Progress in Computational Fluid Dynamics, Vol. 20, No.

5, 2020 249

Effect of design parameters on flow characteristics

of an aerodynamic swirl nozzle

Sharif M. Islam, Md. Tanvir Khan and Zahir U. Ahmed*

Department of Mechanical Engineering,
Khulna University of Engineering and Technology,
Khulna 9203, Bangladesh
Email: [email protected]
Email: [email protected]
Email: [email protected]
*Corresponding author

Abstract: This study numerically investigates the flow characteristics for the effect of various
nozzle design variables, such as tangential inlet number, length of the nozzle and nozzle shape of
an incompressible turbulent swirl air jet. Axial-plus-tangential flow based swirling nozzle is
solved using finite volume method, where turbulence is approximated by Shear Stress Transport
(SST) k-ω model. The results show that swirl into the flow results in a creation of a forced
vortex. The axial velocity and its peak increase with the number of ports, and the tangential
velocity is the maximum after the tangential flow is imparted. The static pressure and turbulent
kinetic energy are significantly influenced by the number of tangential inlets. The most uniform
turbulence is predicted for the tapered cone shape, while the strongest turbulence is observed for
the curved shape with the expense of losing some uniformity near the nozzle wall.

Keywords: turbulent; nozzle; swirl; aerodynamic; CFD; pressure.

Reference to this paper should be made as follows: Islam, S.M., Khan, M.T. and Ahmed, Z.U.
(2020) ‘Effect of design parameters on flow characteristics of an aerodynamic swirl nozzle’,
Progress in Computational Fluid Dynamics, Vol. 20, No. 5, pp.249–262.
Biographical notes: Sharif M. Islam was awarded his BSc in Mechanical Engineering from the
Department of Mechanical Engineering at the Khulna University of Engineering and Technology
in 2018. He also served as a Research Assistant in the same department for a year. His research
interest falls into computational fluid dynamics of fluid flow and earth-to-air heat exchanger.

Md. Tanvir Khan received his BSc in Mechanical Engineering from the Department of
Mechanical Engineering at Khulna University of Engineering and Technology in 2018. His
research interests are CFD and heat transfer with nanofluids.

Zahir U. Ahmed is a Professor of the Department of Mechanical Engineering at Khulna

University of Engineering & Technology, Bangladesh since 2019. He also has industrial
experiences before starting his academic career. He had received his MESc in Mechanical and
Materials Engineering from The University of Western Ontario, Canada and PhD from the Edith
Cowan University, Australia. His research interests lie in the area of jets, two phase flows and
instability, renewable energy and heat transfer enhancements. He is also active in both theoretical
and experimental research in basic fluid mechanics.

1 Introduction encountered in free jets and wake flows behind obstacles

(for geometry induced swirl). The presence of the tangential
Swirl flows are generally complex in nature and have wide
velocity in swirl flow sets up of the radial and axial pressure
range of engineering applications. In swirling flows ideally
gradients which, in turn, influence both the downstream free
either a solid body rotation or free vortex flow emerges
flow development and wall-bounded flows. In case of very
due to the effect of angular momentum. Vortex patterns
strong swirl, the adverse axial pressure gradient is
occurring in real applications can usually be characterised
sufficiently large to result in a reversed flow along the axis,
by a solid body rotation in the core, surrounded by a free
often called as vortex breakdown. Swirl had been generated
vortex like pattern. Tangential velocity increases with the
within a nozzle in many ways in the past, such as rotating
increase of radius for solid body rotation and decreases for
pipe (Rose, 1962), axial-plus-tangential entry (Toh et al.,
free vortex flow. The aerodynamics of the swirling turbulent
2010; Ahmed, 2016) or rotating vane or perforated plate
jet embodies interesting characteristics of rotating turbulent
inside stationary pipe (Gore and Ranz, 1964), ducted
motion as well as the non-isotropic free turbulence
propeller (Felli et al., 2008), helical or twisted tape (Jafaria

Copyright © 2020 Inderscience Enterprises Ltd.

250 S.M. Islam et al.

et al., 2017). When using swirl vanes, the flow was not flows are typically encountered in various industrial
found axisymmetric, and the flow field was complicated by applications with diverse geometrical setups, a generalised
additional secondary flows. However, majority of the study analysis is difficult. Due to the continuous improvement of
associated with swirl had focused on geometry induced flow computer and computational methods over the last few
where swirl was generated by twisted tape insert or helical decades, CFD now becomes one of the reliable and popular
screw (Markal, 2018). The obvious drawback of such research tools to deal with complex flows. In this regard,
system is the lack of control on swirl intensity and difficulty turbulent swirling flows are also investigated numerically
to reproduce flows due to strong dependency on geometrical by many researchers. Numerical modelling of those flows
accuracy. In contrast, aerodynamic swirl generation system still remains a great challenge due to the appearance
or axial-plus-tangential entry flow is advantageous in the of flow induced curvature, strong pressure gradient,
sense that it has no geometry-induced flow reversal or three-dimensional character of the hydrodynamics,
trailing vortices and better control over the large domain of highly anisotropic nature of turbulence and coherent
swirl intensity. Experimental investigation on aerodynamic structure. Several turbulence modelling techniques, such as
swirl flow generation has recently been progressed parallel Reynolds averaged Navier-Stokes (RANS), unsteady RANS
with geometry-induced swirl flow. However, fundamental (URANS), large eddy simulation (LES) and direct
flow behaviour inside an aerodynamic swirl nozzle is still numerical simulation (DNS) had already been applied for
not well understood. turbulent swirling flows and an up-to-date detail of these
Dumouchel et al. (1993) carried out a theoretical techniques are accessible via recent treatises (Batten et al.,
analysis at the orifice of a swirl spray nozzle using a viscous 2009; Argyropoulos and Markatos, 2015; Benim et al.,
flow model. He found that the characteristics of the conical 2017). Whilst DNS requires very fine grid resolutions and
liquid sheet produced at the nozzle’s orifice are mainly large amount of time, as well as is impractical to use at high
functions of the nozzle geometry. Ullrich (1960) used a Reynolds number flows, LES, on the other hand, used
combination of tangential air inlets and adjustable vanes in recently by number of researchers (Benim et al., 2010; Wu
his examination of annular swirling jets. Nouri-Borujerdi et al., 2016). Both the time-mean and turbulence field were
and Kebriaee (2012) observed the boundary layer thickness, found to be captured with convincing accuracy by LES in
which was found dependent on the velocity ratios, Reynolds comparison with experimental data. However, this approach
number and nozzle angle. Due to the short length of the also requires a sufficiently fine mesh to resolve small-scale
nozzle, the momentum change on pressure loss is more turbulence characteristics accurately. Among various RANS
significant than that on the shear stress. In contrast with the turbulence models, the Reynolds stress model (RSM) had
non-swirling flow, the jet with swirl spreads at a larger generally been applied in strong swirling flows due to its
angle, entrains fluid more rapidly, and consequently ability to resolve six individual stress components, leading
displays a more rapid reduction of mean-velocity and to tackle anisotropy better (Jawarneh and Vatistas, 2006;
growth of turbulent intensity. Yajnik and Subbaiah (1973) Wang et al., 2006). However, this hypothesis is not
found that the mean velocity profiles, both away from and universal, and RSM was also found to be inadequate beyond
close to the wall, admit similarity representations at some transitional swirl level (Benim and Nahavandi, 2003;
sufficiently large Reynolds numbers, provided that flow Lu and Semião, 2003; Tsai et al., 1995). Despite Boussinesq
reversal does not take place near the entrance. Toh et al. approximation-based turbulence models (two-equation) are
(2010) observed that higher the spreading rate, higher the believed to be less efficient than the second-moment closure
centreline velocity decay and larger turbulence level. They models, they were also used significantly in the literature. In
also found that, when degree of swirl is sufficiently high, some studies, researchers claimed acceptable accuracy in
vortex breakdown occurs. Chang and Dhir (1994) showed comparison with experimental data and LES (Gorman et al.,
that the axial velocity profile shows existence of a flow 2016; Saqr and Wahid, 2014; Hreiz et al., 2011). In
reversal region in the central portion of the tube and an summary, it appears that no turbulence model performs the
increased axial velocity near the wall. Tangential velocity best for all kinds of swirl flows. Because of the large variety
profiles have a local maximum and the location of which of swirl flows, it is possible that a model which would be
moves radially inwards with distance. Turbulence intensities well suited for an application performs only marginally or
are significantly increased by swirl motion due to the poorly for another type of swirl flow. LES is receiving
destabilising distribution of angular momentum in the free increasing interest in number of swirl flow studies, such as
vortex zone and the large shear stress near the boundary of isothermal and reactive combustor flows. RANS-based
the reversed flow. Chigier and Beer (1964) showed for the simulations, on the other hand, on a relatively coarse grid
case of strong swirl, a vortex is generated in the region close able to provide reasonable and industrially relevant results
to the orifice resulting in a displacement of the maximum with limited computational effort, but with the expense of
axial velocity from the jet axis. After a distance of detailed flow physics.
10 diameters, the influence of the vortex motion becomes Concerning swirl flows in straight pipes, Najafi et al.
small and similarity of profiles is obtained farther (2005) and Escue and Cui (2010) simulated swirl flows
downstream. using an axisymmetric geometry, and they found the
The above experimental studies revealed complex RSM and RNG k-ε models give satisfactory performance.
behaviour of turbulent swirl flows. Since turbulent swirl The former reported that the RSM model coupled to a
 Effect of design parameters on flow characteristics of an aerodynamic swirl nozzle 251

