Vous êtes sur la page 1sur 10

Int. J. Advanced Intelligence Paradigms, Vol. X, No.

Y, xxxx 1

Simulation of longitudinal geometric yarn structure

Ionut Neagu* and Pascal Bruniaux


Ecole Nationale Supérieure des Arts et Industries Textiles,
59100, Roubaix, France
E-mail: ionut.neagu@ensait.fr
E-mail: pascal.bruniaux@ensait.fr
*Corresponding author

Dorin Avram and Vasile Airinei


Technical University,
‘Gh. Asachi’,
Bd. Dimitrie Mangeron, nr. 35, 70050 Iasi, Romania
E-mail: davram@tex.tuiasi.ro
E-mail: airinei@tex.tuiasi.ro

Abstract: A useful yarn model is certainly a valuable tool for yarn


technologists involved in the process of product development that is why this
study is to present a three-dimensional environment for the virtual design and
the simulation of a yarn, the emphasis being on the yarn structure. To integrate
the model of the yarn in the fabric model, we must define a geometry adapted
to the yarn and describe the behaviour of the yarn model, for that, an algorithm
of modelling the fibres in 3D has been developed. Beyond the scientific
character of the forecast, one can use the same computerised base to virtually
simulate the tests and laboratory experiments or to generate realistic animations
in artistic matter and design of textile clothing.

Keywords: yarn; simulation interfaces; model design; virtual reality; computer


aided design.

Reference to this paper should be made as follows: Neagu, I., Bruniaux, P.,
Avram, D. and Airinei, V. (xxxx) ‘Simulation of longitudinal geometric
yarn structure’, Int. J. Advanced Intelligence Paradigms, Vol. X, No. Y,
pp.000–000.
Biographical notes: Ionut Neagu is a Researcher at the ENSAIT/GEMTEX
Laboratory, in Roubaix, France. He received his PhD in Automation and
Industrial Computing from the Université des Sciences et Technologies de Lille
and graduated Magna Cum Laude in 2010.

Pascal Bruniaux is a Professor at the ENSAIT in Roubaix, France and currently


the Head of the Apparel Department at the institution. He received his PhD in
Automation from the Université des Sciences et Technologies de Lille I. He
was involved in several EU research projects. His research interests are
3D simulation, fibber-yarn-garment interface, mass customisation, and 3D
prototyping.

Copyright © 200x Inderscience Enterprises Ltd.


2 I. Neagu et al.

Dorin Avram is a Professor at Technical University ‘Gh. Asachi’, Iaşi, Faculty


of Textiles and Leather Engineering. He received his PhD in Industrial
Engineering from the Université des Sciences et Technologies de Lille I. His
research interest is applied research in yarn structure and technology.

Vasile Airinei received his BA and his MSc in Computer Engineering from the
Technical Univerisity ‘Gh. Asachi’ Iaşi. His research interests are
computational intelligence in hardware design, data structures and algorithms,
methods of graphics programming, and parallel programming. He has
published in Romanian computer science literature. He has a large experience
as C++ Programmer, working for some international companies in Romania.

This article is a revised and expanded version of a paper entitled ‘Simulation of


longitudinal geometric yarn structure’ presented at North-American Simulation
Technology Conference (NASTEC) 2009, held at the Georgia Tech Global
Learning Center, Atlanta, USA on 26–28 August 2009. It has been revised and
updated.

1 Introduction

It is necessary to model the yarn before modelling the fabric in order to develop a
three-dimensional environment for the virtual design and the simulation of clothing
(Bruniaux et al., 2005). Why this approach to the study of yarn geometry? The answers
are simple enough. First of all, the industry is not completely pleased with actual
simulators that achieve good overall performance but lack sensitivity and coherence
(Djaja et al., 1992). Secondly, the absence of a multilevel model that starts from fibres.
That is why we must define a geometry adapted to the yarn and describe the behaviour of
the modelled yarn so we could integrate the yarn model into the fabric model (Jiang and
Chen, 2005; Keefe et al., 1992).
The geometric complexity of textile structures and the anisotropy (non-linear
behaviour of fibres) represent the two main problems that prevent the simulation of
interrelationship between fibre and yarn. The response of fibres to a mechanical load is
influenced by their nature and interactions. Apart the above-mentioned observation, we
also underline the major difficulty in describing the physical behaviour of small fibres
that often take an irregular form (Keefe, 1994; Langenhove, 1997). Generally, the
characterisation of textile material follows a hierarchical strategy based on a
multidimensional and structural concept leading to successive representations of
geometrical models.
Most of the researchers have developed simulators directly from the yarn structure
without taking into account the fibre structure and its characteristics. It is called
geometric structure because the analysis of the yarn structure is based on geometric
elements of yarns and fibres, such as: yarn diameter, position of fibre radius, pitch fibre
angle, fibre length, length of a fibre element, surface and volume of fibres and yarns. It is
known that fibres have a random position in the yarn which is manifested by a
variable number of fibres in cross-section and different radial positions along the yarn
(Legrand et al., 2000b).
As a result, the description of the yarn structure is achieved through the analysis of
the fibre position. There are many research papers which focus on the mechanic
Simulation of longitudinal geometric yarn structure 3

