Journal of the Energy Institute 87 (2014) 372382

Contents lists available at ScienceDirect

Journal of the Energy Institute

journal homepage: http://www.journals.elsevier.com/journal-of-the-energyinstitute

Design and study on performance of axial swirler for

annular combustor by changing different design
parameters
Bhupendra Khandelwal, Dong Lili, Vishal Sethi*
Dept. of Power and Propulsion, School of Engineering, Craneld University, MK43 0AL, UK

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 15 January 2013
Accepted 23 July 2013
Available online 29 March 2014

Recent technologies have been introduced for gas turbine engine to meet with stringent emission regulations. One of the technologies is to introduce recirculation in the combustion zone to control the
residence time and mixing by help of swirling ow. Effect of variation in geometric parameters and inlet
mass ow of swirler have been examined in this study by help of CFD. Detailed design methodologies
have been proposed in this study to design a series of axial swirler with different vane angles and vane
numbers. Substantial variation in swirler performance has been observed by changing vane angle, vane
number and mass ow. Four different types of axial and radial velocity proles have been observed.
Turbulence distribution pattern shows double peaks at all positions and reduces with increasing axial
distance.
2014 Energy Institute. Published by Elsevier Ltd. All rights reserved.

Keywords:
Swirling ow
Swirler
Design methodology
Gas turbine combustor
Combustion chamber

1. Introduction
With the growth in aviation industry and protocols to reduce emissions, there is a need of continuous improvements in technology for
gas turbine engines. A modern combustor must satisfy various requirements for being able to deliver the desired performance. Combustor
should comply with the requirements including capability of easier ignition and stable operation over a wide range of work conditions,
complete combustion, appropriate temperature traverse at outlet and low pressure loss. It is of great importance that all necessary requirements of combustor can be fullled while keeping minimum size, weight, cost, while maintaining operating life. Researchers around
the world are constantly developing new technologies and concepts to improve the performance of combustors and decrease the emissions
from the combustors [1,2].
The swirler is generally mounted in dome of combustors circumferentially and plays an important role in combustion design [36].
Swirlers have three basic functions. Firstly, the swirler can create a stable low pressure central recirculation zone that have merits of stabilized and anchored ame close to swirler exit, enhancing mixing between fuel and air and acting as a continuous source of ignition.
Secondly, the swirling ow produced by swirler can reinforce the secondary holes recirculation, further increasing the turbulence, which
benets stability of the secondary holes. Finally, the air through the swirler can form a lm cooling layer to cool the rst section of liner close
to the injectors. In order to full all three functions, the swirler must impart a high radial component, since air which still has an axial
component reduces the secondary recirculation. In all, the swirler plays a vital role in gas turbine combustors to improve ame stabilization,
fuel air mixing and emissions. High turbulence intensities lead to better atomisation for the liquid fuel sprayed from the injectors, which
benets the combustion stability of the combustor [7]. The length of recirculation zone is generally twice bigger as diameter of swirler if
operating on diffusion without any effects of secondary airow [8,9].
Counter-rotating swirler can generally lead to ne atomization than single or co-rotating swirler. However co-rotating swirler can
generate less noisy ames than single and counter-rotating swirler. So for some civil application the co-rotating swirler are preferred.
Swirlers are also classied according to their vane types, at vane and curve vane. The angles of swirler vane are generally between 45
and
70
[9]. However, for the curve vane swirler, the angles at inlet and outlet are different. Although curve vane swirler have much stronger

* Corresponding author. Tel.: 44 01234 754765.

E-mail address: v.sethi@craneld.ac.uk (V. Sethi).
http://dx.doi.org/10.1016/j.joei.2014.03.022
1743-9671/ 2014 Energy Institute. Published by Elsevier Ltd. All rights reserved.

