M.
Cengiz Kayacan*,
Mehmet Dayik,
Oguz Colak,
Murat Kodaloglu
University of Suleyman Demirel
Department of Textile Engineering
Isparta, Turkey
*Corresponding author
S.D.U. Textile Engineering Department, Isparta, Turkey
E-mail: ckayacan@mmf.sdu.edu.tr
Phone: +90. 246 211 12 46
n Introduction
Weaving machines are classified into
four groups according to their weft inser-
tion systems; shuttle, projectile, rapier
and jet (i.e. air and water jet) looms. Of
these groups, the shuttle and projectile
weft insertion systems have reached the
term of their economic life, because of
their low weaving velocity. The water
jet weft insertion system does not have a
wide application in practice, as it is only
suitable for yarns made of hydrophobic
fibres. Rapier and air jet weft insertion
systems are commonly used for almost
all kinds of fibres and yarns. When air
jet and rapier weft insertion systems are
compared, it is apparent that the rapier
system has a lower velocity than air jet
systems. The air jet weaving system is
commonly preferred due to its high pro-
duction speed.
In air jet weaving, the weft yarn is moved
by the friction created by the high speed
air flow. The forces which are required
to move and accelerate the weft yarn
are produced by the air jet. These forces
have to be higher than the combination of
the force of the inertia and the resistance
forces of the yarn bobbin and the reserve
system. The carrier and the resistive force
characteristics are defined by the consider-
ation of weft yarn properties and physical
properties of the air flow [1,2,9,15]. Air
velocity and yarn structures have complex
interrelations due to their properties, such
as turbulent and laminar air flow, constant
yarn diameter, linear density (count)
and elasticity. Air flow is important to
increase weft velocity and productivity,
and so it must always be controlled, due
to the change in the yarn count and the
coefficient of twist. When the coefficient
of twist increases, the weft velocity de-
FIBRES & TEXTILES in Eastern Europe July / October 2004, Vol. 12, No. 3 (47)
Velocity Control of Weft Insertion
on Air Jet Looms by Fuzzy Logic
Abstract
In this study, a weft insertion system in air jet weaving looms was controlled by using fuzzy
logic. The effects of yarn properties on weft yarn velocity along the tube have been investi-
gated. The effects of yarn linear density (tex) and value of twist coefficient on the change
in weft yarn velocity are determined by the fuzzy logic system in air jet weaving looms.
Experimental data and expert knowledge has been used in the establishment of the fuzzy
logic model and the construction of basic principles. The results obtained from the fuzzy
logic model are compared with experimental results.
Key words: fuzzy logic, air jet, yarn count, linear density, coefficient of twist.
creases. Also, when the yarn count (tex) many speed values between 40-70 km/h.
increases, the average velocity increases In other words, all speed values between
[3,4]. Weft yarn velocity for the Toyoda- 40-70 km/h are accepted as true (1). The
type weaving machine has been studied fuzzy logic approach identifies all speed
by Hasegawa et al. [12]. This velocity values between 40-70 km/h, as shown in
achieves a certain average value along the Figure 1b, and 40-70 km/h are assumed as
weft insertion. It reaches constant speed the lower and upper limits of the average
at a certain time tp which is such a func- speed (logically assumed 0). The average
tion of the weft diameter, that due to the speed of the particle has been accepted as
diameter increase tp also increases [16]. 55 km/h. So, the fuzzy logic solution ap-
According to Uno, et al. [19] the insertion proach is the most suitable technique for
force at a constant diameter weft increases this kind of problem [11,14].
due to the increase in air velocity. With the
decrease in weft diameter, the insertion
force also decreases. For this reason, it a) Weight
was observed that the velocity of the weft
Average speed
yarn has a smaller value. 1
In this study, the effects of the coefficients
of twist and yarn count on the weft yarn
velocity along the tube at constant air-pres-
0
40 70 Speed, km/h
sure are investigated using fuzzy logic.
b) Weight
n Fuzzy Logic Average speed
1.0
The fuzzy logic concept was first proposed
in 1965 by L.A. Zadeh [17]. Since then, 0.5
the importance of fuzzy logic has continu-
ally increased, especially in engineering 0
studies. Fuzzy logic can be defined as a 40 70 Speed, km/h
mathematical model to study and define
uncertainties [20,21]. Statistics and prob- Figure 1. Classic (a) and fuzzy logic (b) speed
ability problem solutions with uncertainty graph.
instead of certainty are well known. Nev-
ertheless, our daily life requires many
uncertainties. Therefore, in our studies
we must use the tool of uncertainty to
understand the capabilities of conclusions
related to themselves [17,21].
Logically, the variables are defined as
false or true. But fuzzy logic states that the
variable could have intermediate values.
For example, the average speed of a parti-
cle can be accepted between 40-70 km/h,
as shown in Figure 1a. In fact, there are Figure 2. Relay nozzles.
29
Figure 3. Weft velocity characteristic along the tube length.
Uncertainty conditions in fuzzy logic are
determined by pre-defined membership
functions. If the most approximate ele-
ment of the results is defined as 1, it can
be understood that the other elements are
between 0 and 1 and they are changing
continuously. The value of every element
with the variation between 0 and 1 are
called membership degrees, and the vari-
ation in a subset is called membership
function [17]. After that, the definition of
the fuzzy logic rule base is implemented
by using either Suggeno or Mamdani’s
method.
After a solution of the implemented
fuzzy logic system has been obtained
by using the selected method, the solu-
