric thickness and L is the thickness of reinforcement. In composite manufacturing
processes involving fibre impregnation, permeability, which has the dimensions of m2, is
a characteristic property of the reinforcing fibres for a given reinforcement volume
fraction.
Significant efforts were made in the past decade to develop computational tools which
can be used, in conjunction with experimental measurements, to determine the perme-
ability of reinforcement preforms used in liquid infusion processing of composite
materials. A successful computational modelling approach capable of predicting fabric
permeability must include both the actual architecture of the fabric and the basic physics
of the flow through it. Prediction of the permeability of porous media has been the
subject of intense research for the last three decades. In 1986, Johnson et al.18 introduced
a transport parameter, based on the electrical conduction principles and directly linked to
viscous flow, which is capable of leading to permeability from more readily measurable
physical properties, such as the porosity and specific surface area. Johnson et al.’s
method was tested numerically and experimentally, with positive results in most cases; it
was found to give very good permeability estimates for various types of porous media
over broad ranges of porosity, though with some exceptions. In 2004, Krawczak et al.19
performed unidirectional compression tests on impregnated fibre reinforcements to
induce in-plane and through-thickness fluid flows. The purpose of the methodology is to
measure hydraulic permeabilities using unidirectional compression experiments. This
methodology is based on the measurement of the total stress szz and the effective stress
s0zz during unidirectional compression test. Then the fluid pressure p can be extracted
and exploited to back calculate the fibre reinforcement permeability. In 2005, Drapier
et al.20 presented the measurement of the through-thickness permeability in continuous
fibre-reinforced composites, more precisely in new multiaxial stitched semi-products,

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

Gennaro et al. 3

developed by Hexcel, and named Non Crimp New Concept (NC2) multiaxial fabrics.21
This study presents the measurement of the through-thickness permeability of a selected
NC2 semi-product intended for automotive applications. Relying on optical fibre sen-
sors, the transient permeability is assessed and appears to be consistent with the measure-
ments in saturated regime presented in Ref. 22. In 2007, Gillespie et al. introduced an
apparatus to measure the saturated permeability of fibrous preforms using both gaseous
and liquid flow. This study proposed a transverse permeability measurement system with
a variable thickness mould cavity that overcomes the limitations of earlier efforts. The
setup creates a uniform one-dimensional flow through the thickness of the reinforcement
by integrating a high-permeability layer on the mould surfaces. A wide range of perme-
ability as a function of fibre volume fraction can be measured in one experiment, while
applying a known load under a hydraulic testing machine.23 The measurement of trans-
versal permeability during filling of fibrous composite has been evaluated by several
authors.24–26 This method consists of evaluating pressure/flow rate couples at a given
fibre volume fraction after injecting a test fluid through the fibrous preforms. In these
studies, the tests were performed using low-viscosity (0.1 Pa s) Newtonian fluids.
On the other hand, in conventional processing of composite materials, impregnation
occurs at different scale levels, that is macro- and micro-impregnation. Macro-
impregnation is the result of matrix flow between bundles or yarns and involves elimi-
nation of large-scale dry spots. Micro-impregnation is the result of matrix flow inside
each bundles, around individual fibres, and affects microporosity and the quality of the
fibre–matrix interface.27–32
In this article, a continuous method for the measurement of transversal permeability
during the impregnation of fibrous reinforcement with a high-viscosity thermoplastic
matrix has been studied. The model system used is composed of a unidirectional flax
woven fabric and ethylene vinyl acetate (EVA). Rheological analysis performed on the
matrix at different temperatures showed that the matrix is characterized by a Newtonian
behaviour.
Optical micrographs of composites were used to determine the fibre volume fraction.
The permeability of the reinforcement was measured by a plot of the measured pressure
as a function of the velocity set during experiments performed with a properly modified
capillary rheometer. The resulting structure of the composite was studied by optical
microscopy. The volume fraction of fibres was used for the prediction of reinforcement
permeability applying proper mathematical models.

