Está en la página 1de 13

Composites Science and Technology 56 (1996) 1303-1315

0 1996 Elsevier Science Limited


Printed in Northern Ireland. All rights reserved
ELSEVIER PII: SO266-3538(96)00092-9 0266-3538/96/$15.00

A THREE-DIMENSIONAL MICROMECHANICAL ANALYSIS


OF WOVEN-FABRIC COMPOSITES: I. GEOMETRIC
ANALYSIS

Ph. Vandeurzen, J. Ivens & I. Verpoest


Department of Metallurgy and Materials Engineering, Katholieke Uniuersiteit Leuuen, de Croylaan 2, B-3001 Leuven,
Belgium

(Received 29 February 1996; accepted 27 June 1996)

Abstract architectures is unlimited. Hence, in order fully to


A three-dimensional geometric description for a large realise the potential of woven-fabric composites, it is
range of two-dimensional woven architectures is important to understand their mechanical perfor-
proposed which can serve as a basis for stiffness and mance. In this, the geometry of the yarns in the woven
strength modelling of fabric composites. We introduce fabric plays a key role.
a strict logical scheme for calculating the full geometry A decade ago, Ishikawa and Chou32 developed
of a general yarn architecture. First, a library of 108 three basic analytical models to predict the in-plane
rectangular macro-cells has been put together to build thermo-elastic behaviour of various woven-fabric
complex material unit cells, This complexity is caused composites. First, the mosaic model treats the
by the yarn interlacing style and the existence of composite as an assembly of asymmetric cross-ply
different yarn sizes in the fabric composite. Next, each laminae. The yarn waviness is neglected. Second, the
macro-cell is divided in a number of micro-cells. The crimp model takes into account the continuity and
geometric model, based on a lenticular yarn cross- undulation, i.e. the yarn waviness or crimp, of the
section, predicts the weave cover factor, the global and yarns in the loading direction. The undulation of the
local fibre volume fractions and the yarn orientations. yarns running perpendicular to the loading direction
Moreover, a logical and simple meshing for each is, however, neglected. Finally, for satin composites
woven structure is created. A Microsoft Excel@ the bridging model was developed to simulate the load
application (TEXCOMP) has been developed to transfer amongst interlaced domains. These models
transform the model into a useful design tool, 0 1996 of Ishikawa and Chou are one-dimensional (1D)
Elsevier Science Limited models because they only consider the undulation of
the yarns in the loading direction. Notice the total
Keywords: woven fabric, geometric model, macro-cell, absence of any geometric analysis. That is, the actual
micro-cell, design yarn cross-sectional shape, the presence of a gap
between adjacent yarns and the influence of crimp on
fabric thickness are not considered. Therefore, no
1 INTRODUCTION
predictions are made for the orientation of each yarn,
A renaissance of interest in textiles and textile for the fibre volume fraction or for the fabric cover
composites has been observed in recent years. Their factor.
ease of handling and low fabrication cost have made The two-dimensional (2D) model of Naik and
woven fabrics attractive for structural applications. Shembekar3.4 considers the undulation of both warp
Moreover, woven-fabric composites can offer the and weft yarns. On the basis of an adequate geometric
potential of balanced in-plane thermo-elastic pro- description, on the classical lamination theory and on
perties and improved impact resistance. a mixed parallel-series arrangement of infinitesimal
Currently, most of the woven fabrics used are composite pieces, the model predicts the in-plane
simple constructions, such as plain, basket and satin elastic properties. This rather complicated method
weaves. However, the fabric architecture can be involves substantial computation. Therefore, Naik and
altered to meet the specific mechanical properties Ganesh developed closed-form expressions for the
required by the composite application. This is in-plane thermo-elastic properties of plain-weave
achieved by changing the weave pattern, fibre type, fabric laminae. Shape functions were used to define
yarn type, yarn count, etc. The number of possible the yarn cross-section and undulation.
1303
1304 Ph. Vandeurzen et al.

The three-dimensional (3D) model of Hahn and


Pandy6 for plain-fabric composites is simple in concept
and mathematical implementation. The yarn undula-
tions are sinusoidal and described with shape
functions. However, in the subsequent elastic analysis
the most questionable assumption is the uniform
strain throughout the plain-weave composite unit cell.
The present work is concerned with the analytical
three-dimensional geometric and elastic modelling of
GO
a large range of two-dimensional woven fabric
laminae. The objective of Part I of this two-part paper
is to present a full geometric analysis. It is new
because of its clear geometric concept (a library of
building blocks for textile composites) and its ability
to describe also non-traditional fabrics with various
yarn sizes or particular yarn crimps. Also, the
presence of special yarns in the fabric such as optical
fibres or shape-memory alloy fibres can be taken into
account. In Part II, we will predict the full set of (W
engineering constants based on the previous mathe-
matical description with macro- and micro-cells. The
model is superior to other analytical analyses because
it predicts correct shear moduli for the fabric
composite. Still, the analysis is sufficiently general to
describe any 2D woven fabric, is easy to perform and
is concise in concept.