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

Nomenclature
A
AF
CFD
C
D
FCV
M
NOx
P
q
r
S
SN
TET
Z

area (unit: m2)

air to fuel ratio
computational uid dynamics
vane chord
diameter (unit: m)
fuel caloric value
mass ow (unit: kg/m3)
oxides of nitrogen
pressure (unit: Pa)
mass ow (unit: kg/m3)
radius (unit: m)
vane spacing
swirl number
turbine entry temperature (unit: K)
vane height

Greek symbols
F
equivalence ratio

q
s

373

vane angle
solidity

Subscripts
3
compressor exit
4
turbine entry
c
core
cd
dome cooling
cs
swirling air
d
dome
f
fuel
h
hub
hub
hub
p
primary zone
ref
reference
s
swirler
sw
swirler
t
total
v
vane

turbulence intensity than at vane swirler, pressure losses are relatively greater and manufacturing is more difcult, which consequently
increase the cost and are comprehensively applied in bigger engines. Therefore the at vane swirler is usually preferable for small engines
due to simplicity and ease of manufacturing.
Several studies have been carried out on understanding working of a swirler [3,7,1018]. To the authors knowledge there is no study
available in literature which explains systematic design procedure of swirler and performance of varying designs. In this study a series of
axial swirler with different vane angles and vane numbers were designed by using pressure drop methods according to the procedure
developed and proposed in this study. A computational and analytical study is done to check the effect of the geometric parameters on ow
characteristics and the effect of mass ow on pressure drop coefcient. Comparison of different CFD models has also been done to suggest
the best model for computational analysis of swirler.
2. Design methodology of swirler
After a comprehensive literature review authors found that there is no literature available on a sound theoretical method of designing an
axial swirler. Therefore the empirical methods are used in designing swirler in practice. Typical parameters of swirler need be determined as
shown in Fig. 1 when designing axial swirler.
Considering the geometrical parameters in Fig. 1 and performance of combustors, there are two basic methods to calculate the geometry
of axial swirler. One is ow drag coefcient method and another is pressure loss coefcient method. The two methods essentially differ in
calculating the effective area of swirler, in one method it is determined by ow drag coefcient and in the other one it is determined by
pressure loss coefcient.
3. Design of axial swirler
Design of swirler is based on preliminary design of combustors. Therefore design of axial swirler is carried out after nishing the preliminary design of combustor, which provides combustor performance parameter. Model engine [15] gives total pressure at the compressor
exit (P3) as 1644.9 kPa, total temperature at the compressor outlet T3 as 775 K, turbine entry temperature (TET or T4) as 1450 K, core mass
ow (MT) as 10.67 kg/s and turbine cooling bleed factor (Ccooling) as 12% and total pressure loss of combustors (DP) as 6%. It gives some
geometric dimensions of the combustors including reference area (Aref) as 0.8186 m2, reference diameter (Dref), chamber depth (H) as
57 mm, length (L) as 180 mm and so on.
The rules of designing the swirler can be summarized as no see through, suitable strong recirculation zone, required mass ow, matching
with fuel injectors. For combustors, the mass ow is more important than the recirculation zone, the design of swirler must be carried out
for the required mass ow for appropriate air distribution of combustors. Secondly the designed swirler must be suitable for the combustor
dimensions and at the same time it must match the fuel injectors.
In this section the axial swirler will be designed according to the given combustors [15] performance parameters and geometric dimensions. Firstly on the basis of preliminary design results of combustor the mass ow through the swirler need be calculated according to
the air distribution, and then the effective area is determined by the pressure loss coefcient method. At the end, sizing of the swirler is
carried out. Therefore the methods used in this section are both theoretical and empirical.
3.1. Procedure of calculating axial swirler mass ow
Firstly the air owing through the core mass ow (Mc) is calculated according to the total mass ow (MT) and turbine cooling bleed factor
(Ccooling) using Eq. (1).

374

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

Fig. 1. Geometric parameters of an axial swirler.




Mc MT  1  ccooling

(1)

Secondly the air to fuel ratio (AF) at the design point, cruise condition of turboprop engine is estimated using Eq. (2), which reects the
energy balance relationship of a typical gas turbine engine. In this equation, the specic heats Cpg and Cpa are found in thermodynamic
properties table of air and have the value of 1210 J/kg/K and 1093 J/kg/K respectively. Fuel Caloric Value (FCV) is 43 MJ here and hence the
air fuel ratio (AF) is 45.5.