modelling of the geometry of the yarn structure but there is no model which takes into
account the cause-effect of fibre integration in the yarn structure (Legrand et al., 2000a;
Ngo Ngoc et al., 2002). Much effort can be saved if we avoid the trial and error approach
and use instead, the simulation model (Grishanov et al., 1997). Even if there are many
studies having as main theme the modelling and simulation of textiles, none of them
focuses on the dynamic yarn behaviour, i.e., none of them takes into consideration fibre
characteristics when it comes to modelling the yarn structure.

2 The geometrical model

We chose the idealised helical yarn geometry for the purpose of simulating the yarn
structure. The main algorithm is described as follows:

Figure 1 The idealised helical yarn geometry

R rs

lSR
h
βR
β
lS

2πr
2πR
where
R yarn radius
rs fibre radius
β twist angle (angle between the fibre tangent and the yarn axis)
βR fibre twist angle located on the yarn radius
lS length of the fibre helix, length corresponding to the layer radius
lSR helix length corresponding to the yarn radius
h helix pitch.
Now we will describe the algorithm in the following steps:
4 I. Neagu et al.

2π rs 2π rsT
1 the radial twist angle is determined by the relation tg β = = , where T is
h 1, 000
the yarn twists (twist/m)

2 the twists are determined by Köchlin’s relation: T = α m * N m , where Nm is the


average yarn count (m/g); and αm is the twist multiplier
T
3 the helix pitch can also be calculated as h =
1, 000
4 we calculate the yarn diameter by means of the following equation
4
D = 2R = , where D is the yarn diameter (mm), ρF the yarn density
NmF ⋅ π ⋅ ρ F
(g/cm3) and NmF the average yarn count (m/g)
5 the yarn density is determined by the equation ρ F = v ⋅ ρ f , where ν is the packing
rate takes values between 0, 6 and 0, 7 and ρf is the average fibre density determined
by the relation v = f(T)
6 in order to calculate the average fibre density, we use the following algorithm
4
ρf = , where d the average fibre diameter (µm)
π ⋅ Nm f ⋅ d 2

4 ⋅16
7 then we calculate the mean fibre diameter using d =
Nm f ⋅ π ⋅ ρ f

8 the number of fibres in a cross-section of the yarn can be obtained by using the
Nm f
equation ns =
NmF

100
9 the limit variation factor is determined by the relation CVlim = , where ns is the
ns
number of fibres in yarn cross section of the yarn
CVef ⋅ D
10 the mean square deviation is determined by the relation σ = , where CVef is
100
the factor of effective variation of yarn diameter and is determined by the relation
CVef = CVlim ⋅ I , and where I represent the irregularity factor that we adopt from the
Uster statistics ranges between I = 1.2 ÷ 1.3.
11 the computation of diameter limits are obtained with the use of equation
Dmin = D –1,96* s
, where σ is the standard deviation of yarn diameter.
Dmax = D + 1,96* s
Simulation of longitudinal geometric yarn structure 5

3 Computation

The twist angle and the fibre trajectory are calculated using the result of in-plane
development of the helix portion between sections (M) and (N). The result is a right
triangle ABA’ as shown in Figure 2, whose sides are AB = 1 and AA ' = rϕ .
The angle (β), formed between the tangent to the helix and the direction of the yarn
axis, is called twist angle.

