Textile Research Journal Article

Numerical Model of an Air-jet Loom Main Nozzle for Drag

Forces Evaluation
G. Belforte1, G. Mattiazzo, V. Viktorov
Abstract The flow field inside an air-jet loom and C. Visconte
main nozzle is studied numerically, by means of a Department of Mechanics, Politecnico di Torino, C.so
two-dimensional model implemented in the com- Duca degli Abruzzi, 24, 10129 Torino, Italy
mercial computational fluid dynamics (CFD) code
Fluent. In order to determine which physical
model could allow a better prediction of the noz-
zle behavior, preliminary simulations were carried
out on a basic geometry configuration, changing
flow models and comparing results with available
experimental data. Having done the model setup,
simulations aimed at evaluating drag force on the
weft yarn were performed on various geometry
configurations; in particular, the influence of accel-
eration tube length, shape and size on drag force
was evaluated. Results gave some guidelines for
future prototyping and experimentation.

Key words numerical modeling, air-jet loom,

main nozzle, drag force

Weft insertion in air-jet looms is operated by high speed air the acceleration tube exit, respectively, highlighting the
jets emitted by a main nozzle; this device works as an ejec- dependence of these pressure values on the acceleration
tor, sucking the yarn and driving it through the warp shed by tube length [1, 2].1 Influence of flow characteristics on weft
means of air friction forces. In fact, it is known that when drag forces was analyzed by the same authors [3, 4].
any body moves through a fluid, an interaction between the Mohamed and Salama carried out experimental tests on
body and the fluid occurs that can be described in terms of nozzles with various geometries. Relationships among air
wall shear stresses, due to viscous effects, i.e. friction, and velocity, turbulence, flow rate at the nozzle exit and nozzle
normal stresses, due to pressure. The resultant force in the structure parameters, such as air tube length and air tube
direction of the upstream velocity is known as drag force. diameter, were reported. A theoretical analysis, based on
Since the yarn inside the main nozzle can be considered as one-dimensional flow, was also carried out to explain main
a zero thickness body parallel to the upstream flow, the drag nozzle performance [5, 6]. Jeong and others analyzed main
is entirely due to the wall shear stresses, i.e. to air friction nozzle flow characteristics, evaluating the effect of using
forces. different acceleration tubes and of installing a suction hole
Large consumption of air and, consequently, increased on them [7]. A numerical approach to analyze main air-jet
manufacturing costs are the main disadvantage of air-jet loom nozzle flow field was proposed by some authors.
looms. Therefore, many research works were addressed to Adamek reported, for example, results on a main nozzle
optimize main nozzle geometry and its functional parame- two-dimensional numerical model, addressed to the analy-
ters, in order to maximize drag forces, as far as allowed by sis of the mixing zone shape [8]. Kim and Song numerically
the yarn material, with acceptable air consumption. Ishida analyzed the influence of air tank pressure, accelerating
and Okajima experimentally studied main nozzle flow
characteristics. They determined tank pressures corre-
sponding to flow critical conditions at the needle tip and at 1
Corresponding author: e-mail guido.belforte@polito.it

Textile Research Journal Vol 79(18): 1664–1669 DOI: 10.1177/0040517508096223 www.trj.sagepub.com © 2009 SAGE Publications
Figures 1–4 appears in color online: http://trj.sagepub.com Los Angeles, London, New Delhi and Singapore

Downloaded from trj.sagepub.com at GEORGIAN COURT UNIV on March 20, 2015

 Numerical Model of an Air-jet Loom Main Nozzle for Drag Forces Evaluation G. Belforte et al. 1665 TRJ

Figure 1 Main nozzle structure.

Figure 2 Flow regions in a main

nozzle.