double-layer approach for the near-wall treatment performs turbulence behaviours for a swirl nozzle is very limited in
slightly better. The latter found two-equation model the literature. As such, the current study will bridge this gap
performed better compared to RSM in swirling flows. by conducting a numerical study by changing different
Saqr and Wahid (2014) achieved good agreement with design parameters in an aerodynamic nozzle, which has a
experiment using realisable k-ε model. Hreiz et al. (2011) flexibility of achieving from no swirl to high swirl from the
carried out RANS, URANS and LES simulations in same nozzle settings. The velocity profiles at the nozzle exit
cylindrical cyclonic flow and found that realisable k-ε plane would serve as boundary conditions for other heat
showed comparable performance with LES. Thomas et al. transfer applications, such as impinging jets. This paper will
(2013) presented a CFD analysis for the effect of examine the effects of number of tangential ports, length of
operational settings on an aerodynamic swirl flow, but only the nozzle and shape of the nozzle on flow patterns, the
at the nozzle exit plane, with limited numerical details to development of the mean flow and turbulence fields, as well
reproduce the work. Additionally, they used standard k-ε as the velocity profiles at exit plane of the nozzle.
model, which is known to have insufficient capacity to
predict strong turbulence or swirl flow. Islam et al. (2017)
carried out a numerical investigation using SST k-ω model 2 Methodology
for exit velocity profiles of a swirl flow. The result showed
An aerodynamically generated base swirl nozzle, replica
that introduction of low-level swirls into an impinging
of the nozzle available at Thermofluids Laboratory,
jet results in centreline velocity decay but a significant
Edith Cowan University, Australia, is modelled in order to
reduction in turbulent kinetic energy at the wall jet region.
achieve realistic results and to the availability of datasets at
In close geometrical relevance, other researchers, such
the nozzle exit to test the current numerical methodology.
as Yang and Kaer (2012) and Vijiapurapu and Cui (2010),
Detailed dimension of this nozzle is available in Thomas
concluded that both LES and RANS-based models able to
et al. (2013) and Ahmed (2016), hence is not repeated here
capture dominant flow structure. Gorman et al. (2016) in
for brevity. The nozzle was equipped with three tangential
swirling flows assessed five different RANS models and
ports (ID 12 mm) and upstream of them uniform axial flow
LES, and concluded that whilst LES shows slightly better
entered. The uniform flow was ensured by the flow settling
result, SST k-ω outperforms other RANS models. They also
chamber with honeycomb. Both axial and tangential flows
reported that the large computation cost (larger than
then mix together downstream of tangential inlets to
ten times) may not be worthy to execute LES for mild
generate aerodynamic swirl. The geometric parameters
improvement in that case. Tamrina et al. (2016) analysed
along with the base swirl nozzle are shown in Figure 1. The
turbulent swirling pipe flow induced by guide vanes using
only modification in this study compared to the nozzle at
numerical simulations in a two-dimensional axisymmetric
Edith Cowan University is that axial ports with settling
channel and, they presented and compared numerical results
chamber, to achieve uniform flow, are omitted here. Instead,
for the three components of the velocity distributions for
a uniform flow is directly applied in the base (bottom) of the
both k-ε and RSM model. They found that the k-ε model
nozzle to reduce numerical complexity and computational
fails to predict the main flow but, RSM yields a better
cost. Figure 1(a) shows the three-dimensional CAD view of
result.
the aerodynamic swirl nozzle. A converging section links
It appears that existing literature mostly is populated in
two straight sections: one in upstream of it where tangential
the area of gas turbine combustors or suddenly expanded
ports are connected and another in downstream. The swirl
flows due to the highly industrial relevance, with flow
nozzle has inlet diameter 50 mm, the diameter after the
physics that are, however, not the same as the current study.
convergent section and at the exit (D) is 40 mm, with a
Round jets are also used in number of engineering
total length of 577 mm. Figure 1(b) shows three variants of
applications whereby jets are either non-swirling or
tangential ports: 2-tangential ports (2TP), 3-tangential ports
swirling. A great geometrical variability also exists in
(3TP) and 4-tangential ports (4TP). Figure 1(c) depicts the
generating swirl motions in the nozzle, which makes
four variations of straight portion of the nozzle above the
difficult to draw summative comments for the effect of
converging segment: 0 mm (L0), 80 mm (L80), 240 mm
design parameters on swirl flow. Majority of existing
(L240) and 400 mm (L400). Finally, Figure 1(d) shows
swirling flows with cylindrical geometry primarily focused
three different nozzle shapes considered in this investigation
on assessing turbulence models against benchmark
to replace the converging section in L0 case: straight (S),
experimental data without much focusing on how flow
tapered cone (T) and 5th degree polynomial curved (C). The
develops along the length of the nozzle, particularly for the
diameters of S, T and C are 50 mm, 40 mm and 30 mm,
same inflow conditions. Thus, a systematic study for the
respectively.
effect of various geometric parameters on mean and
252 S.M. Islam et al.