Materials and methods

The thermoplastic matrix used is a thermoplastic blend of low EVA and wax with a
density 1040 kg/m3 provided by Axel (Italy). The matrix is characterized by a low
melting point (in the range 50–80 C) evaluated by a calorimetric analysis (DSC
METTLER TOLEDO 877). The fibrous reinforcement used is a unidirectional flax
woven fabric 300 HS 45 provided by Fidia (Italy).
An in-plane image of the fabric, showing its architecture, obtained by optical micro-
scopy, is reported in Figure 1. Here the double-scale architecture of the fabric is well

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

4 Journal of Thermoplastic Composite Materials

Figure 1. Optical microscopy image of flax fabric.

evident. The single filaments of fibres are arranged in bundles, which constitute the warp
and weft elements of the fabric.
Matrix viscosity was measured in steady state using a cone and plate rheometer ARES
II from TA Instrument, at four different temperature (85, 90, 95, 100 C) in a range of
shear rate between 0.1 and 10 s1.
Scanning electron microscopy ([SEM] EVO 60 ZEISS) was performed on samples
consolidated at different temperatures in order to measure the fibre–matrix distribution.
A Nikon Model Epiphot 200 was used for optical microscopy analysis in order to
evaluate the inter- and intra-bundle volume fractions. Specimens cut from the centre part
of composite were polished with SiC paper disks and then with diamond particle solu-
tions before optical analysis. Energy dispersive X-ray spectroscopy (EDS) analysis
microprobe (Bruker 127 eV) was used to identify the elemental composition of compo-
site, in order to evaluate the degree of impregnation.

Capillary rheometer experiments

The experimental measurements of impregnation were carried out on samples made of
two layers of flax at different temperatures, using a REO-TESTER 1000 GOTTFERT
capillary rheometer equipped with a 20-bar pressure transducer. This instrument was
modified by substituting the capillary with a tool, holding the reinforcement during
impregnation. A sketch of the equipment used for impregnation of woven glass fibre
is reported in Figure 2. It is made of a capillary rheometer (g), equipped with pressure
transducer (f). The rheometer is modified by substituting the capillary with a tool (e),
supporting the reinforcement (c) during impregnation. During the tests, performed at
four different temperatures (85, 90, 95, 100 C), a piston (a) moves at a constant speed

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

Gennaro et al. 5

Figure 2. Sketch of the equipment developed for the impregnation of woven glass fibre.

downward forcing the molten polymer (b) through the fabric (c). The fabric is clamped
between the tool (e) and the lower surface of the reservoir by means of a threaded nut (d).
A teflon ring (h) is placed between the fabric and the metallic parts to prevent material
leakage.
The pressure transducer measures the values of pressure drop (Dp) above the fabric.
The average velocity of matrix through the reinforcement was obtained as:

v p  Sp ¼ v M  SM ð2Þ

where:

 vp is the velocity of piston,

 Sp is the cross section of the piston or the cylinder,
 VM is the average velocity of matrix through the thickness of woven fabric, and
 SM is the cross section of the woven fabric impregnated.

A typical plot of VM and pressure build as a function of time is reported in Figure 3.

The velocity, imposed as independent variable, was set at different values. In corre-
spondence of each increase in velocity step, the pressure increases in a continuous way,
reaching a plateau value after about 50–100 s. The time necessary for the pressure to
reach the plateau is the time of transient behaviour of the fluid, necessary to reach again
the steady state condition but at higher flow rate. During the transient period, Darcy

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

6 Journal of Thermoplastic Composite Materials

30
0.8 velocity of the piston

velocity set of the piston [µm/s]

experimental pressure [MPa] experimental pressure 25
0.6
20
0.5
15

0.3
10

0.2 5

0.0 0
0 500 1000 1500 2000 2500
time [s]

Figure 3. Velocity of the matrix and experimental pressure versus time.

equation, which is derived in steady state conditions, neglecting all acceleration and or/
convection terms in the momentum continuity equation cannot be applied. After the
transient interval, the plateau value of pressure corresponds to a steady state behaviour
and Darcy equation can again be applied. Therefore, the permeability K of the fabric was
calculated, according to Eq. (1), from the slope, B of the pressure–velocity plot:

K ¼ ZL=B ð3Þ

The pressure measured during a capillary test is actually the sum of two contributions:
the pressure drop inside the empty tool of the instrument (indicated as e in Figure 2) and
the pressure drop in the fabric (indicated as c in Figure 2). In Eq. (3), only the contri-
bution of pressure drop in the fabric must be used. For this reason, a calibration
experiment was performed without the fabric, in order to determine the pressure drop
into the empty tool. The results, reported in Figure 4, show that the pressure drop inside
the tool is comparable to the lower sensitivity limit of the pressure transducer (9E-4 bar)
indicating that:

DPTOOL  DPTOT ) DPTOT ffi DPFABRIC ð4Þ

where:

 DPTOOL is the pressure drop through the empty tool;

 DPTOT is the total pressure drop;
 DPFABRIC is the pressure drop through the fabric.