Figure 2 The position of the fibre within the yarn structure (see online version for colours)


1 Obtaining the equation tg β = , where φ represents the rotation angle (expressed
l
in radians) of section M with respect to section N. If we represent the rotation angle
in terms of the number of complete rotations made by section M with respect to N.
2π nr
2 Then from equation tg β = and since the ratio n/1 represents the twists, we
l
obtain tgβ = 2πrT, where β the twists angle (expressed in degrees), r represent the
fibre radius (mm), and T the yarn twists (twists/mm). This relation is very important
for the tubular structure of yarns and represents the fundamental twist formula. It
brings together the main parameters of the tubular structure of yarns.
3 The result is that for a yarn with tubular structure whose twist is (T), the twist angle
will be wider if the fibres are on the outer boundary of the yarn.
• for fibres at yarn core level: r = 0 and β = 0
• for fibres on the outer boundary of the yarn.
4 If r = R and β = βR, then tgβR = 2πRT, where R is the yarn radius (mm) and the yarn
twists (twists/mm).
As a result of the twisting process, fibres take the shape of cylindrical helixes, except for
the central fibre which remains straight.
6 I. Neagu et al.

There is a cylindrical helix situated on a cylinder whose radius is r, as shown in


Figure 3. On the helix, we consider the point M which is characterised by:
• φ – the angle of helix AOM '
• r – helix radius, OM ' = r.

Figure 3 Calculating the length of the helical arc (see online version for colours)

In order to calculate the fibre length we use the following


x = OA = r ⋅ cos ϕ
1 the parametric equations of point M are: y = Ob = sin ϕ , where k represents the
z = MM ' = k ⋅ ϕ
helix rate and is equal to the ratio between dz and dφ
dz
2 we obtain the helix rate k =
ϕ
3 then using the equation dz = k ⋅ dϕ and through integration we have z = kφ + c

ϕ = o ⇒ z = 0, soc = 0
4 out of these conditions we obtain
ϕ = 2π z=h

h
5 it results k , where h is the helix pitch

6 we consider the helix length ds as shown in Figure 2, the length ds is written as
ds = dx 2 + dy 2 + (dz ) 2

⎛ h2 ⎞
7 by replacing dx, dy and dz, we have ds = ⎜⎜ r 2 sin 2 ϕ + r 2 cos 2 ϕ + 2 ⎟⎟ dϕ 2
⎝ 4π ⎠
Simulation of longitudinal geometric yarn structure 7

8 from the previous computation, we have the relation


⎛ h2 ⎞ h 1
ds = dϕ ⎜⎜ r 2 + 2 ⎟⎟ = rdϕ 1 + 2 2 = rdϕ 1 + 2 2 2
⎝ 4π ⎠ 4π r 4π r T

9 for a helix rotation angle equal to 2π, we obtain the length of loop (corresponding to
a helix pitch), situated at a distance r from the centre of the cylinder, it results
1
s = 2π r 1 + 2
tg β

2π r
10 by replacing, 4π 2 r 2T 2 = tg 2 β the relation can be rewritten as follows s =
sin β

11 thus, the fibre length corresponding to a pitch of the right cylindrical helix represents
the loop length s, which depends on the radial position of the fibre r and the twist
angle of the fibre.

4 Simulation results

We have developed a software programme in C++ language, based on the previously


presented algorithm. We used a multiple-document interface for the dialog windows and
OpenGL API to represent the main 3D window with the 3D structure of the yarn. The
yarn section is rendered by open packing distribution, taking into account the average
number of fibres; the core is represented by a straighten fibre around which the other
fibres are disposed in layers. In Figure 4, the resulted cross-section are presented, every
fibre layer has a different colour in order to be easily identified. Table 1 indicates how we
calculate the layer output.
Table 1 Layers output algorithms

Layer number Layer radius ns/layer Σm*


1 rs3 m*1 m*1
2 rs2 m*2 m*1+m*2
3 Rs3 m*3 m*1+m*2+m*3
Σm* = ns

180
We calculate the number of fibres in every layer by m = , where ns is the
1
arcsin
2( s − 1)
number of fibres in yarn cross-section, rs represent the layer radius.
The resulted cross-section structure is represented in Figure 4 and is similar with the
result of Grishanov et al. (1997) and Grishanov and Harwood (1999) which make our
model valid.
8 I. Neagu et al.

Figure 4 Fibres cross-sectional distribution (see online version for colours)

In Figure 5, we can see the main panel, with the input data at the top and the output data
at the bottom.

