See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233001945

Fabric Engineering by Means of an Artificial Neural Network

Article in The Journal of The Textile Institute · January 2002

DOI: 10.1080/00405000208630570

CITATIONS READS

22 37

3 authors, including:

Arindam Basu
Northern India Textile Research Association
199 PUBLICATIONS 462 CITATIONS

SEE PROFILE

All content following this page was uploaded by Arindam Basu on 28 March 2018.

The user has requested enhancement of the downloaded file.

 Journal of the Textile Institute

ISSN: 0040-5000 (Print) 1754-2340 (Online) Journal homepage: http://www.tandfonline.com/loi/tjti20

Fabric Engineering by Means of an Artificial Neural

Network

A. Basu , K. P. Chellamani & P. Ramesh Kumar

To cite this article: A. Basu , K. P. Chellamani & P. Ramesh Kumar (2002) Fabric Engineering
by Means of an Artificial Neural Network, Journal of the Textile Institute, 93:3, 283-296, DOI:
10.1080/00405000208630570

To link to this article: https://doi.org/10.1080/00405000208630570

Published online: 30 Mar 2009.

Submit your article to this journal

Article views: 61

View related articles

Citing articles: 6 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tjti20
 Fabric Engineering by Means of an Artificial
Neural Network
A. Basu, K.P. Chellamani, and P. Ratnesh Kumar
The South India Textile Research Association, Coimbatore, India
Received 28.1.2002 Accepted for publication 24.9.2002

A study is reported of the relationship between the low-stress mechanical and surface
properties of polyester-fibre air-jet-spun yarns and the handle properties of fabrics made
from them. A total of 81 yam samples in three yarn linear densities were produced by
altering the process variahles in air-jet spinning. An artificial-neural-network (ANN) model
for predicting fabric-handle properties based on the low-stress mechanical and .surface
properties of air-jet-spun yarns is developed. The accuracy of prediction is good. With the
model, the direction of change in fabric-handle properties due to change in yarn flexural
rigidity, compre&sional energy, and hairiness is evaluated. An inverse model of ANN is
also developed.

1. INTRODUCTION
1.1 Background
Many general papers are available on the infiuence of yam properties and spinning
systems on the hand value of fabrics. The Numeri and Fukurami of fabrics are increased
considerably by spinning finer yams from the same fibres (Dbingra et al, 1983). Coarser
yams increase the cover factor, resulting in bigher stiffness of the fabrics. The bard twist
in the yam increases the yam-packing density and hence tbe fabric stiffness significantly.
The fibre mix in tbe blends also significantly influences the low-stress mechanical
properties of the woven fabrics. Fabrics produced from pure wool fibres give a higher
total hand value (THV) than wool-polyester-fibre blended winter fabrics of a similar
constmction. Yams produced on different spinning systems have different structures,
especially in fibre arrangement and twist distribution in the yam, owing to changes in
yam stmcture; the properties of yams vary significantly. Mule-spun yam has therefore
always been considered as being superior to ring-spun yam (Cassidy el al.. 1989). The
properties quoted as superior are the evenness of the yam and its handle. The infiuence
of ring- and rotor-spinning systems has been studied by Subramaniam and Amaravathi
(1994), who reported that fabrics woven from rotor-spun yam bave greater thickness than
fabrics woven from ring-spun yams. The value of compressional energy (We) is therefore
higher for fabrics woven from rotor-spun yams. It has been further reported that the
coefficient of friction of fabric (MIU) increases significantly in fabric woven from open-
end-spun yarns. Most of the primary band values, except the Fukurami, are higher for
fabrics woven from rotor-spun yams than for those woven from ring-spun yams. Fabrics
produced from ring-spun yams exhibit better handle than those produced from rotor-spun
yams. The authors further reported that the use of carded cotton with polyester fibre
enhances the handle of the fabrics.
Fabrics produced from ring-spun yam give lower bending rigidity and lower shear
rigidity than fabrics produced from rotor- and friction-spun yams (Behera et at., 1997).
Fabrics from ring-spun yam also show better compressional bebaviour than fabrics from