tion is defuzzified. The defuzzification
Figure 4. Air-jet weft insertion system.
Figure 5. Air jet system of the weaving machine.
30
Figure 6. Fuzzy logic model for weft yarn insertion system.
method is called as the inverse operation
of fuzzification. There are seven follow-
ing defuzzification methods: Centroid,
Bisector, Moment, LOM (Largest of
maximum), SOM (Smallest of maxi-
mum), MOM (Mean of maximum) and
COG (Centre of gravity) [17,21].
n Air Jet Weaving Systems
The working principles of air jet weaving
machines are based on carrying the yarn
by the friction of the air jet. During the
yarn’s forward movement by the air in
the sheds, the velocity of the weft yarn
decreases because of the decreasing pres-
sure and the disturbance on the air flow
direction [3-5]. The air jet must provide a
constant speed to the weft yarn along the
FIBRES & TEXTILES in Eastern Europe July / October 2004, Vol. 12, No. 3 (47)
weaving wideness. However, weft yarn of
a certain mass is carried by a single jet at
a limited distance. That is why the relay
nozzles are installed at certain distances,
in order to prevent a decrease in yarn
velocity. The relay nozzle system and the
general characteristics of the weft yarn
speed in the shed are shown in Figures 2
and 3 respectively. These are implement-
ed on a movable hollow-needle or slay
system. The basic function of the main jet
is to load the weft yarn into the shed, and
then to carry it to the first relay nozzle at
a certain speed. The weft yarn suddenly
reaches high velocity by means of the
main jet. To preserve weaving defects
and asynchronous beat-up movement, the
weft yarn position and the instant velocity
of the weft yarn must be fully controlled
during the weaving process. A schematic
view of the air jet and its control system
is shown in Figures 4 and 5.
n The Effect of Yarn Properties
on the Weft Yarn Velocity
The effects of twist coefficient and yarn
count, which are discussed below, are the
main effective parameters of the weft
yarn velocity along the tube.
The effects of coefficient of twist on
weft yarn velocity
Higher twist values of the yarn provide
a more compact structure and prevent
mutual movements of fibres. Yarn hairi-
ness decreases with the twist increments.
Lower yarn hairiness cause less friction
between the yarn and the air. It is known
that yarn twist is directly related to hairi-
ness [6,7,10]. For this reason, friction be-
tween the fibres and air decreases when
the yarn has a smooth surface. High
twisted yarn has lower velocity than low
twisted yarn. Experiments carried out on
Open End (OE) rotor yarns showed that
high twist ratio increases the weft inser-
tion time as a result of the decrement
of yarn diameter and the smooth yarn
surface [8].
Figure 7. Membership functions of yarn count.
Figure 9. Membership functions of yarn tube length.
The effect of yarn count (linear den-
sity) on weft velocity
Different yarn counts have different lin-
ear densities, and so the velocity of a thin
weft yarn is much higher than the veloc-
ity of the thick weft yarn, which also has
low starting velocity as a result of the in-
verse relationship between velocity and
mass. During weft insertion, the velocity
of thick weft yarns reaches the maximum
at the end of the shed line, because of its
inertia. Since the inertia of thin yarns is
lower than that of thick yarns, the veloc-
ity at the end of shed of thin yarn is lower
than that of the thick yarn. However, the
effect of inertia is not as important as the
effect of mass. Therefore, thin yarns have
higher average velocity all along the
length of the weft insertion line [8,18].
n Determination of Weft Yarn
Velocity by Fuzzy Logic
The aim of this work was to check
whether the fuzzy logic system can be
used for predicting the weft yarn inser-
tion velocity. To achieve this aim, it was
necessary to compare the results obtained
by fuzzy logic with those from real meas-
urements.
The yarn count and the coefficient of
twist were selected as input variables,
whereas yarn velocity was the output
variable. The fuzzy model of this prob-
FIBRES & TEXTILES in Eastern Europe July / October 2004, Vol. 12, No. 3 (47)
Figure 8. Membership functions of yarn twist.
Figure 10. Membership functions of yarn speed.
lem is given in Figure 6. Real values of
weft velocity for comparison with those
obtained by fuzzy logic have been taken
from the experimental study results of
Adanur [1]. These were carried out on
the basis of the following parameters:
§ jet type - NCSU No 1⁄2
§ weft insertion pressure - 0-689 kPa
§ weft insertion cycle - 500 weft/min
§ weft feed type - drum, loop
§ mean humidity - 45-80%
§ temperature - 15-30°C
The yarn types used in this study [1]
were open-end yarns with yarn counts
between 0-200 tex. The Fuzzy Toolbox in
MATLAB 6.1 was used for mathematical
calculations. The numbers of the input
membership functions and base widths
were obtained using the experimental
results and expert knowledge. These are
shown in Figures 7, 8 and 9. Because we
investigated all output variable specifica-
Figure 11. Some of
the defined rules.
tions that are called yarn velocity, output
membership function is divided into six-
teen intervals, as shown in Figure 10.
n Construction of the Fuzzy
Logic Basic Rules
It is necessary to determine fuzzy logic
rules to make obvious the effect of rela-
tionships between the input membership
functions and the result. Expert knowl-
edge has been used in the formation of
the basic rules to define enough of them
to reach a correct solution. Elements of
these rules are shown in Figure 11.
The calculated 3-D output-input depend-
ency results used are presented in Figures
12, 13 and 14. Figure 12 shows the rela-
tions between weft yarn velocity and the
coefficient of twist & yarn count. Tube
length and yarn count, which influence
the speed of yarn relations, are plotted
31
 Weft velocity, m/s
Tube length, mm