Figure 1 (a) Three-dimensional view of an aerodynamic swirl nozzle (b) Different number of tangential ports
(c) Various lengths of the straight portion of the nozzle (d) Three different shapes for the converging section (see online version
for colours)

(a) (b) (c) (d)

In this study, RANS approach is applied to solve the performing RANS models in swirling and wall-bounded
governing equations, as the paper primarily focuses on flows (Gorman et al., 2016; Ahmed et al., 2017).
time-mean behaviour and does not look into the The above governing equations (1)–(3) and
time-variant turbulence behaviour and vortical structure, in two transport equations (SST k-ω) are solved by finite
which cases LES is preferred for accurate prediction with volume-based commercial software ANSYS Fluent v17.
larger computational cost. Thus, the governing equations The pressure-based coupled algorithm is used to
used for the incompressible three-dimensional steady state simultaneously solve the coupled system of continuity and
turbulent air jets are the mass and momentum conservations, momentum equations, whereas turbulence quantities are
which are in indicial notations: solved separately in a segregated manner; an approach
which significantly improves the rate of convergence when
Ui = 0 (1)
compared to a segregated algorithm (Ahmed et al., 2015b).
 ∂U i ∂U j  ∂P The pressure staggering option (PRESTO) is applied
ρ U i +U j  =− for the pressure discretisation as it is suitable for steep
 ∂X i ∂X j  ∂X i
pressure gradients such as those in swirling jets, and the
(2)
∂   ∂U i ∂U j   second-order upwind discretisation scheme is used for
+  μeff  +  − ρuι′u J′  the momentum, turbulent kinetic energy and dissipation
∂X i   ∂X j ∂X i   equations. Mapped face meshing was applied on the outlet
Here, μeff is the effective viscosity of the fluid, and is equal face, multi-zone meshing was used on the outer parts and
to the sum of molecular viscosity (μ) and eddy viscosity body sizing was used on the part containing tangential
inlets. Hexa-hedral mapped mesh type was used in
(μt). The term uι′u J′ is unknown and called Reynolds
multi-zone meshing with an element size 0.0025 m in body
stresses, which governs the turbulence. The Reynolds stress
sizing. Above numerical settings ensured y+ values less than
components within the term uι′u J′ in equation (2) are 0.2 for the whole domain. The solution is assumed to be
determined via mean velocity gradients by the Boussinesq converged when the residuals of the flow parameters are
hypothesis using turbulent (eddy) viscosity μt. The less than 10–5. Mass conservation is also checked between
Boussinesq approximation is expressed as: inlet ports and the nozzle exit for converged solutions and
the difference between inlets and outlet is found close to
2  ∂U ∂U j 
ui′u ′j = kδij − μt  i +  (3) zero. Another check is done for integral momentum balance
3  ∂X i ∂X j  at immediately after the tangential entry plane and at the
nozzle exit plane. Axial flux of angular momentum at the
where μt is a function of k and ω, which are determined via
plane after the tangential ports and at the nozzle exit is
an appropriate turbulence model. In this research, the SST R
k-ω model, proposed by Menter and Egorov (2010),
is chosen to model turbulence transport quantities:
determined by 2πρ  0
UWr 2 dr , neglecting second-order

turbulence kinetic energy, k, and the specific dissipation turbulence stress component term (Chigier and Beer, 1964).
rate, ω. SST k-ω model was found to be one of the best The angular momentum balance at the exit plane was found
to be reduced by 39.28% than that was at the tangential
 Effect of design parameters on flow characteristics of an aerodynamic swirl nozzle 253

entry plane, which was assumed to be dissipated at the from the bottom of the nozzle. For other proportions,
nozzle wall. A typical mesh independency test is shown in intermediate swirl flows via axial-plus-tangential flow is
Table 1 for 3TP as representative. The final mesh, achieved. Regardless of the swirl intensity, same mass flow
after mesh independency, contains 606,938 elements rate is used through all inlets, as shown in Table 2. The
for 2TP nozzle, 974,001 elements for 3TP nozzle and mass flow rates used here are similar to the experimental
1,698,505 elements for the 4TP nozzle. A typical mesh conditions (Ahmed et al., 2015a). The flow ratio Qr is the
generation of the swirl nozzle with three tangential ports ratio of total mass flow rate through tangential inlets to the
along with magnified view of near-wall mesh of 3TP and total flow rates in the nozzle, i.e., sum of flowrates through
4TP, as well as coordinate system is shown in Figure 2. all axial and tangential inlets. The corresponding swirl
Next a sensitivity test for the number of inflation layers is number used in that experimental study (Ahmed et al.,
examined in Figure 3, and ten inflation layers with 2015a) is also used in the table for convenience. The
enhanced wall treatment function were used around pressure outlet condition is applied at the nozzle exit with
the nozzle wall to better resolve the near-wall flow atmospheric pressure whereby turbulence is specified by 2%
characteristics. Cartesian coordinates are given by (x, y, z) intensity and hydraulic diameter. Finally, no-slip condition
with corresponding velocity components (Ux, Uy, Uz). For at the nozzle wall with ambient temperature is used.
convenience, results of this 3D simulation are presented by
polar coordinates (x, r, θ) with x-axis coinciding for each Figure 3 Sensitivity of the number of inflation layers near the
system. In this regard, the corresponding axial (U) and nozzle wall (see online version for colours)
tangential (W) velocity components are derived from the
Cartesian coordinate data using the axes transformation
rule.

Figure 2 (a) Mesh generation isometric view

(b)–(c) Mesh generation inlet section

(a)
Table 1 Mesh independency test for 3TP case

Number of Number of Axial velocity

Mesh % error
nodes elements U (at r = 0)
Mesh 1 614,650 684,558 5.0674 0.882
Mesh 2 814,900 974,001 5.0204 0.054
Mesh 3 1,134,481 1,310,326 5.0235 0.008
(b) Mesh 4 1,452,959 1,647,089 5.0231 -

Table 2 Mass flow inlet conditions for axial and tangential

flows

Tangential mass
Corresponding Axial mass
Flow ratio flow rate in
swirl number flow rate
Qr each tangential
(Ahmed 2016) (kg/s)
port (kg/s)
0 0 0.0066 0
0.33 0.21 0.0044 0.00073
(c)
0.55 0.32 0.003 0.0012
For the inlet conditions of the model, mass flow inlets
1 0.74 0 0.0022
are used both at the bottom of the nozzle (upstream
of tangential ports) and through the tangential ports. For
no-swirl flow, only axial flow from the bottom of the nozzle
is provided with zero flows from tangential ports. 3 Result and discussion
Conversely, for the strongest swirl flows, only flows from This section initiates with numerical validation against
tangential inlets are allowed to pass with no axial flow experimental data. The effect of number of tangential inlet
254 S.M. Islam et al.