Therefore, the total pressure drop can be used for the determination of the fabric
permeability according to Eqs. (1) and (3).

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

Gennaro et al. 7

Figure 4. Velocity of the piston and experimental pressure versus time in a test without fabric.

360 85°C 90°C

320 100°C 95°C
280
240
200
η[Pa*s]

160

120

80

0.1 1 10
shear rate [s−1]

Figure 5. Viscosity versus shear rate at four different temperatures.

Results and discussion

The results of rheological measurements, reported in Figure 5, show the evolution of
viscosity as a function of shear rate for the matrix at four different temperatures. At each
temperature, the matrix shows a weak shear thinning behaviour, which can be reasonably
approximated by a Newtonian behaviour, with the characteristic values of viscosities
reported in Table 1. As expected, the viscosity decreases with increasing temperature.
The Newtonian behaviour of the matrix is necessary for obtaining a proper definition of
the fabric permeability, independent from the rheological properties of the matrix. In
facts, for non-Newtonian flows, it has been shown that it is not possible to define a

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

8 Journal of Thermoplastic Composite Materials

Table 1. Average viscosity of ethylene vinyl acetate (EVA) and wax measured at different
temperatures

Temperature ( C) Average viscosity (Pa s)

85 254.9 + 14.5
90 136.8 + 4.7
95 116.8 + 4.8
100 88.5 + 3.4

1.2
average experimental pressure [MPa]

I II

0.8

0.4
85°C
90°C
95°C
linear fit according Darcy Law 100°C
0.0
0 20 40 60
average velocity matrix [µm/s]

Figure 6. Measured pressure as a function of the average matrix velocity at four different
temperatures.

permeability according to Darcy equation.16 On the other hand, the viscosity of the
matrix is much higher than those previously used with similar approaches,25 being in the
range of the viscosities commonly used in thermoplastic matrix composites.
The measured pressure as a function of matrix average velocity through the flax fabric
is reported in Figure 6. The slope of the curves decreases with increasing temperature, as
a consequence of the decrease in the matrix viscosity. On the other hand, according to
Darcy law, the plot should yield a linear behaviour in the whole range of pressures. As it
can be observed in Figure 6, the plot at each temperature is characterized by two distinct
linear behaviour zones. The low-pressure zone is characterized by a higher slope,
whereas the zone at high pressures is characterized by a lower slope. According to
Eq. (3), this indicates an increase in the permeability.

Three different explanations can be proposed:

a. The behaviour of the matrix is not completely Newtonian. Therefore, an increase in the
average velocity yields an increase in the shear rate during impregnation and a decrease

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

Gennaro et al. 9

1.2
experimental data at 90°C

average experimental pressure [MPa]

non linear curve fitting, eq.5)

0.8

0.4

0.0
0 20 40 60
average velocity matrix [µm/s]

Figure 7. Nonlinear curve fitting of the pressure–velocity data at 90 C according to Eq. (5).

in the matrix viscosity. A decrease in the matrix viscosity, for a fixed value of perme-
ability, yields a decrease in the slope of the pressure–velocity plot.12 In this case, the
pressure rise during impregnation should be related to the velocity by Eq. (5)17,33:
 1
DP n
vavg ¼ K ð5Þ
Z0 L

where n is the viscosity index of the matrix. The nonlinear curve fitting according to Eq.
(5) of the data at 90 C is reported in Figure 7, and obtained with a coefficient n ¼ 0.64,
which is significantly different from the value obtained by fitting of viscosity data
according to the power law model, which was evaluated to be close to 1 (n ¼ 0.98).