Figure 12. 3-D relationship between coefficient of twist yarn count Figure 15. Velocity of weft yarn twisted 1915/m with different count
and velocity of yarn. (20, 100, 200 tex) determined by fuzzy logic and by experiment.

Weft velocity, m/s

Weft velocity, m/s

Yarn
count, tex Tube length, mm
Distance along tube, mm

Figure 13. 3-D relationship between yarn count, tube length and Figure 16. Velocity of weft yarn twisted 2892/m with different count
velocity of yarn. (20, 100, 200 tex) determined by fuzzy logic and by experiment.
Weft velocity, m/s
Weft velocity, m/s

Coeficient
of twist Tube length, mm
Distance along tube, mm

Figure 14. 3-D relationship between coefficient of twist, tube length Figure 17. Velocity of weft yarn twisted 3830/m with different count
and velocity of yarn. (20, 100, 200 tex) determined by fuzzy logic and by experiment.

in Figure 13. The relationships between
tube length, coefficient of twist and weft
yarn velocity are given in Figure 14.
The values calculated by using the fuzzy
logic model and the experimental results
for different twist value are plotted in
Figures 15, 16 and 17. As tex count de-
creases, the yarn velocity increases. In
contrast, when the twist value increases,
the yarn velocity decreases.
When the graphics are inspected care-
fully, it is seen that while yarn count
increases, despite high changes in yarn
velocity values at the tube entrance, there
are similar velocity values at the end
of the tube. This situation is especially
visible for lower yarn twist values. The
reason for this is that the mixed structure
of fibres with low yarn counts needs
more initial force because of higher fric-
tion effects at the beginning. But by the
timethe yarn is moving by the air effect,
the fibres become parallel to each other,
a more compact yarn structure is formed,
and the yarn velocities become similar