J. Text, Inst.. 2002. 93 Part 1. No. 3 © Te.xtile Institute 283

 Basii, Chellaniani, and Ramesh Kumar

rotor- and friction-spun yams and hence better hand. It has also been reported that fabric
from friction-spun yam shows the highest hysteresis loss, which indicates poor
dimensional stability of the fabric.
Two identically constructed cotton/polyester-fibre fabrics, one from a polyester
staple-fibre core covered with cotton yam and the other from a random-blended yam
showed a significant difference in low-stress mechanical and surface properties
(Radhakrishnaiah et al., 1993; Sawhney et al.. 1993). Differences in fabric properties
mostly reflected the differences in the physical properties of the yam. Fabrics made from
polyester-fibre-core/cotton-covered yam are more resilient to tensile and compressive
defonnation and have higher bending rigidity, lower tensile elongation, and lower shear
modulus. These fabrics also give higher values for all the four primai7 hand qualities
and higher totai band values associated with men's summer-suit applications. The same
fabrics also give higher values for five out of six primary hand qualities for women's
thin-dress applications. It is further reported that the same fabrics offered a cooler
contact sensation and much less variation in contact sensation along their length than
fabrics from random-blended yams. The fabric made of cotton-covered yam had a better
thermal-comfort value for cold and dry (winter) as well as hot and humid (summer)
weather conditions.
Radhakrishnaiah et al. (1993) reported that a core-sheath yarii showed lower values
for bending rigidity, bending hysteresis, compressive resilience, and tensile elongation.
The same yam also showed higher values for compressive softness and tensile modulus.
The lower tensile elongation and higher tensile modulus of core-sheath yam is reflected
in a lower elongation and higher modulus of a corresponding fabric. However, the
bending and compressional properties of core-sheath yam are inversely related to the
bending and compression properties of corresponding fabrics. Cotton/polyester-fibre
core-yarn fabrics have a cotton-like feel and appearance (Sawhney et al., 1993).
Core-wrap composite yam produced by the air-jet-spinning system is relatively weak
and extremely harsh in handle (Lord, 1987). Air-jet-friction-spun composite core yam is
less hairy and bulky than ring-spun yam. which is attributable to the basic difference in
their stmctures. Fabric made from this composite yam has a harsher handle than that of
fabric made from 100% ring-spun cotton yam.

1.2 Objectives of the Research

In spite of the excellent studies in the area of fabric hand as cited above, no integrated
study dealing with the elfect of low-stress mechanical and surface properties of air-
jet-spun yams on various handle properties of fabrics made from these yams has been
reported, and the present study is an attempt in tliis direction.
Smoothness, stiffness, and softness are the three main criteria taken into consideration
in assessing a fabric for hand (Ajayi, 1992). Any fabric that offers little frictional
resistance to motion across its surface is likely to be described as a smooth fabric. Any
fabric that bends easily is likely lo be described as flexible (not stiff), and sucb a fabric
will possess a low bending length. The higher the value of fabric compressional energy,
the softer will be the fabrics and vice versa (Alimaa et al., 2(H)0). Similarly, fabric shear
stiffiiess is also associated with fabric softness in the sense that higher values of shear
stiffness make the fabric less soft. The kinetic frictional resistance, bending length,
compressional energy, and fabric drape were therefore taken as some of the properties of
a fabric indicating its handle. The yam properties considered include (i) flexural rigidity,
(ii) compressional energy, and (iii) hairiness.

284 J. Text. Inst.. 2002, 93 Pan I. No. 3 © Textile Institute

Fabric Engineering by Means of an Artificial Neural Netw>ork

The study consists of three parts, which are:

(a) developing a model by using an artificial neural network (ANN) for
predicting fabric 'handle' properties such as (i) bending length, (ii) kinetic
frictional resistance, (iii) drape coefficient, and (iv) compressional energy
based on air-jet-spun-yam properties such as flexural rigidity, compressional
energy, and hairiness,
(b) evaluating the effect of the three yam properties on each of the four fabric-
handle properties under consideration, and
(c) developing an inverse model for predicting the air-jet-spun-yam properties
required to produce a fabric of a given quality.
Detailed algorithms of backpropagation networks and the relevant theories can be
found in many books that deal with artificial neural networks (Fausett, 1994; Freeman
and Skapura, 1992).