b. Being the fluid non-wetting, capillary pressure of the fibre bundles need to be over-
come to fully wet them. The relevancy of capillary flow compared to viscous flow
can be estimated by calculating the capillary number.34–36
Zv
Ca ¼ ð6Þ
g

Where Z is the viscosity of the matrix, v is the characteristic velocity and g is the
surface or interfacial tension between fibre and matrix, using a value of Z ¼ 100 Pa s
(from Figure 5), a velocity of 10 mm/s (Figure 4) and g ¼ 0.02 N m, a value of Ca ¼ 0.05
was obtained. This value is about 2–3 orders of magnitude higher than the value of
Ca ¼ 104, which according to the work reported by Ermanni and coworkers,35 can be
considered a threshold value below which the capillary flux becomes relevant. As a
further demonstration of the hypothesis that the main contribution to the matrix flow is
given by externally applied pressure, the capillary pressure was calculated as35:

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

10 Journal of Thermoplastic Composite Materials

2g
PCAP ¼ cos y ð7Þ
r
Where r is the capillary radius, which can be approximated as half of the distance
between the fibres. Using a value of r ¼ 2 mm, calculated from a statistical analysis of
optical micrographs and taking cos y ¼ 1, a value of PCAP ¼ 0.02 MPa was calculated.
This value is at least one order of magnitude lower than the lower pressure measured
during the experiments, which confirms that capillary pressure is negligible compared to
the externally applied pressure.

c. The impregnation of the fabric is actually governed by two different mechanisms,

the first one occurring at lower pressures only involves impregnation among bun-
dles, and the second one occurring only at the higher pressures is associated with the
impregnation inside bundles. Each of the two mechanisms is characterized by a per-
meability values.

In order to prove hypothesis (b), a test was performed in the capillary rheometer at
90 C and limiting the pressure drop below 0.47 MPa, which is a value below the slope
variation observed in Figure 6. A SEM image of the fabric after impregnation at low
pressures is reported in Figure 8a and b. It is evident that the matrix has flown around
bundles, a ‘macro-impregnation’ process, while there is no evidence of flow of the
matrix inside bundles. A SEM image of the fully impregnated fabric up to 1 MPa
pressure is reported in Figure 8c and d, showing that impregnation occurred inside the
bundles, a ‘micro-impregnation’ process.
According to the results shown in Figures 6 through 8c, d, macro-impregnation takes
place at the lower pressures, whereas only a threshold value of pressure is achieved, both
the macro- and micro-impregnation processes become active. Therefore, the low-
pressure slope observed in Figure 6 can be associated with permeability at macro-
scale or macro-impregnation permeability, whereas the high-pressure slope reflects the
global permeability, which takes into account the two mechanisms of impregnation.
In order to further support this micro/macro-scale impregnation process, EDS analy-
sis was performed on the sample of the fabric impregnated at low pressure. The micro-
analysis revealed that the carbon amount on the fibre was about 68%, whereas the carbon
amount on the matrix was 95%. The carbon amount in the intra-bundle voids shown in
Figure 8d is the same, 68%, either on the flax fibre or the intra-bundle porosity, indicat-
ing that the EDS microprobe cannot detect any matrix among the fibres.
The values of inter-bundle permeability and global permeability determined from the
data shown in Figure 6 according to Eq. (3), using the viscosity values of Table 1 and a
thickness of fabric equal to 1.63E3[m] which are reported in Table 2.
Transversal permeability as a function of temperature in the low velocity and high
velocity ranges was reported in Figure 9.
As expected, both the inter-bundle and global permeabilities are not dependent on the
test temperature, permeability being a characteristic property of the fabric, independent
on the rheological properties of the matrix.

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

Gennaro et al. 11

Figure 8. (a and b) Scanning electron microscope (SEM) micrographs of polished cross section of
the fabric after impregnation at 90 C and below 0.47 MPa of pressure. (c and d) SEM micrographs
of polished cross section of the unidirectional composite ethylene vinyl acetate (EVA)/flax, after
impregnation up to 1 MPa pressure.