tube length, and needle tip shape on flow velocity distribu- The main nozzle operation as an ejector is highly influ-
tion [9]. enced by the relative position between the needle and the cyl-
This paper deals with the definition of a numerical model inder. In fact, lowering their distance results in a more intense
able to predict in a reliable way the drag force exerted on suction effect; nevertheless, the device pneumatic resistance
weft yarn by the main nozzle air stream. The developed is increased, so that, at constant supply pressure, supply flow
model was used to analyze the influence of some geometry rate is reduced, together with the drag force on the yarn.
parameters on main nozzle performance. Conversely, if the needle is positioned too far away from the
cylinder, a flux inversion can occur through the hollow nee-
dle to the atmosphere; consequently, yarn insertion would
Air-jet Loom Main Nozzle be more difficult. Therefore, a compromise must be found
between these opposite requirements.
Figure 1 shows the typical structure of a main nozzle used Main nozzle flow field was divided into four regions, as
for weft insertion. The device is mainly made up of four shown in Figure 2. Region 1 includes the supply duct and the
components: a body (1), a hollow needle (2), a cylinder (3), annular flow passage up to the nozzle throat region; region 2
and an acceleration tube (4). includes the nozzle mixing zone, where an interaction occurs
Supply air flows from duct (5) to chamber (6); the polar among the annular main jet, the secondary flow induced by
array of holes (8), manufactured on one of the needle shoul- suction through the hollow needle and the sucked yarn;
ders, acts as a local pneumatic resistance, so that air pressure region 3 includes the hollow needle interior, where yarn inser-
in the chamber (6) can be considered constant. As a conse- tion occurs; region 4 includes the acceleration tube.
quence, the supply flow is redistributed inside the annular In typical main nozzle operation, turbulent flow condi-
passage (7), thus generating an axis-symmetric flow. The tions can be considered, corresponding to high Reynolds
annular passage, bounded by the cone-shaped needle exter- numbers Re. Re is defined as
nal surface and the cylinder internal surface, is convergent
with a minimum cross-sectional area at the needle tip (noz- ρ⋅v⋅d
Re = ------------------
zle throat region). At this region, a strong flow expansion µ
occurs: the accelerated flow sucks the weft yarn (9) through
the hollow needle, pulls it by friction into the acceleration where: ρ = fluid density, v = mean velocity at nozzle
tube and releases freely to the atmosphere, while the weft throat section, µ = fluid viscosity, d = nozzle throat sec-
yarn flies into the warp. tion diameter.

Downloaded from trj.sagepub.com at GEORGIAN COURT UNIV on March 20, 2015

TRJ 1666 Textile Research Journal 79(18)

developed, in order to reduce the computational effort and

cost in terms of CPU time and memory.
According to the mass flow-rate continuity equation,
the mass flow-rate from the supply duct must be equal to
the mass flow-rate through the annular duct; this allowed
neglecting both the supply duct and the polar array of
holes, considering an annular passage as the inlet section.
A constant mass flow rate, equal to that of common com-
mercial nozzles, was set as the inlet boundary condition. If
Figure 3 Experimental apparatus for drag force meas- a constant inlet pressure had been chosen as the inlet
urement. boundary condition, the supply duct and the polar array of
holes would have been considered also and a 3D model
would have been needed. In fact, each local pneumatic
resistance causes a pressure drop.
In particular, flow conditions within region 2 are turbu- Figure 4 shows the flow domain as modeled. A nylon
lent, at Reynolds numbers up to 105. Conversely, flow con- monofilament yarn with a diameter of 0.18 mm was consid-
ditions within region 3 are dependent on the nearby ered; since whip effect was neglected, it was placed at the
regions and are characterized by fairly low Reynolds num- nozzle axis. Major geometry parameters of the modeled
bers, in laminar or turbulent conditions. nozzle are summarized in Table 1.
In order to define a numerical model of the nozzle flow About 12,000 nodes were provided to perform simula-
field, described in the following, a commercial nozzle (noz- tions. Different grids were used, in order to obtain a
zle A, in the following) was chosen. Experimental tests were proper mesh resolution in each specific region of interest.
previously performed on it [10] by means of the experimen- In particular, three-node triangular mesh elements were
tal arrangement shown in Figure 3. A piece of weft yarn was
placed inside the main nozzle and the acceleration tube.
The yarn tail was fixed to a piezoelectric load cell, while its Table 1 Major geometry parameters of the simulated
tip stood 5 mm out of the acceleration tube exit section. nozzle.
When supply air was provided, the yarn was strained by the Nozzle A
air flux and the load cell measured the exerted drag force.
Dyarn Yarn diameter [mm] 0.18

L Main nozzle total length [mm] 288

Computational fluid dynamics RS Hollow needle inside diameter [mm] 1.25

Analysis of an Air-jet Loom Main S Distance from the needle tip to the 0.85
acceleration tube inlet [mm]
Nozzle Rt1 Radius of the acceleration tube inlet 1.65
section[mm]
The commercial finite-volume computational fluid dynam-
ics (CFD) code Fluent was employed to determine numer- Acceleration tube
ically main nozzle flow field and drag force exerted on weft Lt Acceleration tube length [mm] 252
yarn. Since the flow inside the main nozzle is mainly an
axis-symmetric flow, a simplified two-dimensional model α/2 Acceleration tube semi-angle of 0.08
divergence [°]
of nozzle A, rather than a three-dimensional model, was

Figure 4 Main nozzle model.