32 FIBRES & TEXTILES in Eastern Europe July / October 2004, Vol. 12, No. 3 (47)
to each other. The comparison of result
obtained by using the fuzzy logic model
and from experimental studies are given
in Figures 15, 16 and 17.
n Conclusions
In this study, the effect of yarn param-
eters on the yarn velocity along the weft
insertion tube was investigated. The
effect of the twist coefficient of weft
yarn on weft yarn velocity at different
yarn linear densities was determined by
using fuzzy logic. When the calculated
and experimental results of the weft yarn
insertion are compared, the following
conclusions can be drawn.
Since a high twist coefficient makes
the yarn more compact and smoother, it
reduces the yarn velocity and increases
the insertion time. Increase in yarn count
increases the velocity of the weft yarn
led through the tube. The yarn speed can
be easily determined in dependence on
twist coefficient and yarn count by fuzzy
logic. It is practically possible to obtain
approaches that will give positive results
with yarn velocity optimisation in a more
economical way, instead of the method of
experimental study.
Acknowledgement
This study is supported by the Prime Ministry
State Planning Organisation (TR), Project No:
DPT 2000 K 120430 and SDU Af 238.
References
1. Adanur, S. and Mohamed, M. H., 1992,
Analysis of Yarn Motion in Single-nozzle
Air-Jet Filling Insertion, Part II: Experimen-
tal Validation of the Theoretical Models
and Statistical Analysis, Journal of the
Textile Institute, Vol. 83, No. 1 pp. 56-68.
2. Adanur, S., Walker, R. P., Broughton,R.
M. and Beale, D. 1994, Weaving Techno-
logy - What Next?, Melliand Textilberich-
te, English/German, Vol. 75, No. 4, April,
pp. 255-263.
3. Adanur, S., and Mohamed, M. H., 2001,
Weft Insertion on Air-jet Looms: Velocity
Measurement and Influence of Yarn
Structure Part II Effects of System Para-
meters and Yarn Structure, Journal Textile
Inst. No 2, 1988 systems, 121 (73-93).
4. Adanur, S. and Mohamed, M. H., 1992,
Analysis of Yarn Motion in Single-nozzle
Air-Jet Filling Insertion, Part I: Theoretical
Models for Yarn Motion, Journal of the Te-
xtile Institute, Vol. 83, No 1, pp. 54-55.
5. Adanur, S., 1989, Dynamic Analysis of
Single Nozzle Air-jet Filling Insertion,
Ph.D. Thesis, NCSU.
FIBRES & TEXTILES in Eastern Europe July / October 2004, Vol. 12, No. 3 (47)
6. Adanur, S., 1995, Effects of Fabric Struc-
tural Parameters on Fabric Modulus, Mel-
liand Textilberichte, Vol. 76, No. 6, June
(German/English), pp. 396-399.
7. Adanur, S., and Bahtiyarov, S., 1996,
Analysis of Air Flow in Single Nozzle Air-
Jet Filling Insertion: Corrugated Channel
Model, Textile Research Journal, 66(6),
pp. 401-406.
8. Adanur,S. and Mohamed, M. H., 1991,
Analysis of Yarn Tension in Air-Jet Filling
Insertion, Textile Research Journal,
Vol.61, No. 5, pp. 253-258.
9. Bayhan, M., Kodaloglu, M., Cengiz, Y.
and Kaplan, S., 2002, Drum ve Loop
Sistemlerinde Atki Hareketinin Dinamik
Modellenmesi ve Hizin Bulanik Mantikla
Kontrolu, Tekstil Maraton Dergisi, Mart-
Nisan. pp. 63-69.
10. Batra, S. K., Ghosh, T. K. and Zeidman,
M., 1987, Dynamic Analysis of Yarn in
Ring Spinning: An Integrated Approach,
NCSU, March, pp. 256-262.
11. Branco, P. J., and Dente, J. A., 2001, “Fuz-
zy Systems Modelling in Practice”, Fuzzy
Sets and System Vol. 121, pp: 73-93.
12. Hasegawa J., et al., 1981, A Study of Weft
Insertion System on Air Jet Loom, Paper
presented at ASME Textile Engineering
conference, Raleigh, NC, pp. 358-45.
13. Kayacan, M. C., Dayık, M., 2001, Bulanık
Mantık Yardımıyla Bornoz Beden Ölçü-
lerinin Tespiti, Tekstil Maraton. Temmuz-
Ağustos, pp. 65-72.
14. Kissling U., 1985, Experimental and
Theoretical Analysis of Weft Insertion by
Air-jet, Melliand Textilberichte. February,
pp. 145-152.
15. Salama, M. M., 1984, Mechanics of Air
Jet Filling Insertion, PhD Thesis, UMI.
16. Salama, M., Adanur, S., and Mohamed,
M.H., 1987, Mechanics of a Single Noz-
zle Air-Jet Filling Insertion System, Part
III: Yarn Insertion through Tubes, Textile
Research Journal, Vol. 57, No.1, January,
pp. 44-54.
17. Şen, Z. 1999, Mühendislikte Bulanık
(Fuzzy) Modelleme İlkeleri, Ders Notları,
İstanbul.
18. Vangheluwe, L., Puissant, P., Kiekens, P.,
1997, Combination of Fuzzy Feedback Lo-
ops and Feedforward Adaptive Systems
for Weft Insertion Control in Air-Jet We-
aving; Proceedings of the 3rd International
Conference on Design to Manufacture in
Modern Industry, University of Maribor, Fa-
culty of Mechanical Engineering, suppor-
ted by the Slovenian Ministry of Science
and Technology, Portoroz, Slovenia, 8-9
September, pp. 471-478.
19. Uno, M., et al., 1972, A Study on an Air
Jet Loom with Substreams Added. Part
5: Analogous Experiment of Flow, J. Te-
xtile Machinery Soc., Japan, Vol. 18, pp.
5-16.
20. Zadeh, L. A, 1965, Fuzzy Sets, Informa-
tion and Control Vol. 8, pp. 338-353.
21. Zadeh, L. A., l978, Fuzzy Sets As A Basis
For A Theory Of Possibility, pp 3-10.
22.http://www.sulzertextil.com/solutions/
weaving/i_airjet.htm
Received 10.03.2003 Reviewed 12.08.2003
33