The global K changes are reasonable, considering for instance that Lundstrom et al.37
presented 15% repeatability. The authors state that this limited repeatability is attribu-
table to the fabric handling. The average experimental global permeability (Table 2) is
equal to 1.24E11 + 1.86E12 (m2). As a conclusion, the experimental global per-
meability of flax woven fabric 300 HS 45 result is the order of 1011 (m2) (Table 2).
At each temperature, the permeability obtained in the first zone, associated with the
inter-bundle impregnation, is lower than that obtained in the second zone, attributed to
the sum of inter- and intra-bundle impregnation processes. Papathanasiou38 numerically
solved the Stokes equation in two dimensions using the boundary element method in
square arrays of permeable multifilament yarns, where each yarn was composed of cir-
cles representing the cross sections of the constituent filaments. The effective permeabil-
ity of the medium was reported as a function of the weave and yarn porosities. Similar

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

12 Journal of Thermoplastic Composite Materials

Table 2. Experimental through-thickness inter-bundle and global permeability coefficient of the

studied flax fabric at different temperatures
T ( C) I slope (MPa s/mm) II slope (MPa s/mm) Inter-bundle K (m2) Global K (m2)

85 0.05169 + 2.88E3 0.03975 + 7.35E4 8.05E12 + 4.48E13 1.04E11 + 1.93E13

90 0.02792 + 4.45E4 0.01951 + 8.50E5 7.98E12 + 1.27E13 1.14E11 + 4.98E14
95 0.01694 + 6E5 0.01306 + 1.7E4 1.12E11 + 3.98E14 1.45E11 + 1.89E13
100 0.01205 + 1.16E3 0.01086 + 1.48E3 1.20E11 + 1.16E12 1.34E11 + 1.85E12

studies were conducted for square and hexagonal arrangement of filaments in yarns,
where the yarns had circular cross-section,39 and for yarns with elliptical cross-section.40
Papathanasiou41 reported a correlation for the effective permeability of two-dimensional
hexagonal arrangements of filament bundles as a function of the weave and yarn perme-
abilities, as follows:
"   #
Kinter n3=2
Kglobal ¼ Kinter 1 þ a ð8Þ
Kintra

where Kglobal is the effective permeability of fabric, Kinter is the inter-bundle permeabil-
ity, Kintra is the intra-bundle permeability and a and n are constants related to the geo-
metric structure of the fabric.
The relationship is constructed based on dimensional arguments and from consid-
eration of the behaviour at high and low Kinter/Kintra ratio asymptotes. Fitting the
numerical simulation data, Papathanasiou41 determined that a ¼ 2.3 and n ¼ 0.59 when
filaments are arranged in a squared array and a ¼ 2.67 and n ¼ 0.89 when filaments are
arranged in a hexagonal array.
In order to use Eq. (8), the experimental values of Kinter reported in Table 2 can be
used. The value of Kintra can be predicted by the Carman–Kozeny equation,42 which
correlates the intra-bundle permeability to the fibre bundle volume fraction (VfINTRA),
the characteristic fibre diameter (d) and a lumped parameter called the Kozeny constant
(C):
2
dfiber ð1  VfINTRA Þ3
KKozenyCarmanintrabundle ¼  2
ð9Þ
C VfINTRA

In order to obtain a correlation between experimental permeability data and model

prediction, the intra-bundle volume fraction was defined as reported in Eq. (10):

Afibers Afibers
vfINTRA ¼ ¼ ð10Þ
pab ATOT
Where Afibres is the actual area occupied by the fibres and a and b are the average
length of the two semi-axes of the bundle. In order to evaluate the intra-bundle volume
fraction, the total area occupied by the fibres was calculated measuring the weight (M) of
a known length (L) in a bundle:

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

Gennaro et al. 13

1.6x10–11
kinter
1.4x10–11 kglobal

transversal permeability [m2] 1.2x10–11

1.0x10–11
8.0x10–12
6.0x10–12
4.0x10–12
2.0x10–12
0.0
85 90 95 100
temperature [°C]

Figure 9. Permeability as a function of temperature in the low velocity and high velocity ranges.