2. MATERULS AND METHODS

Yam of linear densities 19.68, 14.76. and 9.84 tex was spun on an MJS-802 air-jet-
spinning machine by using polyester fibres of 40-mm lengtb and a linear density of
0.111 tex. Some 27 yam samples in each yam linear density were produced by altering
the process variables in air-jet spinning. All the yam samples were tested for flexural
rigidity, compressional energy, and hairiness by using the test methods of Carlene (1950),
Pan et al. (1993), and the Zweigle G 565 instmment, respectively. The 81 fabric samples
were woven by using the 81 experimental air-jet-spun-yarn samples as the weft. The
fabric constructional particulars are given in Table I.
The fabric samples were tested for (i) bending length, (ii) kinetic frictional resistance,
(iii) drape coefficient, and (iv) compressional energy by using the methods of BS
3356:1961 (British Standards Institution, 1961), Ajayi (1992), BS 5058:1973 (British
Standards Institution, 1973), and Pan et al. (1993), respectively.

3. RESULTS AND DISCUSSION

3.1 ANN Model for Predicting Eabric Properties
A single hidden-layer feed-forward artificial neural network based on a backpropagation
algorithm was used to map the low-stress mechanical and surface properties of
air-jet-spun yams to fabric properties. The process data available consisted of 81 sets
of input-output pairs. Of these, 72 sets were used for training the net and the remainder

Table I
Fabric Constructional Particulars
Linear density of warp 9.84 tex X 2
Linear density of weft* Experimental air-jet-spun polyester-fibre yams
Ends/cm^ 19 or 27
Picks/cm^ 19 or 27
Weave Plain
*For 27 fabric samples, 19.68-tex air-jel-spun yam was used as weft. Similarly, 14.76-tex air-
jet-spun yam for 27 fabric samples and 9.84-tex air-jet-spun yam for another 27 fabric
samples were used as weft.
^ 9 ends/cm for fabrics made from 19.68- and 14,76-tex air-jet-spun yams; 27 ends/cm for
fabrics made with 9.84-tex air-jet-spun yams as weft.
*19 picks/cm for fabrics made from 19.68- and 14.76-tex air-Jet-spun yarns; 27 picks/cm for
fabrics made with 9.84-tex air-jet-spun yams as weft.

J. Text, tnst,. 2002, 93 Pan I, No, 3 © Textile Institute 285

 Basu, Chellamani. and Rame.sh Kuniar

Table II
Experimentally Determined and Predicted Values or Bending Length, Kinetic Frictional Resistance,
Drape Coefficient, and Compressional E n e i ^ of FahHcs
Bending Length (cm) Kinetic Frictional Drape Coefficient Compressional Energy
Resistance (cN) (X 10-cN cm/cm')

Experimental ftedicted Experimental Predicted Experimental Predicted Experimental Predicted

Vaiue Value Value Value Value Value Value Value

1 1.99 2.00 31.16 30.05 0.49 0.48 1.07 1.06

2 2.13 2.21 33.15 33.19 0.41 0.44 LOO 1.01
3 2.11 2.23 34.74 33.81 0.50 0.49 1.05 1.08
4 2.13 2.23 32.50 32.03 0.49 0.49 1.06 1.05
5 1.98 2.14 28.38 29.25 0.47 0.48 1.02 1.03
6 2.08 2.17 29.45 30.13 0.47 0.49 L05 1.05
7 2.09 2.17 33.78 33.67 0.48 0.49 1.01 l.OI
8 2.12 2.24 31.92 32.90 0.47 0.49 1.08 1.06
9 2.39 2.42 31.56 31.44 0.49 0.49 1.06 1.12

for testing. For testing, only input parameters were fed to the net. and the output values
from the net (predicted) were compared with the targeted output (determined
experimentally). For the nine testing data, the experimentally determined and predicted
values of the four fabric properties under consideration are given in Table II.
The correlation between the experimentally determined and predicted values of the
four fabric properties is shown in Figures I, 2. 3. and 4. The correlation coefficients of
fabric properties were bigb at (i) 0.92 for bending length, (ii) 0.93 for kinetic friclional
resistance, (iii) 0.91 for drape coefficient, and (iv) 0.90 for compressional energy. Hence
the proposed ANN model could be considered as suitable for the prediction of fabric
properties from yam-quality attributes.

