ISSN 1746-7659, England, UK

Journal of Information and Computing Science

Vol. 5, No. 1, 2010, pp. 071-080
Modelling of Textile Structure for Advanced Applications
Xiaogang Chen
+

School of Materials The University of Manchester Manchester M13 9PL, UK
(Received November 3, 2009, accepted February 2, 2010)
Abstract. In general, textiles are perceived as soft materials and structures and textile fibres can be
organised into fabrics of different complexities to form new materials and structures. Textiles are widely used
in many applications, ranging from clothing to advanced technical applications such as textile composites for
the aerospace and other industries. Accurate understanding the relationships between the construction and
technical behaviour of textiles is the key in engineering textile materials for the intended applications. This
keynote paper introduces the modelling techniques for different textile materials and structures and reports on
the latest progress in modelling textiles for ballistic protection. In this area of work, it is shown that the
modelling results agreed with the experimental work well and modelling tool can be used to guide the
ballistic materials effectively.
Keywords: textiles; structures; behaviour; modelling; ballistic protection
1. Introduction
Textiles refer to fibres and fibre assemblies that are principally used as raw materials for different types
of products. Under this definition, textiles will include fibres, yarns, and fabrics. For garments, beddings,
curtains, floor coverings, as well as technical end-use (such as a type of textile composites), textile fabrics
are the raw materials, providing not only the appearance, texture and decorative features but also the various
properties that make the textile suitable for the intended applications. Textiles are a popular type of materials
that has been widely used domestically and industrially [1].
However, textiles as a type of materials is special when compared to materials such as metal. Textiles are
far from homogenous and isotropic because they are assemblies of fibres. In addition, fibres are made of
wide range of different chemical compositions, and when different fibres are used for making textiles, the
physical and chemical properties can vastly different. Because of all these special features, modelling of
textile structures and behaviour has always been an attention focus.
The textile hierarchy ranges from fibre as the basic element. Fibres are the construction units of yarns
and some non-woven fabrics. Then yarns are used as components for making fabrics based on the weaving
knitting and braiding technologies. It is essential to understand the fibre behaviour which is largely
determined by the chemical structure of the polymer and physical configuration of the molecular chain.
Based on the fibres, it could be claimed that the behaviour of a textile assembly is a function of the property
of the building block and the way how these building block are constructed in the assembly. Following this
logic, the behaviour of yarn depends on the fibre property and the yarn construction and the fabric behaviour
is determined by the composing yarn property and construction of the fabric. A Fabric contains tremendous
amount of fibres of the same or different types, and there are endless ways that a fibre is configured
individually or collectively in a fabric. Phenomena such as these make the modelling of textiles very
challenging.
2. Weave modelling
Woven fabrics are produced by interlacing two systems of yarns perpendicular to each other; the one in
the length direction of the fabric is the warp and the one that goes in the width direction of the fabric is the
weft. There are many different ways of interlacing the warp and weft yarns into a fabric, and a particular plan
for constructing a fabric is know as a weave. Weaves can be classified into four different categories, which
are elementary weaves, derivative weaves, combined weaves, and complex weaves. In addition to these,

+
Corresponding author. E-mail address: xiaogang.chen@manchester.ac.uk
Published by World Academic Press, World Academic Union
Xiaogang Chen, et al: Modelling of Textile Structure for Advanced Applications
72

woven fabrics can be made to have a considerable thickness from multiple sets of yarns in each of the two
directions, and these are sometimes termed as 3D woven fabrics.
The elementary weaves are defined as those having two floats in the weave repeat and with one the float
length being 1. It typically includes the plain weave, simple twill weaves, and satin/sateen weaves [2]. The
weave diagrams of some of these weaves are illustrated in Figure 1. The elementary weaves can be
manipulated to derive new weaves.