Figure 5 The main panel with input and output data (see online version for colours)

The modification of input data determines the automatic computation of output data, due
to an update data function. The 3D viewer window is presented in Figure 6; on the
upper-left side, we have the saving options of the structure, and also a slider for changing
the twist angle.
Simulation of longitudinal geometric yarn structure 9

Figure 6 The 3D view panel with the yarn structure (see online version for colours)

The input data including yarn and fibre characteristics and are indicated in Table 2, and
the output data in Table 3.
Table 2 The input data

Input data Not. Value MU.


Average yarn count NmF 39.50 m/g
Average fibre count Nmf 1,792 m/g
Packing rate Ν 0.50 -
Yarn density ρF 0.76 g/cm3
Fibre density ρfb 1.31 g/cm3
Yarn torsion T 820 t./m
Irregularity index I 1.20 -
Yarn diameter Dmed 0.287 mm

Table 3 The output data

Output data Not. Value MU.


Yarn minimum diameter Dmin 0.27 mm
Calculated yarn diameter Dmedc 0.31 mm
Yarn maximum diameter Dmax 0.34 m/g
Fibre average diameter Dfmed 0.05 mm
Number of fibres in cross-section Ns 45 -
Yarn diameter coefficient of variation CvD 17.82 %
Limit coefficient of variation Cvlim 15 %
Effective coefficient of variation Cvef 18 %
Standard deviation of yarn diameter Σ 51 -

The difference between the average yarn diameter obtained by means of microscopic
measurements and the calculated diameter is of 22%, verifies Barella’s study (Avram,
2001).
10 I. Neagu et al.

5 Conclusions

In conclusion, this modelling strategy that we bring is promising because it corresponds


to the user demand. It has the advantage of reflecting statistically the global problem of
modelling. Of course, these works need to be applied to other yarns to broaden the range
of models validity.
Improving the next generation model can that can merge the fibres with a 3D surface
model to take into account yarn irregularities. Thus, the two merged models can be
beneficial and represent the starting blocks to build a complete mechanical model of yarn
from fibres.

References
Avram, D. (2001) Yarns Structure, ISBN 973-7994-46-9, Iasi.
Bruniaux, P., Legrand, X., Ngo Ngoc, C. and Castelain J.M. (2005) ‘Modélisation mésoscopique
du tissu par le fil’, XIIth Romanian Conference of Textiles and Leadership, Iasi, Romania.
Djaja, R.G., Moss, P.J. and Carr, A.J. (1992) ‘Finite element modeling for an oriented assembly of
continuous fibers’, Textile Research Journal, Vol. 62, No. 8, pp.445–457.
Grishanov, S.A. and Harwood, R.J. (1999) ‘A model fiber migration in staple-fiber yarn’, Journal
of Textile Institute, Part 1, No. 3, pp.298–321.
Grishanov, S.A., Lomov, S.V., Cassidy, T. and Harwood, R.J. (1997) ‘The simulation of the
geometry of a two component yarn, Part II: Fiber distribution in the yarn cross-section’,
Journal of Textile Institute, Part 1, Vol. 88, No. 4, pp.352–372.
Jiang, Y. and Chen, X. (2005) ‘Geometric and algebraic algorithms for modeling yarn in woven
fabrics’, Journal of Textile Institute, Vol. 96, pp.237–245.
Keefe, M. (1994) ‘Solid modeling of fibrous assemblies: Part I: twisted yarns’, Journal of the
Textile Institute, Vol. 85, No. 3 pp.338–349.
Keefe, M., Edwards, D. and Yang, J. (1992) ‘Solid modeling of yarn and fiber assemblies’, Journal
of the Textile Institute, Vol. 83, No. 2, pp.185–196.
Langenhove, L.A. (1997) ‘Simulating the mechanical properties of a yarn based on the properties
and arrangement of its fibers, Part I: The finite element model’, Textile Research Journal,
Vol. 67, No. 4, pp.263–268.
Legrand, X., Bruniaux P. and Ngo Ngoc, C. (2000) ‘Modeling and study of the dynamical behavior
of a yarn’, 16th IMACS World Congress, 21–25 August, Lausanne, Switerland.
Legrand, X., Bruniaux, P. and Vasseur, C. (2000) ‘A new approach for modeling woven fabrics’,
WAC’2000, Fourth Biannual World Automation Congress, Wailea, Maui, Hawaii, USA.
Ngo Ngoc, C., Bruniaux, P. and Castelain, J.M. (2002) ‘Constrained dynamic yarn modeling’,
Textile Research Journal, Vol. 72, No. 11, pp.1002–1008.

Vous aimerez peut-être aussi