ports in the nozzle, shape of the nozzle at the exit and length sufficient for conclusive investigation from this nozzle. For
of the nozzle on fluid flow behaviours of both non-swirling convenience, results of some parameters will be sought at
and swirling flow is systematically discussed next. seven different planes, namely 51 mm (x/D = 1.275),
140 mm (x/D = 3.5), 160 mm (x/D = 4.0), 190 mm
3.1 Validation (x/D = 4.75), 350 mm, (x/D = 8.75), 500 mm (x/D = 12.5)
and at the exit plane, i.e., 577 mm (x/D = 14.425) from the
The results obtained from the simulations of the swirl nozzle inlet. Figure 5 shows the position of these planes
nozzle are first validated by comparing with experimental correspond to the nozzle. The planes are chosen such that
data (Ahmed et al., 2015a) with three tangential ports for the differences of results due to geometric variations of the
the case Qr = 0, 0.33, 0.55 and 1.00, which are equivalent to nozzle can easily be identified.
the swirl numbers, S = 0, 0.21, 0.32 and 0.74 at the exit
plane. The results in Figure 4 shows a good agreement Figure 4 Data validation of, (a) U/Ub vs. r/D chart (b) W/Ub
between the numerical prediction and the experimental data vs. r/D chart for Qr = 0, 0.33, 0.55 and 1 with
for Qr = 0. A slight over-prediction of the axial velocity experimental data from Ahmed et al. (2015a)
U/Ub for Qr = 0.33 is seen, particularly in the shear layer, (see online version for colours)
which is subdued in the tangential velocity. A relatively
higher level of deviation is observed for Qr = 0.55 and
Qr = 1 in the core region and near the wall, respectively,
where large mean velocity and intense turbulence typically
occurs for strong swirling flows. Discrepancy in results
between experimental data and numerical simulation
is primarily due to the possible measurement inaccuracy
associated with CTA X-wire and experimental
flow settings. It is known that measurements of
three-dimensional and highly swirling turbulent flows with
CTA probes are typically challenging and susceptible to
inaccuracies. The inaccuracies mainly stem from larger
resultant velocity vector angle, high velocity gradients,
strong turbulence and out-of-plane contaminations (Chang
and Dhir, 1994; Fitouri et al., 1994; Ovink et al., 2001;
Ahmed et al., 2016b). This results an overestimation of the
mean velocity components than their corresponding true
values. Chang and Dhir (1994), and later Champagne and
Kromat (2000) quantified such overestimation and reported
that mean velocity errors of around 30% of centreline
velocity are expected for strong swirl numbers in the region
of high turbulence. The former also reported that the flow
rate calculated by integrating the axial velocity profile over Figure 5 Different cross-sectional planes along the nozzle for
the cross section is within 8% of the flow rate obtained from contour presentation
the flowmeters. In contrast, tangential velocity (W/Ub) at the
nozzle exit is found to be good agreement in relation
to the profile behaviours and peak locations, with the only
exception of the reduced magnitude in the predication. The
deviations may also be partly attributed to the inability of
RANS approach to accurately capture highly swirling flows.
The ensuing results and discussion section includes
presentation of mean and turbulence controlling parameters,
such as velocity, static pressure, wall shear stress and
turbulent kinetic energy for the effect of nozzle design
parameters. In this regard, two flow conditions (Qr = 0 and
Qr = 1) are chosen in this study. The reason for choosing
only two swirl intensity is that previous studies (Ahmed,
2016; Ahmed et al., 2015a; Khan and Ahmed, 2019) of fluid
flow and heat transfer from the similar nozzle showed that
only two types of flow behaviours exist regardless of the
swirl intensity from the nozzle: a non-swirl-like (Qr = 0)
and strongly swirling (Qr = 1). Intermediate ranges of Qr
normally falls in either of these flow behaviours. As such,
considering these two extremities of Qr may likely be
 Effect of design parameters on flow characteristics of an aerodynamic swirl nozzle 255

3.2 Effect of number of tangential inlet ports Figure 6 Normalised axial (U/Ub) and tangential (W/Ub) velocity
distributions along the length of the nozzle for different
The effect of number of tangential ports (2TP, 3TP and radial locations (see online version for colours)
4TP) on normalised axial (U/Ub) and tangential (W/Ub)
velocity distributions along the nozzle length at the nozzle
centre (r/D = 0), at the half-radius (r/D = 0.25) and near
the wall (r/D = 0.49) are presented in Figure 6. The
normalisation is done by the bulk or average velocity (Ub),
determined from the total volumetric flow rate divided by
the area of the nozzle exit plane. Figures 6(a) and 6(b) show
the axial and tangential velocity component, respectively.
The axial velocity profile at r/D = 0 is very smooth for the
no swirl condition. Gradually increasing from the inlet, the
axial velocity suddenly raises at the nozzle converging
section (x/D = 3–4), followed by steady augmentation in
magnitude. For other radial positions in non-swirl flow, the
axial velocity slowly drops with the increase of both the
radial position and axial distance after converging section.
This is attributed to the viscous layer and applied no-slip
boundary condition at the wall.
In swirl case, the axial velocity is no longer smooth,
rather it is highly fluctuating for all radial positions. The
axial velocity becomes negative at the nozzle wall
proximity (r/D = 0.49) with a trough at x/D ≈ 0.6, the
distance immediately before tangential ports are applied.
This negative peak tends to diminish with increasing Figure 6(b) demonstrates that the tangential velocity at the
number of ports. The velocity again sharply increases and nozzle centre is close to zero. However, for the other
goes to the peak after flow amalgamation of the tangential positions, the velocity is highly fluctuating and quite
flows at x/D ≈ 1.5, regardless of the number of ports. The unpredictable because of the complex mixing of the flow at
velocities then decrease with another trough immediately tangential inlets. The tangential velocity is the highest
before converging section starts, followed by an increase up immediately after the tangential inlets and then decreases
to x/D ≈ 4. The velocity profiles there show some slowly towards the nozzle exit. Although a significant
fluctuations around a steady flow. Such fluctuations at off- variation in velocity magnitude for different number of
centre locations for 2TP and 3TP is clearly not in phase, tangential ports is observed in the upstream, less difference
rather random with a number of weak peaks and troughs. In in magnitude is predicted at the later half of the nozzle.
contrast, a phase-like fluctuations at off-centre locations Similar to the axial velocity profile, 4TP shows stronger
are observed for 4TP case. Such strong fluctuations or repeating fluctuations than 2TP and 3TP.
phase-like fluctuations may be attributed to asymetric Figure 7 depicts the radial distribution of axial
circulation or associated instabilities from vortex [Figure 7(a)] and tangential [Figure 7(b)] velocity
breakdown, which, however, very difficult to ascertain by components at the nozzle exit plane (x/D = 14.425). Such
steady state analysis. Although for the 2TP, 3TP and 4TP velocity distributions help correlating experimental data via
the overall flow velocity profile is similar, the highest and intrusive measurements such as hotwire anemometer or
lowest velocity locations are not the same in all cases. non-intrusive measurements, i.e., LDV or PIV which are
On the other hand, the velocity profiles at or near the nozzle typically applied just above the nozzle exit. This may also
centre is quite different from the near-wall velocities in be useful to consider as an inlet condition of swirling free or
every case. Axial velocity develops initially from the nozzle impinging jets/flames studies for numerical simulation. In
inlet in every case, then reduces before the converging general, the radial distributions of both axial and tangential
section due to the flow shifting towards the nozzle wall, and velocities for 2TP and 3TP appear to be similar, but
then increases again in the convergent section, as expected. completely different profiles are observed for 4TP. An only
Interestingly, velocity at the nozzle centre diminishes exception for 3TP is the slight reduction of axial velocity at
abruptly towards the nozzle exit plane, which perhaps due the nozzle core region. Such variation in magnitude of axial
to the cumulative effect of centrifugal force, appearance of velocity may intense the flow recirculation and affect heat
significant radial velocity at the exit plane and pressure transfer when this jet strikes the impingement surface
variations between inside of the nozzle and ambient. It (Ahmed et al., 2016a). The axial velocity [Figure 7(a)]
appears that axial velocity becomes higher with the increase increases with radius and becomes maximum at the shear
of number of tangential ports, which may have an impact on layer at r/D ≈ 0.47, followed by a sudden drop due to
the free and impinging jet behaviours. no-slip condition at the nozzle wall. The axial velocity for
4TP, however, becomes almost uniform, with a steady
256 S.M. Islam et al.