2.49

2.39

2.29

2.19

2.09

0 2.05 2.15 2.25 2.35 2.45

Actual \^du£s (cm)

Fig. 1 Actual and predicted values of fabric bending length

286 /. Text. Inst.. 2002. 93 Part I. No. 3 © TexHIe Institute

Fabric Engineering by Means of an Artificial Neural Network

30

30 32 34 36
Actual Values (cN)

Fig. 2 Actual and predicted values of fabric kinetic frictional resistance

0.46

0.42 0.44 0.46 0.48 O.SO

Actual Values

Fig. 3 Actual and predicted values of fabric drape coefficient

3.2 Effect of Yarn Flexural Rigidity on Fabric-handle Properties

Fig. 5 shows the change in fabric properties due to changes in yam flexural rigidity as
determined by the neural-network model, assutning a constant yarn linear density of
14.76 tex, yam compressional energy of 2.30 X 10~^cN cm/cm^ and yam hairiness of
140 per 100 m.

J. Text. Inst., 2002. 93 Part I, No. 3 © Textile In.ttitute 287

 sii. Chellatnani, and Ramesh Kttmar

1.10
Actual Values(xiO
Fig. 4 Actual and predicted values of fabric compressional energy

With an increase in yam fiexural rigidity, fabric bending length also shows an increase,
which is understandable in view of the expected increase in fabric flexural rigidity
(Sharma et al.. 1996). An increase in yam flexural rigidity causes a reduction in frictional
resistance, which is rather steep in the flexural-rigidity range from 0.24 to 0,30 X
lO""" cN cm^/tex. In air-jet-spun yams, the clustering effect of core fibres due to their
parallel arrangement and winding by tight wrappers allows little freedom of movement to
fibres during bending, which makes these yams less compressible (Vohs et al., 1985).
With an increase in fiexural rigidity, the yam tends to become more stiflF, which will
further reduce compressibility, from which one would expect a reduction in the contact
area. This in tum could lower the friction factor (Chattopadhyay and Baneriee,1996).
An increase in the fabric-drape coefficient is also noticed with an increase in flexural
rigidity, which is attributed to the increase in bending length. Bending length is one of the
two major parameters influencing the drape coefficient (Cusick, 1965; Hu and Chan,
1998), the other being the shear stiffness. In a fabric specimen, simple shear is influenced
by (i) the force acting on the specimen, (ii) the width of the specimen, and (iii) the weight
of the specimen (Buckenham. 1997). Since these parameters were maintained at the
same level in the present case, the shear rigidity was not expected to be altered, and hence
the observed change in the drape coefficient of the fabrics is mainly attributed to the
corresponding changes in fabric bending length.
A reduction in yam flexural rigidity causes a decrease in fabric compressional energy.
A more flexible yarn can be expected to enhance the cover factor of woven fabric. Any
improvement in the cover factor of an already closely woven plain-weave fabric should
increase the inter-yam and inter-fibre frictional forces. This in tum should reduce the
fabric compressional energy.

3.3 Effect of Yarn Compressional Energy on Fabric-handle Properties

Fig. 6 shows the effect of the yam compressional energy on the fabric bending length,
kinetic frictional resistance, drape coefficient, and fabric compressional energy. With an

288 ./. Te.M. Insl., 2002, 93 Pan I. No. 3 © Textile Institute

Fabric Engineering by Means of an Artificial Neural Network

f., 2002, 9i Part I. No. 3 © Textile Institute 289

 Ba.su. Chellanutni, and Ramesh Kumar

I
I
Ul

(ma

290 /. Text. tnst,. 2002. 93 Part 1, No. 3 © Textile Institute

Fahric Engineering by Means of an Artificial Neural Network

increase in the yam compressional energy, both the bending length and kinetic frictional
resistance of fabrics show an increase; the drape coefficient and compressional energy of
fabrics show a decrease.
With an increase in compressional energy, yam diameter also tends to increase. Yams
with a greater diameter exhibit higher Hexural rigidity (Behera et ai, 1997), and that is
believed to be responsible for the observed increase in fabric bending length.
To ascertain the change in surface characteristics of air-jet-spun yams with an increase
in compressional energy, yams with varying levels of compressional energy (two yams of
14.76 tex and two yams of 9.84 tex) were analysed for structural parameters such as the
incidence of wrappers per unit length (I), the average number of wraps in a wrapped zone
(AN), and the average wrapped length of a wrapped zone (AL) by using a microscopical
method (Basu, 2000). These structural parameters for 14.76-tex and 9.84-tex air-jet-spun
yams are given in Table m.
It can be seen from Table III that, for yams with a higher compressional energy, the
occurrence of wrappings, tbe average number of wraps per unit length, and the average
length of wrapping are all higher for loose wrappers and the wraps per unit length and
length of wrapping are higher for tight wrappers. The extent of the difference in the
structural parameters between yams differing in compressional energy is shown in
Table IV.
Owing to the perceptible increase in the wrapper-related parameters for yams with
higher compressional energy, these yams could be expected to have a rough surface, and
hence fabrics made from them could be expected to possess higher kinetic frictional
resistance.
An increase in fabric cover factor and the attendant increase in inter-yam friction when
yams of higher compressional energy are used are believed to be responsible for the
reduction in compressional energy of fabrics made from these yams.