(a) (b) (c)
Figure 1 Elementary weaves (a) the plain weave (b) 2/1 twill (c) 5-end satin
The derivative weaves are created from the three types of elementary weaves. From the plain weave, two
methods can be applied to create new weaves. The plain weave can be extended in warp, weft or both
directions in order to derive new weaves. Extending the plain weave in the warp direction will result in warp
rib weaves, and extending in the weft direction leads to weft rib weaves. When the plain weave is extended
in both directions, hopsack weaves will be created. Figures 2(a), (b) and (c) show a warp rib, a weft rib and a
hopsack weave respectively. A plain derivative can also be created following the plain weave logic, where
the derivative weave will have 4 quarters with the adjacent ones having opposite images, leading to a basket
weave. This is illustrated in Figure 2(d).

(a) (b) (c) (d)
Figure 2 Plain derivative weaves (a) a warp rib (b) a weft rib (c) a hopsack and (d) a basket weave
Twill weaves are featured by the twill lines in either Z or S direction. Therefore, ways for changing the
twill line thickness and direction play important role in creating twill derivative weaves. When the mirroring
technique is applied, twill lines will be made to change their directions while keeping the continuity. This
technique is used to create waved weaves and diamond weaves. Figure 1.14 shows a horizontal waved weave,
a vertical waved weave, and a diamond weave derived from the
3 1
2 2
S twill. When changing twill line
direction and breaking the continuity, i.e., by applying the inverse mirroring technique, herringbone weaves
and diaper weaves can be achieved. Figure 1.15 displays a horizontal herringbone weave, a vertical
herringbone weave, and a diaper weave based on again the
2 2
1 3
S twill.
More fanciful twill derivatives can also be created using different rules. Figure 3 demonstrates two other
types of twill derivative weaves, an entwined weave and a saw tooth weave.

(a) (b)
Figure 3 Twill derivatives achieved using other principles (a) an entwined weave and (b) saw tooth weave
The traditional weaving technology is also capable of weave fabrics with thickness, or 3D fabrics. 3D
fabrics can be made as broad solid panels (3D solid), with porous cross-sections (3D hollow), or 3D shapes.
3D solid woven fabrics can be manufactured based different principles such the multilayer, orthogonal and
angle interlock. Figure 1.17 shows the 3D model as well as the weave for an orthogonal woven fabric with 4
layers of warp yarn.
Modelling the structure of the woven fabrics is regarded as an important step towards the computerised
generation of weaves. Chen and his colleagues started the structural modelling of woven fabrics by defining
JIC email for contribution: editor@jic.org.uk
Journal of Information and Computing Science, Vol. 5 (2010) No. 1, pp 071-080
73


weaves into regular and irregular. A regular weave is one whose float arrangement and the step number do
not change in a repeat of the weave. All other weaves are defined as irregular weaves. Many commonly used
weaves are regular them selves or further developed from regular weaves. Whilst each type of the irregular
weave would need a distinct mathematical model to describe its construction, all regular weaves will need
same mathematical model for it construction.

(a) (b)
Figure 4 The (a) model and (b) weave for an orthogonal woven fabric with 4 warp layers
A model describing the regular weave construction was reported by Chen et al [2], which yields the 2D
binary weave matrix, W, upon the specification of the float arrangement, F
i
, and the step number, S. Suppose
that w
x, y
is the element of this matrix at co-ordinate (x, y), where 1 s x s R
e
and 1 s y s R
p
with R
e
and R
p

being the warp and weft repeat respectively, then the first column of the weave matrix can be generated
using the following equation:

=
integer even an is if 0
integer odd an is if 1
, 1
i
i
W
y
(1)
where y = ; and 1 s is N to 1
1 1

= =
|
|
.
|

\
|
+
i
j
j i
i
j
j
F F F
f
. N
f
is the number of floats in the float arrangement. Then,
the rest of the matrix will be assigned values as follows:
W
x, z
= W
1, y
(2)
where
( ) | | ( ) | | { }
( ) | | ( ) | | { }
( ) | | ( ) | | { }