increase with radius up to the shear layer. This distribution layer at the nozzle exit plane is very thin compared to
represents likelihood of non-swirl-like behaviour for both non-swirl flow, but the extent of the boundary layer is
free and impinging jets. The normalised tangential velocity almost the same for any number of ports.
increases almost linearly for all number of tangential ports,
indicating a solid body rotation. For 4TP, magnitude of Figure 8 The effect of number of tangential ports on resultant
tangential velocity is significantly low across the nozzle velocity magnitudes (dimensionless) at different axial
planes in the range x/D = 1.275–12.5 (see online
radius.
version for colours)
Figure 7 Normalised axial (U/Ub) and tangetial (W/Ub) velocity
profiles at the nozzle exit plane (x/D = 14.425)
(see online version for colours)

Contours of the pressure coefficient (Cp) at different axial

planes of the nozzle are shown in Figure 9. It appears that
the pressure is the maximum near the inlet and decreases as
the flow moves towards the exit plane and becomes equal to
the atmospheric pressure at the exit for Qr = 0. The pressure
remains uniform for the whole length of the nozzle, as
expected. In contrast, considerably different characteristics
along the length of the nozzle are observed for Qr = 1. At
x/D = 1.275, near the tangential inlets, the pressure is the
highest with a minimum around the centreline. When the
fluid flows through the nozzle towards the exit plane, the
Figure 8 demonstrates the effect of number of tangential pressure starts to reduce and becomes the lowest near the
ports on contours of resultant velocity magnitudes at various outlet. A pressure drop around the centreline always exists
axial locations along the flow direction, i.e., x/D = 1.275, due to the occurrence of strong centrifugal action, regardless
3.5, 4, 4.75 and 12.5 and compared with non-swirl flow of the number of tangential ports and length of the nozzle.
(Qr = 0). For non-swirling flow, the velocity is almost Near the tangential inlets, strong negative pressure
constant before the converging section and uniform coefficient is observed for 2TP. As the number of ports
throughout the planes. After the converging section the increases, the strength of the negative pressure reduces,
velocity begins to increase. The velocity is found to be the eventually leading to a positive value for 4TP. The effect of
most uniform after the converging section, but the number of ports appears to be insignificant after x/D = 4.75.
magnitude is the highest near the exit plane, with a Figure 10 shows the turbulent kinetic energy (k) contour
reduction towards the wall due to boundary layer formation. at the same axial locations mentioned above to elucidate the
In contrast, for swirling flow (Qr = 1), the velocity effect of number of tangential inlet ports. As there is a
magnitude is found to be the highest at some patches on the uniform flow and very little turbulence at the nozzle inlet,
plane at x/D = 1.275 for 2TP, 3TP and 4TP cases. Shape of turbulent kinetic energy is negligible before the converging
such patches is dependent on the number of tangential ports. section, and it starts to increase near the wall due to the
As the flow approaches the converging section the velocity shear layer development after the converging section. The
begins to decrease but becomes uniform across the plane. magnitude is, however, very small compared to the swirl
At x/D = 3.5 the velocity is uniform throughout the plane case, hence not visible here. However, for Qr = 1, turbulent
with slight decay around the centreline. The velocity kinetic energy at the tangential inlet region is the highest
reduces towards the wall due to the formation of the due to the strong mixing in this region and eventually it
boundary layer. After the converging section, the velocity decreases as the fluid flows to the exit. Turbulent kinetic
begins to increase but remains nearly uniform throughout energy is the highest for 2TP, but nearly the same for 3TP
the plane. After leaving the converging section, a very slight and 4TP.
change is observed as the fluid flows through the nozzle
towards the exit. It is worth noting here that the boundary
 Effect of design parameters on flow characteristics of an aerodynamic swirl nozzle 257

3.3 Effect of nozzle length Figure 11 Effect of length of the straight section on streamlines
for both non-swirling and swirling jets (see online
The effect of the length of the straight section is examined version for colours)
in Figure 11 by presenting streamlines for four different
lengths, namely L0, L80, L240 and L400. The results for
swirling jets are compared with non-swirling jets. It appears
that the streamlines for non-swirling flow is similar in all
lengths before and immediately after the converging
section. The effect of length is only visible in the formation
of boundary layers at L240 and L400 only, where a thicker
boundary layer is seen for L400, as expected. L0 and L80
ensure uniform flow at the nozzle exit. For swirl condition,
on the other hand, streamlines are found to be twisting
and fluttering. The intensity of such twisting effect and
magnitude of velocity field is stronger as the length of the
straight pipe increases.

 p − p∞  Figure 12 Effect of length of the nozzle for swirl flow on,

Figure 9 Pressure coefficient  C p = contours
 1 / 2 ρU b2  (a) pressure coefficient (b) normalised axial velocity
at different axial location of the nozzle for various (U/Ub) (see online version for colours)
number of tangential inlet ports (see online version
for colours)

(a)

Figure 10 Turbulent kinetic energy contours (m2/s2) for the

effect of different number of tangential ports for
non-swirling and swirling jets (see online version
for colours)

(b)
258 S.M. Islam et al.