Table III
Structural Parameters of Air-Jet-spun Yams
Yam Compressional Tight Wrappers Loose Wrappers
Linear Energy
Density (X 10 cN cm/cm"^) Incidence Average Average Incidence of Average Average
(tex) of Wraps/10 cm Number of Length of Wraps/10 cm Number of Length of
<I) Wraps/cm Wrap.s/cm (1) Wraps/cm Wraps/cm
(AN) (AL) (AN) (AL)
14.76 2.186 32.0 32.69 0.197 18.80 32.05 0.138
2.515 31.2 33.40 0.214 23.20 35.91 0.150
9.84 2.141 31.4 28.07 0.247 14.40 27.73 0.140
2.467 30.4 32.00 0.260 19.60 40.32 0.160

Tahle IV
Direction and Extent of the Change in Wrapper Parameters for Air-jet-spun Yarns of Higher
Compressional Energy
Yam Tight Wrappers Loose Wrappers
Linear
Density Incidence of Average Average Incidence of Average Average
(tex) Wraps/10 cm Number of Length of Wraps/10 cm Number of Length of
(I) Wraps/cm Wraps/cm (I) Wraps/cm Wraps/cm
(AN) (AL) (AN) (AL)
14.76 -2.5% +2.1% +7.9% +18.9% + 10.7% +8.0%
9.84 -3.2% +12.2% +5.0% +26.5% +31.2% + 12.5%

/ Text. Inst., 2002, 93 Part I, No. 3 © Textile Institute 291

 Basu, Chellamani. and Ramesh Kumar

The fabric-drape coefficient is influenced by the bending length and shear angle; the
expression connecting them due to Cusick (1965) is as follows;

DC - 9.40C - 2.51A + 53.5

where DC is the drape coefficient {%),

C is the bending length in cm, and
A is the shear angle in degrees at a shear stress of 2 gf cm/cm^.
With an increase in fabric bending length when yams of higher compressional energy
are used, one would expect the fabric-drape coefficient to go up. However, an exactly
opposite trend is noticed. This could probably be due to a reduction in shear rigidity and
the consequent increase in shear angle. This aspect, i.e. the effect of yam compressional
energy on fabric shear rigidity, needs further exploratory studies.
Yam compressiona! energy therefore shows an inverse relationship with fabric
compressional energy. This is in line with the observations of Alimaa et al. (2000) on
plain-knitted fabrics and by Radhakrishnaiah and Sawhney (1996) on core-spun yam.
Yams with higher compressional energy generally have greater diameter.

3.4 Effect of Yarn Hairiness on Fabric-handle Properties

Fig. 7 shows the effect of yam hairiness on fabric bending length, kinetic frictional
resistance, drape coefficient, and compressional energy.
Higher values of yam hairiness show a tendency to increase the bending length and
kinetic friction^ resistance of fabrics and to decrease the compressional energy and drape
coefficient of fabrics.
With an increase in yarn hairiness, fibre-to-fibre interlacements between adjacent
threads in the fabric will be more, which should make the fabric more difficult to bend
(Radhakrishnaiah and Sawhney, 1996) and also increase the kinetic frictional resistance
during yam-over-yam or fabric-over-fabric movement.
The greater extent of fibre-to-fibre interlacements between adjacent threads in the
fabric when yams of higher hairiness are used also makes the fabric more difficult to
compress, which in tum reduces tbe fabric compressional energy. With an increase in
yam hairiness, one would also expect a decrease in effective yam diameter (Punj et ai.
1996), which in tum would increase the shear angle (Buckenham, 1997). The relation
between yam diameter and shear angle, 9, is given by:

where F is the force acting on the specimen (gf);