> + +
s + s +
< + + +
=
; 1 if 1
; 1 1 if 1
; 1 1 if 1
p p
p
p
R x S y R x S y
R x S y x S y
x S y R x S y
z 2 s x s R
e
; and 1 s y s R
p
.
Chen and colleagues also worked on weaves for other 2D fabrics and 3D fabrics [3] [4].
3. Geometrical modelling of woven fabrics
The performance of a textile fabric is basically a function of the property of the constituent fibres/yarns,
and the geometrical construction of the fibres/yarns in the fabrics. Study on the geometry of fabrics has been
continuing for almost a century. Geometrical models of fabrics have led to the estimation of some structural
and physical properties of the fabrics, such as the areal mass and the porosity, and the results from the
modelling have been used as guidance to fabric manufacture in giving the maximum areal density of the
fabrics. Geometrical modelling of textile assemblies becomes more important nowadays as the geometrical
models are, arguably, the most reliable solution in providing geometrical information of the textile
assemblies for finite element (FE) analysis for performance simulation.
Peirces work in 1937 [5] is regarded as the beginning of modelling woven fabric geometries. Under
certain assumptions including circular yarn cross-section, complete flexibility of yarns, incompressible yarns
and arc-line-arc yarn path, he derived the following equations describing the geometry of the plain woven
fabrics. The cross-section of the plain woven fabric based on Peirces assumption is shown in Figure 5.
JIC email for subscription: publishing@WAU.org.uk
Xiaogang Chen, et al: Modelling of Textile Structure for Advanced Applications
74


Figure 5 Peirces model for the plain woven fabric
p e
d d D + =
(3) (4)

D h h
p e
= +
1 =
p
e
e
p
l
c
(5)
1 =
e
p
p
p
l
c
(6)

e e e
D
e p
D l p u u u sin cos ) + ( =
(7) (8)
p p p
D u u u sin cos ) +
p e
D l p ( =
) cos 1 ( sin ) (
e e e e e
D D l h u u u + = (9)
) cos 1 ( sin ) (
p p p p p
D D l h u u u + =
(10)
where
e
h
, - the modular heights of the warp and weft yarns normal to the neutral plane of the fabric
p
h
e
c , - the crimps of the warp and weft yarns
p
c
D -the sum of the diameters of the warp and weft yarns
e
d
, - the diameters of the warp and weft yarns
p
d
e
p
, - the thread spacing between adjacent warp and weft yarns
p
p
e
l
, - the modular lengths of the warp and weft yarns in one repeat
p
l
e
u
,
p
u
- the weaving angles of warp and weft yarns
Subscripts e and p in the variables above refers to warp (ends) and weft (picks) respectively.
There are thirteen variables in these eight equations. Therefore, with five variables known, such as the
two spacings ( , ), the two yarn diameters ( , ) and one crimp (either or ), these simultaneous
equations can be solved. Ai [6] presented an algorithm to calculate the geometry assuming that five
variables, , , , and one of and are specified. If , , , and are known, the other
fabric parameters can be worked out.
e
p
p
p
p
p
e
d
e
d p
d
e
c
p
p
c
e
c
e
p p
d
e
c
p
c
e
p p
p
e
d
d
The yarn cross-section in a real fabric is hardly circular because of the pressure between the war and
weft yarns during the weaving process. Peirce himself proposed an alternative model for the plain woven
fabric assuming the yarn cross-section to be elliptical. It proved to be mathematically too complicated to
describe the relationship among the structural parameters. Peirce model of plain woven fabrics was extended
by others notably Kemp [7] who assumed the yarn cross-section is racetrack shaped and Shanahan and
Hearle [8] who proposed a lenticular yarn-cross section. These extended models kept all assumptions Peirce
used except for the yarn cross-section, and are regarded as Peirce derivative models.
4. FE modelling of woven structures
Based on the achievements made in weave modelling and fabric geometric modelling, solid models for
different types of woven fabrics are created using the algorithm. For example, a programme called
UniverWeave was developed to create woven fabric geometrical models efficiently. The defined geometry
can be picked up by major FE software packages to carry out FE simulations for material and component
analysis.
4.1. FE modelling of filtration through fabrics
Since the geometry and porosity of the fabric filter is determined by the weave pattern and the various
JIC email for contribution: editor@jic.org.uk
Journal of Information and Computing Science, Vol. 5 (2010) No. 1, pp 071-080
75