Figure 12 shows the effect of nozzle length on the static profiles vary linearly with radius up to the shear layer
pressure coefficient [Figure 12(a)] and the axial velocity (r/D = 0.45) and behaving solid body type tangential
[Figure 12(b)] for swirl flow (Qr = 1) only. A low pressure velocity. The strongest tangential velocity is obtained for
zone around the nozzle centreline is found at x/D = 1.275 L0, as expected. With the increase of length the magnitude
for all lengths, with three circumferential peaks. As the flow of tangential velocity in the shear layer (only) reduces
moves downstream, low pressure region spreads towards significantly.
the wall with an increase in magnitude up to the converging Turbulent kinetic energy for the effect of different tube
section. The pressure is found significantly small but lengths (L0, L80, L240 and L400) at various axial locations,
uniform due to the converging effect. This behaviour is i.e., x/D = 1.5, 3.5, 4.25 and the exit plane is shown in
common to all nozzle lengths, i.e., L0, L80, L240 and L400. Figure 14 only for swirling flow. It appears that strong
The axial velocity is found to be the maximum at the edge turbulent kinetic energy is localised around the centre
of a triangle formed around the centreline at x/D = 1.275 and the length of straight section has little effect on the
after tangential flow imparts. This strong non-uniformity, magnitude and distribution of turbulence immediately after
however, completely vanishes after the converging section. the tangential imparts. In contrast, a uniform turbulence,
Again, the effect of different lengths an axial velocity is with a ring of low turbulence near the wall, is predicted for
minimal and only relatively thicker boundary layer is the whole length of L0 and L80, and up to x/D = 4.25 for
predicted for L280 and L400 cases. The static pressure and L240. The low turbulence region moves towards the centre
axial velocity are also examined (not shown here) for as the length of the straight section further increases. The
non-swirling flow and found that uniform contours with turbulent kinetic energy is found to be very small, but
varied magnitudes are present along the nozzle, with little or uniform across the nozzle diameter, at the exit plane for
no effect of length increase. L240 and L400. This turbulent kinetic energy behaviour is
expected to be more significant when used in heat transfer
Figure 13 The effect of nozzle length on the normalised axial applications, as surface heat transfer has strong dependency
(U/Ub) and tangential (W/Ub) velocity profiles at the on turbulence intensity (Hoogendoorn, 1977).
nozzle exit plane (see online version for colours)

3.4 Effect of nozzle shape

Figure 15 shows the axial distribution of normalised axial
(U/Ub) and tangential (W/Ub) velocity distributions along
the nozzle length at three different radial locations in the
range r/D = 0–0.49 for different nozzle shapes. For
non-swirling flow, the centreline velocity for the shape S is
found to be constant for the length considered, and it may
remain within the potential core range. For both T and C,
the centreline velocity increases in the converging section to
conserve mass flow rate and reaches to a peak at the exit of
the section. The off-centre locations axial velocity largely
follow the trends of centreline locations, but with reduced
magnitude. For swirling flows, however, a weak peak is
predicted immediately after the tangential flow inlets for all
radial locations and these peaks reduce slowly and merge
all the profiles together before the converging section.
At this location, while the axial velocity remains constant or
slightly decreases with the nozzle length for the shape S,
U/Ub increases sharply with a maximum at the end of the
section for the shape C and a weak peak for the shape T.
Figure 15(b) presents the tangential velocity distribution and
The effect of different lengths of the nozzle for swirling a significant variation in the radial profiles is observed for
flow (Qr = 1) on velocity profiles at the nozzle exit is different shapes. For instance, shape S shows a small
presented in Figure 13, where Figure 13(a) and Figure 13(b) negative velocity followed by a positive peak at x/D ≈ 1.4.
depict the normalised axial velocity and tangential velocity An almost linear increase of tangential velocity from the
profiles, respectively. The velocity profiles, in general, inlet up to the peak at x/D ≈ 1.3 with a weak increase after
show similar radial distributions, but varied magnitude. The the trough at x/D ≈ 2.3 shows for shape T. Such an increase
lowest velocity in the nozzle core region (r/D ≤ 0.3) is in velocity from x/D = 0.7 and peak at x/D = 1.4 is also
predicted for the longest length of the nozzle, i.e., L400. observed. It appears that shapes T and C show similar trends
The highest magnitude is found to be for L280, followed by in velocity distribution before and after the peak.
L0 and L80, respectively. To balance the central deficit,
L400 becomes the maximum near the wall. All tangential
 Effect of design parameters on flow characteristics of an aerodynamic swirl nozzle 259

Figure 14 Contours of turbulent kinetic energy (m2/s2) for and C). The axial velocity profile for S and T shape
different tube length for Qr = 1 (see online version is comparable, both in profile and magnitude, but the
for colours)
profile for C is completely different and resembles to the
non-swirl-flow-like behaviour. A large increment in
velocity magnitude is also predicted. These aspects are
attributed to the contraction of the nozzle, which also forces
the portion of fluids in the shear layer into the central core
region. Likewise axial velocity, tangential velocity for S and
T are similar in nature, with a slight overshoot in shear layer
for S. In contrast, a very strong tangential velocity is
predicted for C shape almost for the whole radius of the
nozzle. The peak tangential velocity for C in the shear layer
is about 1.5 times larger than the other shapes.

Figure 15 Normalised axial and tangential velocity distributions

at different radial locations within the nozzle for
various shapes (see online version for colours)

The effect of different shapes (S, T and C) on turbulent

kinetic energy for swirl flow is shown in Figure 16. For
non-swirling case, very low and uniform turbulent energy
are predicted for S and C, except at the exit plane of C
(x/D = 4.4375), where a strong turbulence is observed at the
wall vicinity. A relatively uniform turbulence distribution
is predicted for the shape T. For swirl conditions, a
significantly different turbulence behaviour is observed for
all shapes. The turbulent kinetic energy becomes stronger
around the centre at x/D = 3. Then it slowly spreads across
the nozzle diameter and tends to be uniform as it travels
along the nozzle length. The higher turbulence is
consistently maintained around the centre for all shapes.
The most uniform turbulence is predicted for T, the
strongest turbulence for C, but with the expense of losing
uniformity near the nozzle wall.
Finally, Figure 17 presents axial and tangential profiles
at the nozzle exit for the effect of different shapes (S, T

Figure 16 Turbulent kinetic energy (m2/s2) contours for both non-swirling and swirling jets (see online version for colours)
260 S.M. Islam et al.

Figure 17 The effect of nozzle shape on the normalised axial twisting is found stronger as the length of the nozzle
(U/Ub) and tangential (W/Ub) velocity profiles at the increases.
nozzle exit plane (see online version for colours)
• The static pressure and turbulent kinetic energy are
found to be significantly influenced, particularly up to
the length of four nozzle diameters, i.e., x/D = 4, by the
number of tangential inlets. More uniform and higher
turbulence at the nozzle exit are seen for 4TP. Again, a
uniform turbulence, with a ring of low turbulence near
the wall is predicted for the whole length of L0 and
L80. For the effect of shape, higher turbulence is
consistently maintaining around the centre. The most
uniform turbulence is predicted for the shape T, and the
strongest turbulence for the shape C with the expense
of losing some uniformity near the nozzle wall.

Acknowledgements
The first author acknowledges the Department of
Electronics and Communication Engineering at Khulna
University of Engineering and Technology, Bangladesh for
part of the computation facility provided during the study
program.