Table 3. Values of the geometric parameters calculated for the four tests and the intra-bundle
volume fraction

Test temperature ( C) ATOT (mm2)  1E3 VfINTRA

85 180 + 9 0.58
90 200 + 25 0.545
95 201 + 30 0.538
100 216 + 41 0.502

M
Afibers ¼ ð11Þ
Lr

where r is the material density. From optical microscopy and knowledge of the ATOT, the
intra-bundle volume fraction was calculated according to Eq. (10). The values of the geo-
metric parameters calculated for the four tests and the intra-bundle volume fraction are
reported in Table 3.
As expected, the intra-bundle volume fraction decreases with increase in test tempera-
ture, due to the fact that decreasing the viscosity of the matrix involves an higher intra-
bundle permeation, which results in an increase in the size of impregnated bundle
(ATOT).
Table 4 shows the values of intra-bundle permeability calculated according to
Carman–Kozeny model, using a value of constant C ¼ 5043 and the parameters reported
in Table 3. Substituting the values of experimental inter-bundle permeability (Table 2)
and predicted intra-bundle permeability (Table 4) into Papathanasiou model, a global
permeability was obtained. The values of global permeability predicted by Papathana-
siou model for an hexagonal array are also reported in Table 4. The application of Eq.

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

14 Journal of Thermoplastic Composite Materials

Table 4. Model predictions for intra-bundle and global transversal permeability

Test temperature Kintra_Carman–Kozeny KPapathanasiou exgonal array KPapathanasiou a ¼ 3.2 n ¼ 0.93

( C) (m2) (m2) (m2)

85 1.36E13 9.81E12 1.06E11

90 1.53E13 9.36E12 1.01E11
95 1.56E13 1.29E11 1.39E11
100 1.70E13 1.44E11 1.54E11

(8) to the permeability obtained in our study leads to a ¼ 3.2 and n ¼ 0.93 close to the
values reported in previous work.44
As expected Carman–Kozeny model prediction for intra-bundle permeability is much
lower than inter-bundle experimental permeability reported in Table 2. The values of
global permeability predicted by Papathanasiou are of the same order of magnitude of
those reported in Table 2.

Conclusions
In this work, a new methodology for the measurement of fibrous reinforcement trans-
versal permeability has been described. The presented setup makes use of a properly
modified capillary rheometer, and allows the determination of the reinforcement per-
meability by a direct approach, which does not require any hypothesis but allows for the
measurement of the physical parameters appearing in the Darcy equation. The results
obtained in this work, derived for a Newtonian matrix, show that the pressure–velocity
curves are characterized by different regimes. Initially, a higher slope of the pressure
versus velocity curve is indicative of a lower permeability of the reinforcement. Rein-
forcement samples impregnated in the low-pressure zone showed the presence of poly-
meric fluid in the spaces between the bundles, but there was no flow of the matrix inside
each bundle. This phenomenon, which is referred to as ‘macro-impregnation’ is charac-
terized by a relatively high value of permeability that allows for impregnation even at
low values of the applied pressure.
On the other hand, at higher pressures, the slope of the pressure versus velocity curve
decreases, which indicates an increase in the overall permeability of the reinforcement.
In fact, when impregnated in the high-pressure zone, the reinforcement shows significant
flow of the matrix inside each bundle. This phenomenon, which is referred to as micro-
impregnation, is characterized by a lower value of permeability, requiring higher pres-
sures. The overall permeability, in the high-pressure zone, is given by a proper combi-
nation of macro- and micro-scale permeabilities.
In facts, the slope in the low-pressure zone was used for evaluating the macro-scale
permeability. As expected, this property is independent of the test temperature, being
dependent only on the reinforcement architecture.
The micro-scale permeability was evaluated by means of the Carmen-Kozeny equa-
tion, using the geometric properties of the reinforcement, measured by SEM and optical
microscopy.

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

Gennaro et al. 15

The experimental macro-scale and predicted micro-scale permeabilities were com-

bined by following the approach proposed by Papathanasiou, and the global permeability
was therefore calculated. The estimated value of the global permeability is in very good
agreement with the experimental value determined in the high-pressure zone.
Such results indicate the sensitivity of the developed measurement techniques. Future
works will be directed towards flow analysis of non-Newtonian matrices in dry
reinforcements.

Funding
This research received no specific grant from any funding agency in the public, com-
mercial, or not-for-profit sectors.