Downloaded from trj.sagepub.com at GEORGIAN COURT UNIV on March 20, 2015

 Numerical Model of an Air-jet Loom Main Nozzle for Drag Forces Evaluation G. Belforte et al. 1667 TRJ

specified for the flow field, with the exception of the near-
wall region. Near the nozzle wall and the weft yarn, a four-
node quadrilateral boundary layer mesh was provided; the
importance of the boundary layer mesh size will be discuss
in detail below. Mesh density was increased at the nozzle
mixing region and at the weft yarn wall. Grid sensitivity
tests were carried out, evaluating the results trend varying
the mesh size.
The following assumptions were made:

1. Air was assumed to obey ideal gas law.

2. Air viscosity was calculated as a function of absolute
2⁄3
temperature T: µ = µ 0 ⋅ ( T ⁄ T 0 ) , where µ0 and T0 Figure 5 Drag force, Fyarn, versus height, h, of the first
are, respectively, air viscosity and temperature under row of mesh elements.
standard conditions.
3. Inlet mass-flow was set constant and equal to 4.08 g/s;
this value was chosen on the basis of the average
consumptions of many commercial nozzles. Reynolds stress model accounts for the effects of rapid
4. The exit static pressure was specified as atmospheric changes in strain rate in a more rigorous manner than one-
pressure. equation and two-equation models and has greater poten-
5. Turbulent flow conditions were investigated; results tial to give accurate predictions for complex flows.
from different turbulent models were compared. In order to predict shear stress on fluid at the wall, Flu-
6. The near-wall region was modelled by means of ent provides three different wall treatments: standard wall
standard wall function, non-equilibrium wall func- function, non-equilibrium wall function, enhanced wall
tions, and two-layer model (enhanced wall treat- treatment. In fact, in turbulent flow, shear stress calcula-
ment), comparing results. tion does not simply depend on velocity gradients at the
wall, as in laminar flows. Both standard and non-equilib-
By default, Fluent stores discrete values of the variable rium wall functions are semi-empirical formulas used to
of interest at the cell (element) centers; the variable values bridge the viscosity-affected region near the wall and the
at the cell faces are obtained interpolating the cell center fully-turbulent region, where turbulence plays a major role.
values by using an upwind scheme. In particular, a first- Conversely, the two-layer model divides the whole domain
order upwind scheme was used to interpolate pressure, into a viscosity-affected region and a fully-turbulent region,
density and turbulence kinetic energy values, while a sec- without bridging them.
ond-order upwind scheme was used for momentum values Preliminary simulations highlighted that the k–ε turbu-
interpolation. lence model, both with standard and non-equilibrium wall
functions, easily converged. Conversely, the k–ε turbulence
model with enhanced wall treatment tended to diverge.
Influence of flow conditions modeling In all cases, grid sensitivity tests were performed refin-
In order to determine the physical model that allowed a ing the mesh step by step. In particular, it was found that
better prediction of main nozzle behavior, simulations the resolution of the boundary layer mesh nearby the weft
were carried out using various turbulence models (Spalart– yarn highly affected drag force calculation. Simulations
Allmaras, standard k–ε, RNG k–ε, Realizable k–ε, Rey- were carried out providing on the weft yarn a boundary
nolds stress). In fact, unfortunately, no single turbulence layer mesh made up of six rows; the height of the first row,
model is universally accepted as being superior for all classes h, i.e. the distance between the yarn surface and the first
of problem: for each specific problem an optimum model row of mesh nodes, was varied. The distance growth factor,
allowing to avoid numerical diffusion must be found. i.e. the distance between each row and the previous one
The Spalart–Allmaras turbulence model, solving only divided by h, was taken as constant. Figure 5 shows, as an
one turbulence transport equation, is a low-Reynolds- example, the drag force variation changing the height, h, of
number model designed specifically for flows involving the first row of mesh elements, as calculated by the k–ε tur-
separation and reattachment. The k–ε turbulence model, bulence model with non-equilibrium wall functions. Mass
available in three variants (standard, RNG, Realizable) is a flow rate, q3, exiting the hollow needle, which character-
semi-empirical model developed for high-Reynolds-number ized the calculated flow field, was not influenced by h.
flows. Up to now it has been used in most practical engi- Numerical results, summarized in Table 2, were com-
neering flow calculations, because of its economy and rea- pared with available experimental data in terms of drag
sonable accuracy for a wide range of turbulent flows. The force on the yarn, Fyarn, and mass flow rate, q3, through