2 WOVEN-FABRIC STRUCTURE Cc)


Fig. 1. Basic weave constructions: (a) plain, (b) twill and (c)
In general, an orthogonal 2D woven fabric is made by 5HS satin weave. The grey shaded zone represents the fabric
weaving together yarns. Each yarn is a bundle of unit cell.
fibres and its size is determined by the number of
fibres in the yarn, the diameter of one fibre and the
packing density of fibres. The fabric is produced on a 3 DEFINITION OF THE FABRIC UNIT CELL
loom that interlaces yarns at right angles to one
another. The lengthwise yarns are called warps, while Since the mechanical properties of woven-fabric
the yarns that are shuttled across the loom are called composites have a very strong dependence upon the
fills or wefts. The interlacing of the yarns causes yarn reinforcing yarn geometry, it is essential to create a
undulation or yarn crimp. The weave type is geometric concept or scheme for describing the fibre
determined by the method of interlacing both sets of architecture. The idea is to include just enough
yarns. Figure 1 shows three basic constructions. The details in the idealised geometry to predict the
plain weave is the most commonly used basic most important aspects of the composite behaviour.
reinforcement for woven composites. Shifting of the This type of geometric concept has not yet been
yarns, or (in other words) displacement of the yarns, is introduced in the existing analytical models described
difficult. The twill weave has a looser interlacing and in Section 1.
the weave is characterised by a diagonal line. Each In this research work, the woven fabric is treated as
warp yarn floats over two or more consecutive weft an assembly of unit cells, i.e. grey areas in Fig. 1. By
yarns. The satin weave has a good drapability, a definition, the unit cell is the smallest repeating
smooth surface and minimum thickness. Here, one pattern in the structure. Moreover, the model deals
warp yarn is floating over at least three weft yarns and with a perfect, regular, one-layer fabric composite.
under a single one. Even in rather simple woven Hence, the presence of voids, the misorientation of
fabrics, there are important geometric differences yarns and the nesting of fabric layers are neglected in
between the warp and the weft direction. These our unit cell concept, just as they are neglected in
differences are the result of numerous constructional most other fabric composite models.
and process parameters like weaving density, warp The checkerboard pattern, a well-known represen-
tension, weft tension, beating motion, etc. tation in the textile industry, is extended now to
30 analysis of fabric composites: I 1305

woven constructions such as optical glass fibres or


shape-memory alloy fibres with particular sizes.

4 GEOMETRIC PARAMETERS

Half a century ago, Peirce8,9 developed the essential


Carbon yarn geometric relationships between parameters such as
-. Flax yarn yarn crimp, yarn length and yarn count, to solve
Ei
practical problems for designing cloths. The analysis
f* P f f treats the yarns as flexible, inextensible cylinders. As
such, this hypothesis is not completely valid for fabrics
for composites because the yarns never have a circular
cross-section, but rather a flattened elliptical or

1
W
w / lenticular one. The model does not take into account
W
the setting of the yarn crimp in the fabric, i.e. the
residual crimp of the yarn after it has been released
W*
from the fabric. Bending stiffness and torsional
n warp over weft stiffness of the yarn are also neglected. From a purely
textile point of view the model is acceptable; for the
c-7 weft over warp plain-woven fabric, formulae are presented that can
Fig. 2. Extended checkerboard pattern of a hybrid be useful when studying shrinkage effects of the
twill-weave unit cell. The w- and f-type yarns are natural flax textile, cloth extension, limits of setting, etc.
fibres, and the w*- and f*-type yarns are carbon fibres. Our analysis provides the basic description of the
microgeometry of a large range of 2D woven fabrics.
Not only standard weaves such as plain, twill or satin
describe the most complex material unit cells. Figure 2 weaves, but also more complex weaves with a special
shows an example of a checkerboard pattern for such yarn interlacing style and with different yarn types are
a complex fabric, namely a hybrid carbon/flax twill considered. The most detailed analysis would consider
weave. The hybrid fabrics of carbon fibre and natural the path of each single fibre in the unit cell. The
flax fibre are actually under study in our research greatest practical problem, however, is caused by the
group. The rows of the board represent the warp fact that the complete set of input data necessary for
yarns, while the columns are the filling or weft yarns. such a detailed geometric description is very large and
At an interlacing point, the square is coloured black if difficult to quantify. Therefore, the geometric analysis
the warp yarn runs over the weft yarn. The main is carried out on the yarn level. We assume that all
complexity arises from the fact that the fabrics fibres in the yarn run in the same direction as the
considered here can contain two different warp yarn yarn. For twisted yarns, the existence of the twist
types, labelled w and w*. The fibre mechanical angle could be taken into account by altering the
properties, the size, the shape and the inter-yarn elastic properties of the impregnated yarn. The
distances can vary. Moreover, two filling yarn types, f intra-yarn fibre volume fraction or packing factor,
and f*, can be present in the fabric unit cell. Although defined as fibre-to-yarn area ratio, is assumed to be a
they are equally sized and shaped, the fibre constant for the woven-fabric composite. From a
mechanical properties and the inter-yarn distances can textile point of view, the air-filled space trapped by
differ. Because of the possible existence of different the fibres affects the insulating and warmth properties
types of yarns in one fabric structure, an extra column of the fabric. It is clear that interlacing of the yarns
and row are added to the classic checkerboard. If a and processing of the composite leads to thread or
square in this extra row or column is marked, the yarn flattening. Therefore, the assumption of a
corresponding yarn is an asterisk-type yarn. First, the circular yarn cross-section is unrealistic. On the basis
extension is necessary to describe hybrid weave styles. of microscopic observations, a lenticular shape was
Next, the need for this refinement is based on selected to describe the cross-sectional shape of the
microstructural observations of woven-fabric compos- yarn (Fig. 3).
ites. Even when yarns are woven separately for the Figure 4 depicts the geometric parameters on a
construction of a fabric, they can unite to compose a cross-section of a woven composite. Three groups of
big yarn or super yarn in the composite. The geometric parameters are identified to describe a
existence of those super yarns in the skin of 3D woven general two-dimensional weave geometry (Table 1).
core sandwich composites was predicted theoretically The subscripts f, f*, w and w* distinguish quantities
and confirmed experimentally.? Of course, the associated with filling yarns, filling* yarns, warp yarns
extension is also needed when using special fibres in and warp* yarns, respectively. The first group- the
1306 Ph. Vandeurzen et al.