References
1. Phelan FR Jr. Simulation of the injection process in resin transfer molding. Polym Compos
1997; 18(4): 460.
2. Abrate S. Resin flow in fiber performs. Appl Mech Rev 2002; 55(6): 579.
3. Simacek P, Heider D, Gillespie JW Jr., and Advani SG. Post-filling flow in vacuum assisted
resin transfer molding processes: theoretical analysis. Compos A: Appl Sci Manuf 2009; 40(6-
7): 913–924.
4. Heider D, Simacek P, Dominauskas A, Deffor H, Advani S and Gillespie JW Jr. Infusion
design methodology for thick-section, low permeability preforms using inter-laminar flow
media. Compos Part A 2007; 38(2): 525–534.
5. Foley ME and Gillespie JW Jr. Modeling the effect of fiber diameter and fiber bundle count on
tow impregnation during liquid molding processes. J Compos Mater 2005; 39(12): 1045–1065.
6. Gillio EF, Gillespie JW, Jr., Fink BK and Advani SG. Investigation of the role of transverse
flow in co-injection resin transfer molding. Poly Compos 1998; 19(6): 738–746.
7. Zhang Y, Lopatnikov S, Heider D and Gillespie JM Jr. Modeling of distribution media and
vacuum bag properties on permeability variations during vacuum assisted resin transfer mold-
ing (VARTM). ASME International Mechanical Engineering Congress and Exposition,
Orlando, FL, 2005.
8. Khanafer K and Vafai K. The role of porous media in biomedical engineering as related to
magnetic resonance imaging and drug delivery. Int J Heat Mass Transfer 2006; 42(10): 939.
9. Yang N and Vafai K. Modeling of low-density lipoprotein (LDL) transport in the artery-
effects of hypertension. Int J Heat Mass Transfer 2006; 49(5–6): 850.
10. Pan C, Hilpert M and Miller CT. Pore-scale modeling of saturated permeabilities in random
sphere packings. Phys Rev 2001; 64(6): 066702.
11. Pan C. Use of pore-scale modeling to understand transport phenomena in porous media. PhD
dissertation in environmental science and engineering. University of North Carolina at Chapel
Hill, 2003.
12. Lu WM, Tung K and Hwang J. Fluid flow through basic weaves of monofilamentfilter cloth.
Textile Res J 1996; 66(5): 311.
13. Tung KL, Shiau J and Chuang C. CFD analysis on fluid flow through multifilament woven
filter cloths. Separat Sci Technol 2002; 37(4): 799.
14. Koido T, Furusawa T and Moriyama K. An approach to modeling two-phase transport in
the gas diffusion layer of a proton exchange membrane fuel cell. J Power Sources 2008;
175: 127.