4 Conclusions References
An incompressible turbulent swirling air jet through a Ahmed, Z.U. (2016) An Experimental and Numerical Study of
especially designed nozzle is investigated numerically. In Surface Interactions in Turbulent Swirling Jets, PhD thesis,
this regard, an axial-plus-tangential swirl nozzle is Edith Cowan University, Australia.
considered that is capable to seamless transition from Ahmed, Z.U., Al-Abdeli, Y.M. and Guzzomi, F.G. (2015a)
non-swirling (Qr = 0) to highly swirling (Qr = 1) flow. ‘Impingement pressure characteristics of swirling and
Finite volume-based commercial software is used in the non-swirling turbulent jets’, Experimental Thermal and Fluid
Science, Vol. 68, pp.722–732.
simulation to investigate mean flow and turbulence
Ahmed, Z.U., Al-Abdeli, Y.M. and Matthews, M.T. (2015b)
characteristics. Governing equations are approximated by
‘The effect of inflow conditions on the development of
the RANS equations and turbulence is characterised by the non-swirling versus swirling impinging turbulent jets’,
SST k-ω model. The study examines the effect of number of Computer & Fluids, Vol. 118, pp.255–273.
tangential inlets, lengths of the swirl nozzle and nozzle Ahmed, Z.U., Al-Abdeli, Y.M. and Guzzomi, F.G. (2016a)
shapes on swirl flow behaviours. A summary of the ‘Heat transfer characteristics of swirling and non-swirling
conclusions drawn is given below. impinging turbulent jets’, International Journal of Heat and
Mass Transfer, Vol. 102, pp.991–1003.
• Both the axial velocity and tangential velocity are
Ahmed, Z.U., Al-Abdeli, Y.M. and Guzzomi, F.G. (2016b)
found to be strongly dependent on the number of ‘Corrections of dual-wire CTA data in turbulent swirling and
tangential inlets, lengths of the nozzle and shapes of non-swirling jets’, Experimental Thermal and Fluid Science,
nozzle. A more uniform axial velocity and lower Vol. 70, pp.166–175.
tangential velocity are predicted for 4TP. The smaller Ahmed, Z.U., Al-Abdeli, Y.M. and Guzzomi, F.G. (2017) ‘Flow
nozzle lengths (L0 and L80) show stronger velocity field and thermal behaviour in swirling and non-swirling
profiles. Both the axial and tangential velocity are the turbulent impinging jets’, International Journal of Thermal
largest for contoured nozzle shape (C), followed by the Sciences, Vol. 114, pp.241–256.
tapered (T) shape. Axial velocity for C shape becomes Argyropoulos, C.D. and Markatos, N.C. (2015) ‘Recent advances
more uniform compared to others. on the numerical modelling of turbulent flows’, Applied
Mathematical Modelling, Vol. 39, No. 2, pp.693–732.
• The velocity magnitude becomes larger with the Batten, P., Goldberg, U., Peroomian, O. and Chakravarthy, S.
increase of inlet ports for swirling flow. The boundary (2009) ‘Recommendations and best practice for the current
layer at the nozzle exit is found thinner than state of the art in turbulence modelling’, International
non-swirling counterparts and the extent of boundary Journal of Computational Fluid Dynamics, Vol. 23, No. 4,
layer is independent of the number of ports. The nozzle pp.363–374.
length affects the formation of boundary layer, and Benim, A.C. and Nahavandi, A. (2003) ‘A computational analysis
smaller lengths ensure uniform flow. The intensity of of turbulent swirling flows’, in Hanjalic, K., Nagano, Y. and
Tummers, M.J. (Eds.): Turbulence, Heat and Mass Transfer,
Vol. 4, pp.715–722, Begell House, New York, NY.
 Effect of design parameters on flow characteristics of an aerodynamic swirl nozzle 261