parameters of yarns constituting the fabric [9], it is important to optimise its structure to achieve the most
efficient filtration. The past and current practice is to rely on the practical skill and experience of the fabric
designers and empirical trials. Readily available computational power provides an opportunity to develop
computer-aided design (CAD) procedures. CAD software would enable predictions of filter performance to
be made, leading to improved filters and reduced cost of trials. In the first stages of developing CAD
programs, it is necessary to produce good models of fabric structure and then predict the flow through the
fabrics.
Among the numerous outputs from the analysis, the fluid pressure, fluid velocity and shear stress on the
fabric are used as the performance indices [10]. Fluid pressure is read on the front face and the back face of
the filter fabric. Fluid velocity is taken on the planes that are one mesh size away before and after the filter
fabric and on the fabric centre (middle) plane. The shear stress, on the other hand, is measured at the front
side and inner-side of the yarns constituting the fabric. The fluid velocity and pressure through the chamber
are also simulated, which give an overall effect of the fabric filter on the fluid flow. Figure 6 shows the
positions where the data were extracted in relation with the fabric and to the direction of the flow.

Figure 6 Fabric in relation to planes before, middle, after, and through
In each analysis, 3 different fabric models (all plain) are used. Every fabric model is specified using warp
yarn linear density (t
1
), weft yarn linear density (t
2
), warp density (d
1
), weft density (d
2
), warp crimp (c
1
),
cross-sectional shape of the yarns and the Width to Height Ratio (WHR) of the yarn cross-section. Warp
crimp is then a dependent parameter. The inlet pressure for all cases is 3 bars and the operating pressure 1
bar. Density and viscosity of the fluid were assumed to be constant, corresponding to the isothermal
approach. Liquid-water was used as the Newtonian fluid (with density of 998.2 kg/m
3
and viscosity at 10
-3

kg/m.s). As an example, when the cross-section of the yarns in the filter fabrics are taking circular, racetrack
and lenticular shapes, Figure 7 summarises the effect of yarn cross-sectional shape on fluid pressure on the
front and back surfaces.

Figure 7 Effects of varying yarn cross-sectional shape on fluid pressure
In this experiment three rather ideal yarn cross-sectional shapes namely circular (FS4), racetrack (FS5)
and lenticular (FS6) shapes are investigated. From circular to racetrack to lenticular, the yarn width increases
as its height decreases. Figure 7 illustrates influence of variation of the yarn cross-sectional shape on fluid
pressure exerted to the front and back surfaces of the fabric. By changing yarn cross-sectional shapes from
circular to racetrack to lenticular, the fluid pressure is increased on the front face of the fabric due to higher
flow resistance and drops dramatically at the back face.
JIC email for subscription: publishing@WAU.org.uk
Xiaogang Chen, et al: Modelling of Textile Structure for Advanced Applications

4.2.
JIC email for contribution: editor@jic.org.uk
76
FE modelling of ballistic impact through fabrics
FE modelling of ballistic impact on fabrics is another field that is much needed in order to understand the
strain/stress distribution in each of the fabric layers and among all fabrics layers. Results from such work
provide guiding information for body armour engineering. Modelling of ballistic fabrics is carried out on two
levels, one a single layer of fabric, another layered panels of fabrics.