Table 1. Classification of the geometric parameters

Know-group Nr Number of fibres in the yarn


d Diameter of the fibre
P Yarn spacing
Measure-group f Aspect ratio of the yarns
4 Crimp parameter for the filling yarn
D Thickness of the composite
K Packing factor of fibres in the yarn
Calculate-group h,, h,,,, Crimp parameter for the warp yarns
t Thickness of the yarn cross-section
Width of the yarn cross-section
Fig. 3. Lenticular yarn cross-section.
; Orientation of the yarn
v, Fibre volume fraction
CF Cover factor
know group-contains those parameters that are
supplied by the weaving company: the number of
tibres Nf and the fibre diameter d for each different parameter. Notice that those two geometric para-
yarn; the warp and weft yarn counts from which we meters are always positive numbers because they are
derive the yarn spacings or pick lengths p. For linear measures. For the packing factor K of fibres in
example, pw,* is the distance between two neighbour- the yarn, one can make a very good estimation. This
ing yarns, a w-type and a w*-type warp yarn. All the estimation is based on real yarn data such as the
parameters we have to measure on a real fabric number of fibres in the yarn and the diameter of a
composite are put together in the second group-the fibre, but also on microscopic determination of the
measure group. This fabric information can be area of the yarn cross-section. Usually, the value of
obtained by microscopic observation of warp and weft the packing factor is about 75%. One can easily
sections of the fabric composite. The aspect ratio f of calculate that an absolute maximum of 91% is reached
the yarn, defined as the width over the thickness of when the fibres are hexagonally close packed. Finally,
the cross-section, is the most important one. The the third group-the calculate group-contains all
crimp parameter hf describes the undulation of the values that are calculated from the previous
filling yarns. Of course, the undulation of the yarns in parameters, using formulas based on simple geometric
the warp direction is related to this parameter, considerations. They will be introduced in Sections 5
because the increase of undulation in one direction of and 6. This group contains the crimp parameters h for
the fabric reduces the undulation in the other
direction. Figure 5 demonstrates the effect of hf on the
fabric geometry of a plain-woven structure: if hf
equals zero, the path of the filling yarns is totally
straight. If, on the other hand, h, equals zero, the
undulation of the filling yarns is at its maximum. The
thickness D for a one-layer fabric composite can be
measured easily. However, when conducting paramet-
ric studies we will usually set D equal to the fabric
thickness, Dfabric. The fabric thickness has a strong
dependence on the fabric geometry and is given by:
Dfabric= max(t, + h,, t,.,*+ h,*, tf + hf, tp + hp) (1)
where t is the thickness of the yarn and h is the crimp

D/

DI

Fig. 4. Cross-section of a fabric composite along the warp Fig. 5. Influence of the crimp parameter h on the
direction. plain-weave fabric geometry: (a) h, = 0 and (b) h, = 0.
30 analysis of fabric composites: I

the different warp yarns, and the thickness t and width


w of each yarn cross-section. The most important
parameters are certainly the yarn orientations p, the
fibre volume fraction V,, and the cover factor CF.
Cover factor is herein defined as the projected yarn
area to unit cell area ratio. Yarn orientation and fibre
volume fraction are crucial for understanding and
modelling the mechanical properties of the fabric
composite. The cover factor, on the other hand, is of

l!29
primary interest in studies of the porosity and the tw
permeability of fabrics to air, water and polymers.
tf
i
(cwcm
] FIWW
5 PARTITION OF THE UNIT CELL: MACRO- ffl2
AND MICRO-PARTITION

tw
There are two major motivations in creating a strictly
logical partitioning scheme for the unit cell. First, it
should be easy then to calculate the geometric
parameters which fully describe the yarn architecture, [p&J tf
Pff
only based on the presented know and measure
group. The bookkeeping of geometric data becomes a