Downloaded from trj.sagepub.com at GEORGIAN COURT UNIV on March 20, 2015

TRJ 1668 Textile Research Journal 79(18)

Table 2 Numerical results obtained using different Table 3 Main geometry parameters of nozzles under
turbulence models. study.
Fyarn q3 L Lt RS Rt1 S
Flow model Nozzle
[mN] [g/s] [mm] [mm] [mm] [mm] [mm]

Standard k-ε turbulence model with 55 –0.82 A 288 252 1.25 1.65 0.85
non-equilibrium wall functions B 303 280 0.75 1.25 0.65
RNG k-ε turbulence model with 43.5 –0.75
non-equilibrium wall functions
Table 3 allows the comparison of the major geometry
Realizable k-ε turbulence model with 41.8 –0.63
non-equilibrium wall functions parameters of nozzles (see Figure 4).
A constant inlet mass-flow equal to 4.08 g/s was set in
Reynolds stress turbulence model with 36.5 –0.41 all simulations.
non-equilibrium wall functions
Standard k-ε turbulence model with 36 –0.53 Influence of Acceleration Tube Angle of
standard wall functions
Divergence
Spalart-Allmaras turbulence model 22 –0.29 Initially, the influence of acceleration tube semi-angle of
Experimental data 62 +0.17 divergence, α/2, on drag force was evaluated. In the case of
nozzle B, because of technological difficulties in manufac-
turing small and highly divergent acceleration tubes, only
region 3. Mass flow rate positive values refer to flow enter- the following values of α/2 were examined: α/2 = 0°, α/2 =
ing the hollow needle, while negative values refer to flow 0.2° and α/2 = 0.36°.
exiting the hollow needle. Figure 6 shows drag force, Fyarn, versus acceleration tube
As shown, a satisfactory matching of numerical and semi-angle of divergence, α/2; results from simulations are
experimental results in terms of yarn drag force, Fyarn, was compared with available experimental data. As shown, the
obtained using the standard k–ε turbulence model with influence of geometry modifications on drag force is well
non-equilibrium wall functions. This wall treatment dif- reproduced by the numerical model. For each geometry con-
fers from the standard one in that it takes into account figuration an optimum value of angle of divergence, α, able
pressure-gradient effects in the wall-neighbouring cells; to maximize drag force, can be found. By increasing α, the
for this reason it is recommended in the case of flows acceleration tube pneumatic resistance is reduced, so, tak-
involving separation and reattachment, giving improve- ing constant nozzle consumption, supply pressure is reduced
ment in skin friction forces prediction, as confirmed by to a constant value. Nozzle B exhibited the higher value of
the results herein obtained. Conversely, the flow field was drag force, confirming that reducing nozzle dimensions can
better calculated by means of the Spalart–Allmaras turbu- be an optimization criterion. By increasing α, the accelera-
lence model; in fact, even if suction cannot be calculated tion tube pneumatic resistance is reduced, so, taking con-
with this model, inlet mass flow rate convection toward the stant nozzle consumption, supply pressure is reduced to a
acceleration tube has a better correspondence to the real constant value.
case.
Simulations highlighted that, in the specific case under
study, different models must be used for a reliable predic-
tion of each parameter of interest. Accordingly, all follow-
ing simulations were performed using the k–ε turbulence
model with non-equilibrium wall function, for drag forces
evaluation. Flow characteristics were evaluated using the
Spalart–Allmaras model.

Influence of Geometry Parameters on Main

Nozzle Operation
Having tuned the model, the influence of some geometry
parameters on the device behavior was analyzed to
improve performance. In order to evaluate the effect of Figure 6 Drag force, Fyarn, versus acceleration tube
radial size reduction, simulations were also carried out on semi-angle of divergence, α/2.
a new nozzle geometry, called nozzle B in the following.