Figure 8 The geometrical model and constraints
Figure 8 is the geometric model of the fabric and the impacting projectile. A quarter of the fabric and
that of the projectile is considered because of symmetry, the fabric involved 8 warp and 8 weft yarns
constructed in the plain weave. This is a square fabric with the same value of warp and weft densities (7.6
threads/cm). The material type is Kevlar whose specific density is 1.55 g/cm
3
. The following describes the
boundary conditions For the projectile, the translational freedom along X and Y axes and the rotational
freedom along Z axis are constrained and set to zero, i. e., v
x
=v
y
=0. For the fabric, the translational freedom
perpendicular to the symmetrical plane and the rotational freedom in the symmetrical plane are constrained
and set to zero, i.e., U
x
=UR
y
=UR
z
=0 and U
y
=UR
x
=UR
z
=0. The circumference of the quarter of the target
fabric is fixed, i.e. U
x
=U
y
=U
z
=UR
x
=UR
y
=UR
z
=0.
460
470
480
490
500
510
520
460 470 480 490 500 510 520
Residual velocity - experiment (m/s)
R
e
s
i
d
u
a
l

v
e
l
o
c
i
t
y

-

s
i
m
u
l
a
t
i
o
n

(
m
/
s
)

Figure 9 Agreement between the model and experiment

Figure 10 Stress distribution on an impacted fabric at (a) t=0 s (b) t=0.75 s (c) t=1.37 s (d) t=5.25 s (e) t=6.0 s
and (f) t=8.12 s
Figure 9 shows that the modelled exit velocities agree well with the measured exit velocities, with the
Journal of Information and Computing Science, Vol. 5 (2010) No. 1, pp 071-080
77


correlation coefficient being 0.9939. With the models validated by the experimental data, a series of FE
simulation was carried out. Figure 10 reveals the ballistic impact process, with the projectile impact velocity
v0=494.217m/s.
It is clear in Figure 10 that the distribution of stress caused by the ballistic impact on the fabric is mainly
along the warp and weft yarns in the fabric before the projectile penetrates the fabric. This implies that the
currently used plain woven fabric may not be the most efficient construction for ballistic applications,
because such constructed fabric is unable to mobilise more areas of fabric to absorb the impact energy. To
improve the fabrics ability to absorb impact energy, fabrics with better yarn gripping have been designed and
manufactured. The test results confirmed the superiority of such fabrics.
475
480
485
490
495
500
505
510
0 2 4 6 8 10 12 14 16
Time t(s)
P
r
o
j
e
c
t
i
l
e

v
e
l
o
c
i
t
y

v
(
m
/
s
)
v=494.217
v=506.466

Figure 11 Modelled project velocity change due to impact
The models also give information that enhances understanding on of the impact process. Figure 11
shows the change of the projectile velocity due to the impact on the fabric. The curves for both cases
demonstrate similar trend. The curves indicate that the initial impact causes a sharp reduction of projectile
velocity before the fibres start to break at about 1 s into the impact. Projectile going through the fabric
further reduces its velocity. The penetration of the projectile for both cases seems to have taken place at
around 8 s.
2.10
2.11
2.17
2.24
2.14
2.12
2
2.05
2.1
2.15
2.2
2.25
32 32 34 36 37 40
Impact energy E(J)
E
n
e
r
g
y

a
b
s
o
r
b
e
d

b
y

f
a
b
r
i
c


E
(
J
)

Figure 12 Energy absorption vs. impact energy
Figure 12 reveals that for a given assembly of ballistic fabrics there exists an impact energy that relates
to the maximum energy absorption of the assembly of the fabrics. The impact energy is only a quarter of the
energy applied to the fabric assembly due to the geometrical symmetry of the projectile. Under this
circumstance, the fabric assembly absorb most of the energy at 36 J. This provides support to the use of V50
in ballistic test.
The results from simulating single layer fabrics are valuable information for fabric design so that the
fabric is more absorbent to impact energy. It is also important to consider how the fabric is used in the fabric
assembly. Obviously, different fabrics/materials can be used at different position in the assembly because the
impact action and reaction are different at different layer. Even if the same fabric is used, the fabric
orientation in the assembly will also matter as fabric orientation has a direct influence on the strain/stress
JIC email for subscription: publishing@WAU.org.uk
Xiaogang Chen, et al: Modelling of Textile Structure for Advanced Applications
78

distribution. Figure 13 (a), (b) and (c) shows that fabric layers with different orientation angle leads to better
impact energy absorption.