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

16 Journal of Thermoplastic Composite Materials

15. Wang L and Afsharpoya B. Modeling fluid flow in fuel cells using the lattice-Boltzmann
approach. Math Comput Simul 2006; 72(2–6): 242.
16. Kaviany M. Principles of heat transfer in porous media. New York: Springer-Verlag, 1991.
17. Gennaro R, Greco A and Maffezzoli A. Numerical simulation of the microscale impregnation
in commingled thermoplastic composite yarns. Adv Polym Tech 2010; 29 (2): 122.
18. Johnson DL, Koplik J and Schwartz LM. New pore-size parameter characterizing transport in
porous media. Phys Rev Lett 1986; 57(20): 2564.
19. Comas S, Binetruy C and Krawczak P. Unidirectional compression of fibre reinforce-
ments. Part 2: continuous permeability tensor measurement. Compos Sci Technol 2007;
67: 638–645.
20. Drapier S, Monatte J, Elbouazzaoui O and Henrat P. Characterization of transient through-
thickness permeabilities of Non Crimp New Concept (NC2) multiaxial fabrics. Compos A:
Appl Sci Manuf 2005; 36: 877.
21. Drapier S, Pagot A, Vautrin A and Henrat P. Influence of the stitching density on the trans-
verse permeability of non-crimped new concept (NC2) multiaxial reinforcements: measure-
ments and predictions. Compos Sci Technol 2002; 62(15): 1979.
22. Elbouazzaoui O, Drapier S and Henrat P. An experimental assessment of the saturated trans-
verse permeability in Non Crimp New Concept (NC2) multiaxial fabrics. J Compos Mater
2005; 39(13): 1169.
23. Scholz S, Gillespie JW Jr, and Heider D. Measurement of transverse permeability using gas-
eous and liquid flow. Compos A: Appl Sci Manuf 2007; 38(9): 2034–2040.
24. Mehri D, Michaud V, Kampfer L and Manson JAE. Transverse permeability of chopped fibre
bundle bed. Compos A: Appl Sci Manuf 2007; 38: 739.
25. Ouagne P and Bréard J. Continuous transverse permeability of fibrous media. Compos A: Appl
Sci Manuf 2010; 41(1): 22.
26. Van de Ven EC, Loendersloot R and Akkerman R. Proceedings of the 8th International
Conference on Flow Processes in Composite Materials (FPCM8) Douai, France, 2006.
27. Greco A, Strafella A, La Tegola C and Maffezzoli A. Assessment of the relevance of sintering
in thermoplastic commingled yarn consolidation. Polym Compos 2011; 32(4): 657.
28. Mayer C, Wang X and Neitzel M. Macro-and micro-impregnation phenomena in continuous
manufacturing of fabric reinforced thermoplastic composites. Compos A: Appl Sci Manuf
1998; 29: 783.
29. Lekakou C and Bader MG. Mathematical modelling of macro- and micro-infiltration in resin
transfer moulding (RTM). Compos A: Appl Sci Manuf 1998; 29: 29–37.
30. Long AC, Wilks CE and Rudd CD. Experimental characterisation of the consolidation of a
commingled glass/polypropylene composite. Compos Sci Technol 2001; 61(11): 1591.
31. Zhou F, Kuentzer N, Simacek P, Advani SG and Walsh S. Analytic characterization of the
permeability of dual-scale fibrous porous media. Compos Sci Technol 2006; 66: 2795.
32. Gennaro R, Montagna F, Maffezzoli A, Fracasso S and Fracasso F. On-line consolidation of
commingled polypropylene/glass roving during filament winding. J Thermoplast Compos
Mater, in Press.
33. Chhabra RP, Comiti J and Machac I. Flow of non-Newtonian fluids in fixed and fluidised
beds: a review. Chem Eng Sci 2001; 56: 1.
34. Reddy S, Schunk PR, Roger T and Bonnecaze RT. Dynamics of low capillary number
interfaces moving through sharp features. Phys Fluids 2005; 17: 122104.
35. Schell JSU, Deleglise M, Binetruy C, Krawczak P and Ermanni P. Numerical prediction and
experimental characterization of meso-scale-voids in liquid composite moulding. Compos A:
Appl Sci Manuf 2007; 38: 2460.

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014

Gennaro et al. 17

36. Breard J, Saouab A and Bouquet G. Numerical simulation of void formation in LCM. Compos
Part A 2003; 34: 517.
37. Lundstrom TS, Gebart BR and Sandlund E. In-plane permeability measurements on fiber
reinforcements by the multi-cavity parallel. Polym Compos 1999; 20(1): 146.
38. Papathanasiou TD.On the effective permeability of square arrays of permeable fiber tows. Int
J Multiph Flow 1997; 23(1): 81.
39. Markicevic B and Papathanasiou TD. On the apparent permeability of regular arrays of non-
uniform fibers. Phys Fluids 2002; 14(9): 3347.
40. Papathanasiou TD, Gravel EM and Barwick SC. Non-isotropic structured fibrous media: the
permeability of arrays of fiber bundles of elliptical cross section. Polym Compos 2002; 23(4):
520.
41. Papathanasiou TD. Flow across structured fiber bundles: a dimensionless correlation. Int
J Multiph Flow 2001; 27(8): 1451.
42. Carman PC. Fluid flow through granular beds. Chem Eng Res Des 1937; 15: 150.
43. Bates PJ, Taylor D and Cunningham MF. Compaction and transverse permeability of glass
rovings. Appl Compos Mater 2001; 8: 163.
44. Nabovati A, Llewellin EW and Sousa ACM. Through-thickness permeability prediction of
three-dimensional multifilament woven fabrics. Compos A: Appl Sci Manuf 2010; 41: 453.

Downloaded from jtc.sagepub.com at IOWA STATE UNIV on October 1, 2014