;r=-
simple task. Second, to calculate the elastic properties A
of the woven-fabric composite, one needs a logical
and simple geometric meshing of the unit cell. tw
Basically, our meshing has the same purpose as the
meshing of elements over the geometry in the
well-known finite element analysis and, in this way, it
[FG] tf Pff
w* *
could be considered as an intelligent mesh generator
for woven-fabric composites. Fig. 6. Library of macro-cells: four exemplars.
The macro-partition of the unit cell consists of
discretising the unit cell into a number of rectangular construction pattern provided by the weaving
macro-cells. Each square of the checkerboard pattern company is sufficient to determine automatically the
is split up into four macro-cells. The macro-partition number of each type of macro-cell present in the unit
includes a split into two layers in the thickness cell. For this, the extended checkerboard pattern is
direction of the composite. That is, at each cross-over transformed into an algebraic matrix, where each
point of a warp and a weft yarn, one needs two matrix element is completely determined by the
macro-cells in one layer to define the path for the geometric positions and yarn types of the two
warp yarn and two in the other layer for the weft yarn.
Depending on the orientation of the yarn, we can
distinguish A-, B- and C-type cells for the warp yarns

Bd3
and D-, E- anf F-type cells for the weft yarns.
In order to describe a general weave geometry, a
library of 108 macro-cells has been put together. Even
the most complex 2D woven structures can be
composed with this library of rectangular macro-cells
or building blocks. Figure 6 shows examples of UPPER
macro-cells. Again the subscripts w, w*, f and f* are
LAYER
used, now to define a name for each cell; e.g. Bw$
describes the path of a w*-type warp yarn running
from an f-type yarn to an f-type yarn. Moreover,
special macro-cells containing pile yarns of 3D woven
core/sandwich composites7,2 were developed and LOWER
added to the library to extend the model. As can be LAYER
seen in Fig. 7, we need 16 macro-cells to compose the
unit cell of a simple plain-woven fabric composite.
It should be pointed out that only the weave Fig. 7. Macro-partition of the plain-weave fabric unit cell.
1308 Ph. Vandeurzen et al.

interlacing yarns. The number of rows and columns of


the matrix equals the number of warp and weft yarns
in the unit cell, respectively. The matrix element at
any position is 1 if the warp yarn is running over the
weft yarn, i.e. the checkerboard field is black, and -1
in the other case. Then, any row corresponding to an
asterisk-type warp yarn is multiplied by 2, while any
column corresponding to an asterisk-type weft yarn is
multiplied by 3. On the basis of the value of a matrix
element and the values of the four neighbouring
elements, the code defines the unique set of four types
of macro-cell at the corresponding fabric position. No
extra fabric information, nor any geometric assump-
tions, nor operator interventions are needed. There-
fore, the macro-partition is simple in concept and easy
to apply.
Next, each macro-cell itself is automatically
partitioned into micro-cells. This process is called the
micro-partition. Two different methods of micro-
partition will be presented: an elementary 1D
partition and a more advanced 2D partition, The (b)
micro-partition results in a fully three-dimensional
division of the unit cell, so that the variation of
properties within the unit cell can be properly
analysed. This is essential for understanding the q straight yarn path
mechanical performance of woven-fabric composites. curved yarn path
Now, on the basis of the developed geometric
Fig. 8. (a) 1D micro-partition of a Bw&ype macro-cell; (b)
concept, we are able to predict with analytical models
2D micro-partition. Only half the Bw&ype macro-ceil is
the mechanical stress variations within the composite shown.
unit cell. This shall be illustrated in Part II of this
paper.
There is no obvious geometric nor mechanical
5.1 1D micro-partition reason to use, for the description of the yarn
In this case, the assemblage of four micro-cells creates curvature, micro-cells of different sizes. Therefore, the
a macro-cell. The size of each micro-cell depends on curved part of the warp yarn is split up in three
the weave geometry (Fig. 8). Part of the yarn in each equally sized micro-cells. They have all the same
macro-cell is assumed to be straight between adjacent width, %%d celi. In this way, the unit cell is partitioned

crossing yarns. Although woven fabrics show a into rectangular micro-cells:


smoothly curved yarn between adjacent crossing
yarns, microscopical observations of woven-fabric
composites justifies the assumption of straightened
yarns. The width of the micro-cell containing the Wcurved cell =