(a) A 2-layer assembly

(b) A 3-layer assembly

(c) A 3-layer assembly
Figure 13 Effect of fabric orientation on energy absorption

(a) Aligned 4-layer (b) Angled 4-layer
Figure 14 Stress distribution of the 4-layer fabric assembly
From Figure 13 demonstrates that up to 4 layers of fabrics, a more evenly aligned fabric assembly
absorbs the most of impact energy, and the aligned fabric assemblies are related to the least energy
JIC email for contribution: editor@jic.org.uk
Journal of Information and Computing Science, Vol. 5 (2010) No. 1, pp 071-080

JIC email for subscription: publishing@WAU.org.uk
79
tal results show the same trend.
ppens when a high velocity
so demonstrated that fabric assemblies with evenly
impact energy than the aligned fabric assembly.
9/1 and RT/COM/5/030. The author also appreciates the important
yaraj, Ali Zanarboland, Danmei Sun and Ying Wang as
d assistants.
xford: Woodhead publishing Ltd, 2009.
3D woven
1, No. 2): 356.
ications. Text. Res. J. 1999, 69(9): 648.
fabrics. J Text Inst. 2006, 97(1):
03.
lar threads. J. Text. Inst. 1958, 49: T44.
J. Text. Inst. 1978, 69(4): 92.
ure
0] M. A. Nazarboland, X. Chen, J. W. S. Hearle, R. Lydon, and M. Moss. Modelling and simulation of filtration
through woven media. International Journal of Clothing Science and Technology. 2008, 20(3): 150-260.

absorption. Experimen
5. Conclusions
As an important field for modelling and simulation, this paper explained the technique developed by the
author for woven textiles. The weave modelling is a mathematical description where the weave is
parameterised. This technique has been adopted in CAD software for weave design. Geometrical modelling
is explained by adopting Peirce model. This work enables fabric models to be created and exported for
property analysis using the FE packages. Two examples were used to demonstrate the application of the
geometrical models. Through the filtration modelling, information was created giving the relationship
between the fabric geometry and the fluid behaviour passing through the fabrics. The second example is on
the performance of ballistic fabrics. The modelling lead to understanding what ha
rigid projectile impacts on a soft woven fabric. It al
angled orientation absorb more
6. Acknowledgement
The author thanks the EPSRC and the MOD/Dstl for the continued support in the research, in particular
for the two grants EP/G04282
contributions made by Andrew Ai, Pranut Poti
research students an
7. References
[1] X. Chen (Ed.). Modeliing and predicting textile behaviour. O
[2] X. Chen, R.T. Knox, D.F. McKenna, R.R. Mather. Automatic generation of weaves for the CAM of 2D and
textile structures. J. Text. Inst.. 1996, 87( Part
[3] X. Chen, P. Potiyaraj. CAD/CAM of the orthogonal and angle-interlock woven structures for industrial
appl
[4] X. Chen, H. Wang. Modelling and computer aided design of 3D hollow woven
79.
[5] F.T. Peirce. The Ggeometry of cloth structure. J. Text. Inst. 1937, 28: T45
[6] X. Ai. Geometrical modelling of woven and knitted fabrics for technical applications. MPhil Thesis, UMIST, 20
[7] A. Kemp. An extension of Peirce cloth geometry to the treatment of noncircu
[8] W.J. Shanahan, J.W.S. Hearle. An energy method for calculations in fabric mechanics, part II: examples of
application of the method to woven fabrics.
[9] S. Backer. The Relationship Between the Structural Geometry of Textile and Its Physical Properties, I: Literat
review. Text. Res. J. 1948, 18: 650-658.
[1














Xiaogang Chen, et al: Modelling of Textile Structure for Advanced Applications

JIC email for contribution: editor@jic.org.uk
80