Benim, A.C., Escudier, M.P., Nahavandi, A., Nickson, A.K., Khan, M.H.U. and Ahmed, Z.U. (2019) ‘Fluid flow and heat
Syed, K.J. and Joos, F. (2010) ‘Experimental and numerical transfer characteristics of multiple swirling impinging jets at
investigation of isothermal flow in an idealized swirl various impingement distances’, International Journal of
combustor’, International Journal of Numerical Methods for Thermofluid Science and Technology, Vol. 6, No. 4, pp.1–12,
Heat & Fluid Flow, Vol. 20, No. 3, pp.348–370. pp.19060403:1–12.
Benim, A.C., Iqbal, S., Meier, W., Joos, F. and Wiedermann, A. Lu, P. and Semião, V. (2003) ‘A new second-moment closure
(2017) ‘Numerical investigation of turbulent swirling flames approach for turbulent swirling confined flows’, International
with validation in a gas turbine model combustor’, Applied Journal for Numerical Methods in Fluids, Vol. 41, No. 2,
Thermal Engineering, Vol. 110, pp.202–212. pp.133–150.
Champagne, F.H. and Kromat, S. (2000) ‘Experiments on the Markal, B. (2018) ‘Experimental investigation of heat transfer
formation of a recirculation zone in swirling coaxial jets’, characteristics and wall pressure distribution of swirling
Experiments in Fluids, Vol. 29, No. 5, pp.494–504. coaxial confined impinging air jets’, International Journal of
Chang, F. and Dhir, V.K. (1994) ‘Turbulent flow field in Heat and Mass Transfer, Vol. 124, pp.517–532.
tangentially injected swirl flows in tubes’, International Menter, F. and Egorov, Y. (2010) ‘The scale-adaptive simulation
Journal of Heat and Fluid Flow, Vol. 15, No. 5, pp.346–356. method for unsteady turbulent flow predictions. Part 1: theory
Chigier, N.A. and Beer, J.M. (1964) ‘Velocity and static pressure and model description’, Flow, Turbulence and Combustion,
distributions in swirling air jets issuing from annular and Vol. 85, No. 1, pp.113–138.
divergent nozzles’, Journal of Basic Engineering, Trans. Najafi, A.F., Saidi, M.H., Sadeghipour, M.S. and Souhar, M.
ASME, Series D, Vol. 86, No. 4, pp.788–798. (2005) ‘Numerical analysis of turbulent swirling decay pipe
Dumouchel, C., Bloor, M.I.G., Dombrowski, N., Ingham, D.B. and flow’, International Communications in Heat and Mass
Ledoux, M. (1993) ‘Viscous flow in a swirl atomiser’, Transfer, Vol. 32, No. 5, pp.627–638.
Chemical Engineering Science, Vol. 48, No. 1, pp.81–87. Nouri-Borujerdi, A. and Kebriaee, A. (2012) ‘Simulation of
Escue, A. and Cui, J. (2010) ‘Comparison of turbulence models turbulent swirling flow in convergent nozzles’, Scientica
in simulating swirling pipe flows’, Applied Mathematical Iranica, Vol. 19, No. 2, pp.258–265.
Modelling, Vol. 34, No. 10, pp.2840–2849. Ovink, R., Lamers, A.P.G.G., van Steenhoven, A.A. and
Felli, M., Falchi, M. and Fornari, P. (2008) ‘Impinging swirling jet Hoeijmakers, H.W.M. (2001) ‘A method of correction for the
against a wall: experimental investigation by PIV and high binormal velocity fluctuation using the look-up inversion
speed visualizations’, 14th Int. Symp on Applications of Laser method for hot-wire anemometry’, Measurement Science and
Techniques to Fluid Mechanics, Lisbon, Portugal, 7–10 July. Technology, Vol. 12, No. 8, p.1208.
Fitouri, A., Khan, M.K. and Bruun, H.H. (1994) ‘A multiposition Rose, W.G. (1962) ‘A swirling round turbulent jet’, Journal of
hot-wire technique for the study of swirling flows in vortex Applied Mechanics, Trans. ASME, Series E, Vol. 29, No. 4,
chambers’, Experimental and Thermal Fluid Science, Vol. 10, Vol. 84, pp.616–625.
No. 1, pp.142–151. Saqr, K.M. and Wahid, M.A. (2014) ‘Effects of swirl intensity
Gore, R.W. and Ranz, W.E. (1964) ‘Back flows in rotating fluids on heat transfer and entropy generation in turbulent
moving axially through expanding cross sections’, American decaying swirl flow’, Applied Thermal Engineering, Vol. 70,
Institute of Chemical Engineers Journal, Vol. 10, No. 1, No. 1, pp.486–493.
pp.83–88. Tamrina, K.F., Sheikhb, N.A. and Rahmatullaha, B. (2016)
Gorman, J.M., Sparrow, E.M., Abraham, J.P. and ‘Numerical analysis of swirl intensity in turbulent swirling
Minkowycz, W.J. (2016) ‘Evaluation of the efficacy of pipe flows’, Jurnal Teknologi (Sciences & Engineering),
turbulence models for swirling flows and the effect of Vol. 78, Nos. 5–10, pp.133–141.
turbulence intensity on heat transfer’, Numerical Heat Thomas, B., Ahmed, Z.U., Al-Abdeli, Y.M. and Matthews, M.T.
Transfer, Part B, Vol. 70, No. 6, pp.485–502. (2013) ‘The optimisation of a turbulent swirl nozzle using
Hoogendoorn, C.J. (1977) ‘The effect of turbulence on heat CFD’, Proceedings of the Australian Combustion Symposium,
transfer at a stagnation point’, International Journal of Heat Perth, Australia, 6–8 November.
and Mass Transfer, Vol. 20, No. 12, pp.1333–1338. Toh, I., Honnery, D. and Soria, J. (2010) ‘Axial plus tangential
Hreiz, R., Gentric, C. and Midoux, N. (2011) ‘Numerical entry swirling jet’, Experiments in Fluids, Vol. 48, No. 2,
investigation of swirling flow in cylindrical cyclones’, pp.309–325.
Chemical Engineering Research and Design, Vol. 89, No. 12, Tsai, J.H., Lin, C.A. and Lu, C.M. (1995) ‘Modelling dump
pp.2521–2539. combustor flows with and without swirl at the inlet using
Islam, M.M., Tasnim, S. and Ahmed, Z.U. (2017) ‘Numerical Reynolds stress models’, International Journal of Numerical
study of a swirl nozzle at moderate swirl number’, Methods of Heat & Fluid Flow, Vol. 5, No. 7, pp.577–588.
International Conference on Mechanical Engineering Ullrich, H. (1960) ‘Stromungsvorgange in drall
and Renewable Energy, Chittagong, Bangladesh, brennernmitregelbaren drall und beirotation symmetrischen
20–22 December. freistrahlen’, Forschungs berichte auf dem Gebiete
Jafaria, M., Farhadia, M. and Sedighia, K. (2017) Ingenieurwesens, Vol. 26, No. 1, pp.19–28.
‘An experimental study on the effects of a new swirl Vijiapurapu, S. and Cui, J. (2010) ‘Performance of turbulence
generator on thermal performance of a circular tube’, models for flows through rough pipes’, Applied Mathematical
International Communications in Heat and Mass Transfer, Modelling, Vol. 34, No. 6, pp.1458–1466, 2010.
Vol. 87, pp.277–287. Wang, J., Priestman, G.H. and Tippetts, J.R. (2006) ‘Modelling
Jawarneh, A.M. and Vatistas, G.H. (2006) ‘Reynolds stress model of strongly swirling flows in a complex geometry using
in the prediction of confined turbulent swirling flows’, unstructured meshes’, International Journal of Numerical
Journal of Fluids Engineering, Vol. 128, No. 6, Methods for Heat & Fluid Flow, Vol. 16, No. 8, pp.910–926.
pp.1377–1382.
262 S.M. Islam et al.

Wu, Y., Carlsson, C., Szasz, R., Pengh, L., Fuchs, L. and Nomenclature
Bai, X.S. (2016) ‘Effect of geometrical contraction on vortex
breakdown of swirling turbulent flow in a model combustor’, A Area of the nozzle at the exit plane (πD2/4)
Fuel, Vol. 170, pp.210–225.
Cp  p − p∞ 
Yajnik, K.S. and Subbaiah, M.V. (1973) ‘Experiments on swirling Pressure coefficient  C p =
turbulent flows. Part 1. Similarity in swirling flows’, Journal  1 / 2 ρU b2 
of Fluid Mechanics, Vol. 60, No. 4, pp.665–687. D Nozzle diameter at the exit plane (40 mm)
Yang, Y. and Kaer, S.K. (2012) ‘Comparison of Reynolds k Turbulent kinetic energy (m2/s2)
averaged Navier-Stokes based simulation and large-eddy
simulation for one isothermal swirling flow’, Journal of P Pressure (Pa)
Thermal Science, Vol. 21, No. 2, pp.154–161. Q Volumetric flow rate (m3/s)
Qr Flow ratio (dimensionless), ratio between Qt and
Qtot
Qt Total flow rate through all tangential ports (m3/s)
Qtot Total flow rate through all axial and tangential
ports (m3/s)
r Radial direction in polar coordinate (mm)
Re Reynolds number (dimensionless)
S Swirl number (dimensionless), Wb/Ub
U Time mean velocity component in axial direction
(x-direction) (m/s)
Ub Bulk velocity at the nozzle exit plane (m/s), Qtot/A
Ui or Uj Time mean velocity component (m/s)
Ux Time mean velocity component in x-direction in
Cartesian coordinate (m/s)
Uy Time mean velocity component in y-direction in
Cartesian coordinate (m/s)
Uz Time mean velocity component in z-direction in
Cartesian coordinate (m/s)
ui′ or u′j Fluctuating (rms) velocity components (m/s)
W Time mean velocity component in tangential (θ)
direction (m/s)
x Axial direction in Cartesian or polar coordinate
(mm)
y Abscissa of the impingement plane (mm)
z Ordinate of the impingement plane (mm)
θ Tangential direction in polar coordinate (rad)
ρ Density of air (1.025 kg/m3 at 293 K)
μt Turbulent or eddy viscosity
μeff Effective viscosity of the fluid, sum of molecular
and eddy viscosity
υ Kinematic viscosity of jet (air) (15.16 × 10–6 m2/s,
at 293 K)
ω Specific turbulent dissipation rate
δij Kronecker delta