straight part of the warp yarn can be calculated as


follows:
5.2 2D micro-partition
Straight cell Also, a two-dimensional approach was used to
partition the macro-cells. Here, cuts were made not
Here, pff is the yarn spacing for f-type yarns, Pmax is only perpendicular to the yarn path but also parallel
the maximum orientation of the w-type yarn, Rf is the to the yarn path. The 2D micro-partitioning procedure
radius of curvature of the lenticular f-type yarn over is an extension of the 1D case. First, the 1D
which the considered w-type yarn is running (Fig. 3) micro-partition should be carried out for each
and t, is the thickness of the w-type yarn. The radius macro-cell in the composite unit cell to describe the
of curvature R can, of course, be written as a function undulation of the considered yarn (A). Next, to make
of the aspect ratio f and the thickness t of the yarn a distinction between the yarn and matrix areas, two
cross-section: extra cuts are made parallel to the yarn at the two
ends of the lenticular shape of the considered yarn
(B). It should be stressed that Fig. 8 only shows half
30 analysis of fabric composites: I 1309

the macro-cell, namely in the top layer of the unit cell. The assumption of f 2 1 is implied in the
Finally, every cut made in every macro-cell is expressions above. The parameters h, w and t are
extended to the other layer of macro-cells of the linear measures and therefore positive numbers. Next,
woven fabric composite (C, and C,). As explained the orientation of the yarn in each micro-cell has to be
before, every single-ply woven-fabric composite has calculated. The orientation Pmax in eqns (2) and (4)
two layers of macro-cells because of the division in the can be expressed as a function of yarn spacing p,
thickness direction. This automated partitioning radius R, thickness t and crimp parameter h. The
procedure originates in a regular grid of micro-cells to following geometric relationships were calculated for
describe the unit cell. Fifty micro-cells compose one an Awfr-macro-cell (Fig. 4). Only these relationships
macro-cell. Not all of those micro-cells contain fibres, are given as an example:
i.e. some of the micro-cells are pure matrix material
cells. Each micro-cell is characterised by its local fibre
volume fraction, yarn orientation and fractional
arccos
(2%+ tw> cos[ _l+h,)]
arctan( 2Rf

Pmax =
volume relative to the unit cell. 2R,-t,+h,

S
6 MATHEMATICAL ANALYSIS + arctan
2R,-t,+h,
A strategy for calculating the geometric output is or
presented in Fig. 9. First, the number of each type of
macro-cells is automatically determined from the P,..,,=arcsin(y] if 2R,-t,+h,=O
weave construction pattern, provided by the weaving
company, as explained in Section 5. The crimp In the expressions above the constraint eqn (8) is
parameters, and thus the undulation, of the warp implied. This important constraint symbolises that it
yarns are calculated as follows: must be geometrically possible for the warp yarn to
interlace with the filling yarns. That is, if the
h,=t,+t,-hf
(5) constraint is false then the fabric architecture cannot
h,* = t,.,* + tf - hf be created:
The width w and thickness t of each yarn cross-section
are computed as a function of the number of filaments
N,, the filament or fibre diameter d, the aspect ratio f
and the packing factor K: The orientation of the warp yarn in the four different
micro-cells of the Aw, macro-cell, considering the 1D
N,nd2
micro-partition, is then calculated as a function of the
j/*K[arcsin(&J f (fT)z - (f- 1). ([)I maximum orientation Pmax. In case of a 2D
micro-partition, the presented equations are still valid.
w=tXf (6) Pstraight cell = Pmax (9)

II Nf, d, p and weave style KNOW


GROUP
with k = 5, 3 and 1 (10)
MEASURE From the calculated yarn orientation for each
GROUP micro-cell and the composition of the unit cell, one
. - can easily recalculate and draw an orientation
distribution function for the warp and weft yarns.
Such a graph plots the probability to hold a certain
yarn orientation in a randomly chosen infinitesimal
piece along warp or weft direction. Finally, the global
vf CALCULATE fibre volume fraction V, and the cover factor CF for
CF GROUP the unit cell can be expressed as a function of the
E i fabric parameters:
108
K ~ niA;li
V,= i=l
Fig. 9. Geometric calculations: strategic scheme. Knit cell
(11)
1310 Ph. Vandeurzen et al.