R is the shear experienced by the specimen (gf/cni);
d is the width of the specimen (or diameter of yam) (cm); and
IV is the tensioning weight applied on the specimen.
The combined infiuence of the two opposing factors, i.e. an increase in bending length
and increase in shear angle, resulted in the observed reduction of drape coefficient with
an increase in yam hairiness.
The ANN model discussed in Section 3.1 is capable of predicting fabric properties
(output) for a given set of yam properties (input). If the properties of yam required to
manufacture a fabric of a given quality and the process variables to be adopted in the
spinning line to produce yams of given quality could be predicted by using ANN. then

292 / Text. Inst.. 2002. 93 Part I. No. 3 © Textile Institute

Fabric Engineering by Means of an Anificial Neural Network

(103) ipftrai Sinprog

/.. 2002. 93 Pan I, No. 3 © Texiile 293

 Basti. Chellamani, and Ramesh Kumar

the model would be of use in the area of fabric-engineering. Hence an attempt in this
direction was made, and the results will be discussed in Section 3.5.

3.5 Training of ANN for Prediction of Yarn Properties

The 81 pairs of yam properties-fabric-quality data as presented in Section 3.1 were used
for the purpose. Fabric quality in terms of bending length, kinetic frictional resistance,
drape coefficient, and compressional energy was used as input parameters, and the
corresponding yam-quality attributes (flexural rigidity, compressional energy, and
hairiness) were used as output parameters.
After training, an attempt was made to predict the air-jet-spun-yam properties required
to produce air-jet-spun-yam fabrics that had the minimum difference in properties from
ring-spun-yam fabrics.
In this connection, air-jet-spun yams were produced by using process parameters that
were found to help to produce air-jet-spun yams with lower flexural rigidity. Fabrics were
made from these yams, and ring-spun-yam fabrics were also produced for comparison
purposes. Both the sets of fabrics were tested for bending length, kinetic friction
resistance, drape coefficient, and compressional energy. The properties of air-jet- and
ring-spun-yam fabrics are given in Table V.
Air-jet-spun-yam fabrics are inferior to their ring-spun counterparts by about 12% in
bending length and about 25% in drape coefficient.
The inverse model of ANN was made use of to narrow down this difference.
Two predictions were made, one for a fabric bending length of 1.75 cm (with kinetic
frictional resistance at 30.04 cN, drape coefficient at 0.47, and compressional energy at
1.038 X 10"^ cN cm/cm^ and the second for a drape coefficient of 0.43 (with bending
length at 1.85 cm, kinetic frictional resistance at 30.04 cN, and compressional energy at
1.038 X 10"^ cN cm/cm^). Predicted values of the yam properties for the above two
situations are given in Table VT.
By using the predicted values of yam properties as input variables, process parameters
in air-jet spinning were predicted and are given in Table VII.
Air-jet-.spun yams were spun by using the predicted values of process parameters and
fabrics woven from them. These fabrics were tested for bending length and drape
coefficient. Targeted and experimentally determined values of fabric properties are given
in Table Vin.
The difference between the targeted and experimentally determined values of the two
fabric properties was within the statistical limits. The inverse model of ANN is therefore
found to be of use towards narrowing down the gap in bending length and drape
coefficient between air-jet-spun and ring-spun-yam fabrics.

Table V
Handle Properties of Fabrics Made from Air-jet-spun and Ring-spun Yarns
Bending Kinetic Drape Compressional
Property Length Frictional Coefficient Energy
Type of \ ^ (cm) Resistance (cN) (x lO^'cN cm/cm^)
Yam
Air-jet-spun* 1,85 30.04 0.470 1.038
Ring-spun 1.65 32.64 0.380 1.039
DifiFerence Air-jet-spun Air-jet-spun Air-jet-spun Air-jet-spun
higher by 12% lower by &% higher by 24% lower by 0.1%
•Fabrics made frotn air-jet-spun yams with process parameters for producing yam of lower flexural rigidity.