Downloaded from trj.sagepub.com at GEORGIAN COURT UNIV on March 20, 2015

 Numerical Model of an Air-jet Loom Main Nozzle for Drag Forces Evaluation G. Belforte et al. 1669 TRJ

son with experimental data was satisfying. It was found that

it is possible to identify an optimum value of the accelera-
tion tube length, for which drag force on the yarn can be
maximized, only slightly increasing supply pressure. In any
case, it must be taken into account that the nozzle response
is reduced if the tube length is increased; the nozzle suc-
tion capability is reduced with longer tubes.
Reducing nozzle and acceleration tube sizes allows higher
drag force to be obtained.
Divergent tubes are advantageous with respect to
cylindrical tubes. The optimum value of the semi-angle of
divergence, obtained from simulations and confirmed by
Figure 7 Drag force, Fyarn, versus acceleration tube experimental tests, is about 0.2°.
length, Lt.

Literature Cited
1. Ishida, M., and Okajima, A., Flow Characteristics of an Air
Influence of Acceleration Tube Length
Jet Loom with a Modified Reed and Auxiliary Nozzles. Part 1.
Subsequently, the influence of acceleration tube length, Lt, Flow in a Main Nozzle, J. Textile Machinery Soc. Japan 44(4),
on yarn drag force was evaluated. Simulations were per- 43–54 (1991).
formed on nozzle A connected to a cylindrical acceleration 2. Ishida, M., and Okajima, A., Flow Characteristics of Main
tube (α/2 = 0°). Figure 7 shows drag force, Fyarn, changing Nozzle in an Air-jet Loom, Textil. Res. J., 64(1), 10–20 (1994).
tube length, Lt. As shown, by increasing tube length, drag 3. Ishida, M., and Okajima, A., Flow Characteristics of an Air
force is increased. Jet Loom with a Modified Reed and Auxiliary Nozzles. Part 2.
Measurements of a High Speed Jet Flow from a Main Nozzle
Since numerical simulations were carried out setting the
and a Weft Traction Force, J. Textile Machinery Soc. Japan
same constant inlet mass-flow rate, it is clear that when 45(12), 65–77 (1992).
lengthening the acceleration tube, supply pressure must be 4. Ishida, M., and Okajima, A., Flow Characteristics of an Air
increased with Lt, because the device pneumatic resistance is Jet Loom with a Modified Reed and Auxiliary Nozzles. Part 3:
increased. Nevertheless, in practical applications supply High Speed Jet Flow from a Main Nozzle with Different
pressure is limited to 6 bar. Therefore, drag force will reach a Acceleration Tube Length, J. Textile Machinery Soc. Japan
maximum value at a certain optimum length: further increas- 48(1), 9–19 (1995).
ing this length, the acceleration tube pneumatic resistance 5. Mohamed, M. H., and Salama, M., Mechanics of a Single
will become the most important factor, most of the supply Nozzle Air-jet Filling Insertion System Part I: Nozzle Design
mass flow rate will exit through the hollow needle and a drag and Performance, Textil. Res. J., 56(11), 683–690 (1986).
6. Mohamed, M. H., and Salama, M., Mechanics of a Single
force reduction will occur.
Nozzle Air-jet Filling Insertion System Part II: Velocity Distri-
bution and Design of the Air Guide System, Textil. Res. J.,
56(12), 721–726 (1986).
7. Jeong, S.Y., Kim, K. H., Choi, J.H., and Lee C.K., Design of
Conclusions the Main Nozzle with Different Acceleration Tube and Diam-
eter in an Air-jet Loom, Int. J. Precision Eng. Manufact. 6(1),
Simulations highlighted that it was not possible to define a 23–30 (2005).
single model able to predict in a reliable way both the drag 8. Adamek, K., Numerical Modelling the Air Flow in Parts of Air
force on the yarn and the flow field inside the main nozzle. Jet Loom, J. Comput. Assist. Mech. Eng. Sci., 4, 251–261 (1999).
In particular, it was necessary to use the standard k–ε tur- 9. Kim, S. D., and Song, D. J., A Numerical Analysis of Tran-
sonic/supersonic Flows in the Axisymmetric Main Nozzle of
bulence model, with non-equilibrium wall functions, to
an Air Jet Loom, Textil. Res. J., 71 (9), 783–790 (2001).
analyze the influence of various geometry parameters on 10. Belforte, G., Costamagna, A., Mattiazzo, G., and Testore F.,
drag force. The Spalart–Allmaras turbulence model could Test methodologies for the measure of main nozzles efficiency
give better results in terms of flow distribution. in air jet looms, in “Proceedings of the 5th World Textile Con-
Using the developed numerical model, various main ference AUTEX 2005”, 27–29 June, Portorož, Slovenia, 2005,
nozzle geometry configurations were examined. Compari- pp. 762–767.

Downloaded from trj.sagepub.com at GEORGIAN COURT UNIV on March 20, 2015