In eqn (11) K is the packing factor, ni is the number A custom Microsoft Excel@ application, called
of macro-cells of type i in the unit cell (i = l.lOS), Ai is TEXCOMP, has been developed to perform all
the area of the yarn cross-section, Ziis the length of the calculations.3 Dialogue boxes, which work and look
yarn in macro-cell i, and Vunit_,, is the volume of the like the built-in Excel@ dialogue boxes, were created
unit cell. to standardise the user interface. Different macro
The cover factor is important for permeability functions are included in the software to perform
properties of the fabric. Since this factor is defined as specialised tasks like drawing the orientation distribu-
the projected yarn area to unit cell area ratio, its value tion function for the warp and weft yarns or
is determined by the width of the yarn sections w and calculating the macro-cell composition for a given unit
the yarn spacings p: cell. One can choose standard weave styles such as
CF= plain, twill and satin weaves, but also self-designed
fabrics can be treated by the model. Because of the
use of a powerful spreadsheet program, it is easy to
add missing aspects or to exchange data with other
KnitcelllD applications. In this way the geometric model is
transformed into a useful design tool.
(12)
Certainly, the most important output of the
calculation procedure are the fibre volume fraction, 7 PARAMETRIC STUDY
the orientation of the yarn and the fractional volume
of each micro-cell. These data are the basis for a The TEXCOMP custom application was used to
further modelling of mechanical properties, discussed analyse different 2D woven fabric composites. A
in Part II of this paper. Moreover, the geometric parametric study has been conducted to evaluate the
model developed is most useful in determining some effect of various fabric parameters, i.e. the yarn aspect
textile properties such as cover factor and fabric ratio f and the crimp ratio H, on the global fabric
thickness, but also in determining the allowable geometry. The crimp ratio H of a fabric is herein
microstructural states of fabrics. Three possible facts defined as the ratio of the difference between the
that limit the fabric architecture can be distinguished fabric crimp parameters hf and h, to the maximum
(Fig. 10). First, the interlacing of the yarns must be crimp value h,,,. The maximum crimp value
possible. The conditions at which yarns jam each other corresponds to a maximum yarn undulation in either
are essential in identifying the processing windows of the warp or the weft direction:
fabric preforms. Second, interaction of the yarns
h--h,
(13)
H=---.-.-
should be avoided. This interaction error is due to a
h max
mistake in the input data or it indicates the formation
of a combined yarn or super yarn in the real fabric The presented example considers a plain woven-
composite. Finally, also the undulation of the yarns fabric composite. The yarn characteristics for this
has limits: the crimp parameters h, and h,, should be parametric study are given in Table 2. Notice that the
positive numbers. warp and weft yarns have the same size. A yarn fibre
packing density K or intra-yarn fibre volume fraction
of 041 was used in all calculations. For a given crimp
ratio H, the filling and the warp crimp parameters can
be calculated from:

L +6 H x hmx
h,=
60 2 2
(14)
t, + tf I H x k,ax
hf =
2 2

in which the thickness t, in turn, is derived from eqn


(6). From the definition of the crimp ratio H, one can
easily see that both yarn systems have the same
undulation if the crimp ratio H = 0. When the crimp
ratio H is set equal to 1, we describe a fabric where
the warp yarns show no undulation while the filling
yarns show a maximum undulation. If H = -1, the
filling yarns are totally straight. Finally, the composite
Fig. 10. Woven construction limits: (a) impossible interlac-
ing of yarns, (b) interacting of yarns and (c) crimp error thickness is set equal to the fabric thickness.
h,<O or h,<O. The format of the study (Table 3) is to start with a
30 analysis of fabric composites: I 1311

Table 2. Yam characteristics used in the TEXCOMP parametric study

Plain-weave composite E fibre


= 73 GPa, vfibre = O-2 and E,,,, = 3-13 GPa, v,,,,, = 0.34

Thickness, D fI = {Dtabric = m=G + k, tf + 41

Yarn properties d (mm) K

Warp 415 o-01 O-81


Filling 415 0.01 ON

Table 3. Limiting yam spacing p in the TEXCOMP parametric study

Limiting yarn spacing, p (mm) H=-1 H = -0.75 H = -0.5 H= -0.25 N=O

f = 1 circular yarn - - - - 0.4


f=5 - - - - 0.6
f = 10 1.15 3 0.95 0.85 0.8
f = 20 flat yarn 1.1

constant crimp ratio (H = 0). Then, for different yarn for different crimp ratios, the yarn spacing is lowered.
aspect ratios, the yarn spacing is lowered until the The range of values for the aspect ratio f and the
geometric limit of the fibre architecture is reached. crimp ratio H used in this study, together with the
This geometric limit is controlled by the condition of corresponding limiting yarn spacing p, are shown in
impossible interlacing of both yarn systems (Fig. 10). Table 3.
Next, the aspect ratio is kept constant (f = 10) and, Figures 11 and 12 show the fibre volume fraction,

100
s 90
; 80
3 70 4-f=l
2 60 -a- f=5
50 +f=lO
1 40
+ f=20
7 30
8 20
2 10
0
0 0.5 I 1.5 2 2.5 3 3.5 4

0a Yarn spacing p (mm)

0.5
h
J 0.4
-c-f=1
z
8 0.3 + f=5
3 -t- f=lO
8 0.2
* f=20
II
[ 0.1
c;l
0
0 0.5 1 1.5 2 2.5 3 3.5 4
@)
Yarn spacing p (mm>

Fig. 11. Process windows (H = 0): (a) fibre volume fraction and (b) lamina thickness as a function of yarn spacing p.
1312 Ph. Vandeurzen et al.