J. Text. Inst., 2002. 93 Pan 1. No. 3 © Textile Institute

Fabric Engineering by Means of an Artificial Neural Network

Table VI
Fabric Quality (Input Variables) and Predicted Values of Yarn Properties
Prediction Input Variables ft-edicted Values (Yam Properties)
No. (Fabric Properties)

Bending Drape Flexural Compressional Hairs/100 m

Length Coefficient Rigidity Energy
(cm) (X 10 ^ cN cm^/tex) (X 10"' cN cm/cm^)
1.75 0.205 2.289 107
0.430 0.209 2.412 103

Table Vn
Yarn Quality (Input Variables) and Predicted Values of Process Parameters
Prediction Input Variables (Yam Properties) Predicted Values (Process Parameters)
No.
Rexural Compressional Hairs/ Delivery Main First- Second- N| to Feed
Rigidity Energy 100 m Speed Draft nozzle nozzle F.RolI Ratio
{X 10"' cN cm"/tex) (X 10'-^ cN cm/cm") (m/min) Ratio Pressure Pressure Distance
(kgf/cm^) (kgf/cm^) (mm)
I 0.205 2.289 107 187 28.30 1.96 3.44 40.00 0.98
2 0,209 2.412 103 206 28.16 2.10 3.43 39.87 0.98

Table VIII
Targeted and Experimentally Determined 'Values of Fabric Bending 1Length and Drape Coefficient
Prediction Targeted Values Experimentally Standard Control Predicted
No. Determined Values Error of Limits* Value Lies
Experiments,1 between
Bending Drape Bending Drape Value Lower Upper Control
Length Coefficient Limits?
Length Coefficient
(cm) (cm)
1 1.75 1.82 0.027 1.730 L910 Yes
2 0.430 0.440 0.006 0.420 0.460 Yes

4. CONCLUSIONS
4.1 A model for predicting fabric-handle properties such as bending length, kinetic
frictional resistance, drape coefficient, and compressional energy based on air-jet-
spun-yam properties such as flexural rigidity, compressional energy, and hairiness
was developed. The accuracy of prediction is good.
AJl By using the ANN model, ihe direction of change in fabric-handle properties due to
changes in yam flexural rigidity, compressional energy, and hairiness was evaluated.
With an increase in yam flexural rigidity, the fabric bending length, drape coefflcient,
and fabric compressional energy tend to increase, whereas the fabric frictional
resistance reduces. With an increase in the yam compressional energy, the fabric
bending length and frictional resistance increase, whereas fabric drape and fabric
compressional energy show a tendency to decrease. The trend in fabric properties
observed with a change in yam hairiness is similar to that in yam compressional energy.
4.3 An inverse model of ANN for predicting process variables in air-jet spinning with a
view to producing yams of given quality attributes and subsequently to producing
fabrics of specified quality characteristics, particularly in terms of the handle
properties considered in the study, was also developed.

/. Text. Inst,, 2002, 93 Pan I, No. 3 © Textile Institute 295

 Basu. Chellamani, and Ramesh Kumar

ACKNOWLEDGEMENTS
The authors are greatful to Ms Indra Doraiswamy. Director, SITRA, for ber guidance at
various stages of the progress of tbe work. Tbe assistance rendered by Mr M.ICViitopa of
tbe Spinning Division, SITRA, is also gratefully acknowledged.