1
0.9
0.8
8 0.7 +-f=l
-u- f=5
-A- f=lO
-A- f=20
u 0.3
0.2
0.1
0
0 a 1 1.5 2 2.5 3 3.5 4

Yarn spacing p (mm)

+f=l
-n- f=5
+f=lO
+ f=20

0 0.5 1 1.5 2 2.5 3 3.5 4


(b) Yarn spacing p (mm)
Fig. 12. Process windows (H = 0): (a) cover factor and (b) maximum yarn orientation as a function of yarn spacing p.

the lamina thickness, the cover factor and the equal. This is due to the strong influence of the crimp
maximum yarn orientation as a function of yarn ratio on lamina thickness (Fig. 13(b)). The lamina
spacing p for different yarn aspect ratios f. The limits thickness is minimum if both yarn systems have the
of these process windows are caused by the situation same undulation. The cover factor is not influenced by
of impossible yarn interlacing together with the the crimp ratio H, because this geometric parameter
situation of a yarn aspect ratio lower than 1. The was defined as the projected yarn area to unit cell area
maximum fibre volume fraction reached, even for very ratio. Therefore, its value is only a function of yarn
flat yarns, is about O-54. This is a rather low value, spacings p and yarn widths w. Figure 14 shows the
compared with the high intra-yarn fibre volume maximum yarn orientation of warp and weft yarns for
fraction. The large differences in fibre volume fraction different crimp ratios H as a function of yarn spacing
for the various aspect ratios can be attributed to the p. By definition, the filling yarns show no undulation
difference in composite thickness (Fig. 11). When the and the warp yarns a maximum undulation if H equals
yarns have a circular yarn cross-section, the maximum -1. By lowering the yarn spacing p, this maximum
cover factor is 0.81. However, when flattening the undulation will increase significantly until 12 is
yarns, it is easy to reach a cover factor of 1. The reached. If H = 0, both yarn systems have the same
maximum yarn orientation Pmax can reach as much as maximum yarn orientation. For the case of yarn
50 for a dense weave with circular yarn cross-sections. spacings close to the geometric limit of the woven
This value drops to 18 if the yarn aspect ratio is five. construction, small variations in yarn spacing result in
Figure 13(a) shows the influence of yarn spacing p and large variations in yarn orientation. Generally, the
crimp ratio H on the fibre volume fraction. The predicted geometric and elastic constants are very
presented process window is a narrow strip. Notice sensitive to the yarn spacing in the vicinity of the
that the case of H = 1 shows the same dependence as geometric limit of the fabric.14 Finally, Fig. 15 shows
H = - 1 for properties not related to warp or weft the allowable microstructural states of the fabric as a
direction, i.e. fibre volume fraction, lamina thickness function of the yarn aspect ratio f and the crimp ratio
and cover factor. The fibre volume fraction is H. That is, for different conditions of the yarn aspect
maximum if the undulation of both yarn systems is ratio f and the crimp ratio H, the yarn spacings in
30 analysis of fabric composites: I 1313

100
g 90 1Fibre nackina in yarn
80
B -m- IHI=l
T, 70
-u- IHW.75
8 60
50 + IHW.5
g 40 + IHI=O.25
2 30 + IHI=O
c 20
2 10
0
0 0.5 1 1.5 2 2.5 3 3.5 4
0 a
Yarn spacing (mm)

0.25

1- 0.2 + IHI=l
n -n- IHI=0.75
B 0.15
+ IHl=O.5
4 + IHW.25
3 0.1
+ lHI=O
3
# 0.05
1 IHI=l 1 s
3
0
(b) 0 0.5 1 1.5 2 2.5 3 3.5 4
Yarn spacing p (mm)
Fig. l3. Process windows (f = 10): (a) fibre volume fraction and (b) lamina thickness as a function of yarn spacing p.

warp and weft direction are decreased until the different types of warp and weft yarns. Next, each
geometric limit is reached. This plot shows a very macro-cell itself is automatically partitioned in four or
different behaviour for low and high yarn aspect 50 micro-cells: the 1D or 2D micro-partition,
ratios, with a transition zone between fequals 1 and 5. respectively. The most important output of the model
This kind of parametric study shows the need for a is the fibre volume fraction, the orientation of the yarn
refined modelling of the yarn architecture and the and the fractional volume of each micro-cell.
yarn crimp to study the effects of fabric parameters on Moreover, the geometric model as such is most useful
overall mechanical properties, through the fibre in determining some textile properties as cover factor
volume fraction and the yarn orientation, and also on and fabric thickness, but also in determining the
textile properties, through the cover factor and the allowable microstructural states of fabrics. The above
fabric thickness. The geometric limit of the woven geometric concept and mathematical analysis was
fabric is strongly dependent on the microstructural implemented in a user-friendly custom application for
parameters. Microsoft Excel*, called TEXCOMP. The software
can treat self-designed fabric weave styles. It is easy to
change existing formulations, to include additional
8 CONCLUSIONS aspects, or to exchange data.
Certainly, geometric descriptions are never true
In this paper, a model describing the geometry of a representations of reality. They are based on
woven composite with arbitrary weaving architecture simplifying assumptions. A closer look at a fabric or
was presented. After carrying out an extensive fabric composite will show that our assumptions
geometric analysis, a library of macro-cells has been concerning the fabric geometry are not completely
put together to describe complex material unit cells. true. For example, the yarns do not follow totally
This partitioning of the unit cell in macro-cells is straight yarn paths in the warp and weft directions, the
called the macro-partition. The main complexity arises yarns do not have a perfect lenticular shape, and voids
from the fact that the weave considered can contain are always present in the composite. Also, pure
1314 Ph. Vandeurzen et al.

a-Hz-1

-o- H=-0.75
+ H=-0.5
- H=-0.25
-O-H=0
+H=l

0
0 3.5 4
(a > 0.5 1 1.5 2 2.5 3

Yarn spacing p (mm)

14

+-Hz-l
-o- H=-0.75
-t H=-0.5
+ H=-0.25
- H=O
-o-H=1

0
0 0.5 1 1.5 2 2.5 3 3.5 4
09
Yarn spacing p (mm)
Fig. 14. Process windows (f = 10): ( a ) maximum warp yarn orientation and (b) maximum weft yarn orientation as a function of
yarn spacing p.

geometry is not sufficient for the design of of the composite on the other hand. We shall
woven-fabric composites. A link should be established demonstrate that the geometric assumptions made are
between the weave geometry and material properties very useful for the translation of geometry into elastic
on one hand, and the resulting mechanical properties properties of the fabric composite in the second part

1.6

B 0.8
ii 0.6

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

pm (mm)
Fig. 15. Process window: allowable microstructural fabric states as a function of yarn aspect ratio f and crimp ratio H.
30 analysis of fabric composites: I 1315

of this paper, where an analytical model with the 5. Naik, N. K. & Ganesh, V. K., An analytical method for
plain weave fabric composites. Composites, 264 (1995)
potential to predict correct shear moduli for
281-289.
woven-fabric composites will be established. 6. Hahn, H. T. & Pandy, R., A micromechanics model for
Moreover, we shall indicate how the presented thermoeiastic properties of plain weave fabric compos-
geometric concept can serve as a solid basis for the ites. J. Engng Mater. Technol., 116 (1994) 517-523.
stress analysis problem. 7. Vandeurzen, Ph., Geometrische en elastische modellering
van drie-dimensionale weefsel-sandwichcomposieten.
Master thesis, Katholieke Universiteit Leuven, Belguim,
ACKNOWLEDGEMENTS 1993.
8. Peirce, F. T., Geometry of cloth structure. J. Text. Inst.,
This paper presents research results of the Belgian
28 (1937) T45-96.
programme on Interuniversity Poles of Attraction, 9. Peirce, F. T., Geometrical principles applicable to the
funded by the Belgian state, Prime Ministers Office, design of functional fabrics. Text. Res. J., 173 (1947)
Science Policy Programming. The scientific respon- 123-147.
sibility is assumed by its authors. Ph. V. and J. I. are 10. Olofsson, B., A general model of a fabric as a geometric
mechanical structure. J. Text. Inst., 5511 (1964) T541.
financed through grants of the Flemish Institute for
11. Hoffmann, R. M., Some theoretical aspects of yarn and
the Promotion of the Scientific-Technological Re- fabric density. Text. Res. J., (1952) 170-178.
search in Industry (IWT). 12. Ivens, J., Vandeurzen, Ph., Van Vuure, A. W.,
Verpoest, I., Ko, F. K. & Meerding, K., Modeling of the
skin properties of 3D fabric sandwich composites In
REFERENCES
Proc. 9th Int. Conf. on Composite Materials (lCCM/9),
Ishikawa, T. & Chou, T.-W., Stiffness and strength Vol. IV, ed. A. Miravete. Madrid, 1993, pp. 540-547.
behaviour of woven fabric composites. .I. Mater. Sci., 17 13. Vandeurzen, Ph., Ivens, J. & Verpoest, I., TEXCOMP:
(1982) 3211-3220. a 3D analysis tool for 2D woven fabric composites.
Ishikawa, T., Matsushima, M. & Hayashi, Y., SAMPE J. (in press).
Experimental confirmation of the theory of elastic 14. Vandeurzen, Ph., Ivens, J. & Verpoest, I., Structure-
moduli of fabric composites. J. Camp. Mater., 19 (1985) performance analysis of two-dimensional woven fabric
443-458. composites. In Proc. 10th Int. Conf on Composite
Naik, N. K. & Shembekar, P. S., Elastic behavior of Materials (ZCCM-lo), Vol. IV, ed. K. Street & A.
woven fabric composites: I. Lamina analysis. .J. Camp. Poursartip. Whistler, Canada, 1995, pp. 261-270.
Mater., 2615 (1992) 2196-2225. 15. Vandeurzen, Ph., Ivens, J. & Verpoest, I., A
Shembekar, P. S. & Naik, N. K., Elastic behavior of three-dimensional micromechanical analysis of woven-
woven fabric composites: II. Laminate analysis. J. fabric composites: II. Elastic analysis. Comp. Sci.
Comp. Mater., 2615 (1992) 2226-2246. Technol., 56 (1996) 1317-1327.

También podría gustarte