REFERENCES
Ajayi, O., 1992. Fabric Smoothness. Friction and Handle. Te.xt. Re.s. J,. 62. 52-59.
Alimaa. D.. Matsuo, T.. Nakajima. M., Takahashi. M., and Yamada. E.Y.. 2000. The Compressionai Property of
Knitted Fabrics in Terms of Fabric Structure and Yam Compressional Property. In Proceedings of the 80th
World Conference of the Textile ln.ititure, the Textile Institute. Manchester.
Basu. A.. 2000. Air-jei Spun Yams: Structure-Properties Relationship. Asian Text. J.. 9. No. 10, 63-66.
Behera. B.K.. Ishtiaque. S.M.. and Chand, S.. 1997. Comfort Properties of Fabrics Woven from Ring-. Rotor-.
and Friction-spun yams. / Text. Inst.. 88, Part I. 255-264.
British Standards Institution. I96t. Method of Tesi for the Detemiination of Stiffness of Cloth. BS 3356:1961.
British Standards Institution. 1973. Method for the Assessment of Drape of Fabrics. BS 5058; 1973.
Buckenham, P.. 1997. Bias-extension Measurements on Woven Fabrics. J. Text. Inst.. 88. Pan I. 33-40.
Carlene, P.W., 1950- The Relation between Fibre and Yam Flexural Rigidity in Continuous-filament Viscose
Yams. J. Text. Inst.. 41. TI59-TI7I.
Cassidy. T.. Weedall. P.J.. and Harwood. R.J.. 1989. An Evaluation of the Efifeet of Different Yam-spinning
Systems on the Handle of Fabrics. Part I: A Comparison of Ring-spun and Mule^pun Woollen Yams J Text
M.vr.. 80,537-545.
Chattopadhyay. R.. and Banerjee. S.. 1996. The Frictional Behaviour of Ring-. Rotor-, and Friction-spun Yam
J. Text. Inst., 87. Part 1. 59-67.
Cusick. G.E.. 1965. The Dependence of Fabric Drape on Bending and Shear Stiffness. / Text. Inst.. 56
T596-T606.
Dhingra, R.C., Mahar. T.J., Postle. R.. Gupta. V.B.. Kawabata. S.. Niwa, M.. and Camaby. G.A.. 1983. The
Objective Specification of the Handle of Men's Suiting Materials: A Comparison of Fabric Handle Assessment
in India, Australia. Japan, and New Zealand. IndUtn J. Text. Re.s.. 8. 9-15.
Fau.sett. L.. 1994. Fundamentals of Neural Networks: Architectures. Algorithms and Applications, ^entice Hall
Englewood Cliffs. NJ. USA.
Freeman. J.A.. and Skapura. D.M., 1992. Neural Networks: Algorithms, Applications and Programming
Techniques. Addison-Wesley. New York. NY. USA.
Hu. J.. and Chan. Y.F.. 1998. Effect of Fabric Mechanical Properties on Drape. Text. Res. / . 68. 57-64.
Lord. P., 1987. Air-jet and Friction Spinning. Text. Horii., 7. No. 10. 20-24.
Pan. N.. Zeronian, S.H.. and Ryu. H.S.. 1993. An Altemative Approach to the Objective Measurement of
Fabrics. Text. Res. J.. 63, 3 3 ^ 3 .
Putij, S.K., Mukhopadhyay. A., and Saha. R.N.. 1996. Air-jet Spinning of Acrylic. Text. Asia. 27. No. 5, 53-56.
Radhakrishnaiah. P.. and Sawhney. A.P.S.. 1996. Low Stress Mechanical Behavior of Conon/Poiyester Yams
and Fabrics in Relation lo Fiber Distribution within ihe Yam. Text. Re.f. J.. 66. 99-103.
Radhakrishnaiah. P.. Tejatanaleri. S.. and Sawhney. A.P.S.. 199.V Handle and Comfort Properties of Woven
Fabrics Made from Random Blend and Cotton-covered Cotton/Polyester Yams. Text. Res. ./.. 63. 573-579.
Sawhney. A.P.S.. Kimmel. L.B.. Ruppenicker. G.F.. and Thibtxieaux. DR.. 1993. A Unique Polyester Staple-
core/Cotton-wrap Yam Made on a Tandem Spinning System. TcKt. Res. J.. 63. 764-769.
Shamia, I.C.. Chatterjee. K.N.. and Mukhopadhyay. A.. 1996. Air-jet and Ring-spun Yams—Comparative
Study of Fabrics. Indian Text. J.. 107. No. I. 54-55.
Subramaniam. V., and Amaravathi. T.B.C.. 1994. Effects of Fibre Linear Density and the Type of Cotton on the
Handle and Appearance of Polyester-fibre-Cotton Fabrics Produced from Ring-spun and Open-end-spun Yams
J. Text. ln.it.. 85. 24-28.
Vohs. KM.. Barker. R.L.. and Mohamed. M.H., 1985. CM)jective Evaluation of Fabric Woven wife Air-jet
Yams. Part I; Mechanical and Surface Properties. In Proceedings of 3rd Japan/Australia Symposium on
Objective Measurement: Application to Product De.ugn and Process Control (edited by S. Kawabata, R, Postle,
and M. Niwa), Textile Machinery Society of lapan. Osaka. Japan, pp. 121-134.
Zweigle G 565 Hairiness Tester: Instrument Manual.

./. Text. Inst.. 2002. 93 Part I. No. 3 © Teml,- Institute

View publication stats