 

AF FCV  Cpg TET = Cpg TET  Cpa T3

(2)

Thirdly the fuel mass ow (Mf) needs to be again calculated by using Eq. (3), although the fuel mass ow has been given in overall
performance calculation, because here it is calculated from the prospective of combustion. The core mass ow (Mc) and air fuel ratio (AF)
have been obtained by above two steps and so the fuel mass ow is 0.21 kg/s.

Mf Mc =AF

(3)

Fourthly some intermediate variables during calculation need be assumed because designing combustors cannot completely be theoretical, it is still said to be black art by many researchers [2,4]. These assumed variables include ratio of dome cooling mass ow (Mcd) to core
mass ow, Mcd/Mc, and ratio of air mass ow for atomizing fuel to fuel mass ow Mca/Mf if for air blast fuel jet. In order to ensure sufcient
and effective atomization the ratio of air mass ow for atomizing fuel to the fuel mass ow Mca/Mf has the value of 23, in this study value of
3 has been used. For annular combustors, the air for cooling dome is around 8% for older engine and 1015% for modern engine, here the
value of 15% is chosen [16].
Fifthly the equivalence ratio in primary zone (Fp) should be assumed in order to calculate swirling air mass ow (Mcs). For a conventional
combustors, the equivalence ratio at primary zone ranges from 0.9 to 1.1. In this range there are relatively maximum ame temperatures and
consequently the emission pollutants such as NOx may be observed. If using rich burn quick quench technology, the equivalence ratio is
between 1.4 and 1.5. At this range the ame temperatures are lower than that of the conventional combustors and the level of emission are
lower. Therefore here the rich equivalence ratio is selected and has the value of 1.5.
Finally once the equivalence ratio is selected in primary zone, the swirling air mass ow can be calculated by using Eq. (4). AFth is the
stoichiometric air fuel ratio and the value is 14.7 (Table 1).

Mcs

AFth

fP

 Mf  Mca  Mcd

(4)

3.2. Calculation of swirler area

The pressure loss coefcient method is used in this study to calculate the area of axial swirler. In general pressure loss of swirler should be
6070% of total pressure loss through combustor. Therefore the pressure loss of swirler is here assumed as 65% of the total pressure loss
through the combustor. Pressure drop coefcient of swirler DPsw/qref can be obtained by pressure loss of swirler over reference dynamic
head. Pressure loss relationship of swirler is given in Eq. (5).

DPs
qref

"
k

Aref
As

2

 #


Aref 2 Msw
sec q 
Al
MT
2

(5)

Vane shape factor k is 1.3 for at vane and 1.15 for curved vane, which indicates that pressure loss of curved vane is less than that of at
vane swirler. In Eq. (5), all items except Aref/As are known, so Aref/As can be calculated and the area of swirler (As), can be obtained.

Table 1
Design results of air distribution (unit: kg/s).
MT

Mf

Md

Mp

Vp

Mcs

9.53

0.21

2.45

7.08

1.25

1.02

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

375

In typical combustors, swirl induced by recirculation is augmented by the airow through the primary hole; a certain amount of air
should be involved in mass ow of swirler for calculating area of swirler. Out of the quantity of airow that enters in the primary zone of
combustor through primary holes, 3070% takes part in the recirculation [8]. Consequently the area of swirler that can be calculated from
Eq. (5) need be corrected with augmented mass ow. Here allowance of 50% more airow has been used.
3.3. Swirler sizing
The ow area of swirler (Asw) is calculated by both pressure loss method and geometric structure. The area of swirler can be calculated by
Eq. (6) if using the geometric dimensions.

Asw

p

4


D2t  D2h  0:5ndDt  Dh

(6)

The hub diameter (Dh) is calculated by the outer diameter of fuel injector that has been roughly estimated to a value of 14 mm. The tip
diameter of swirler (Dt) should be calculated from Eq. (6) and have the value of 28 mm. The hub tip ratio is the ratio of the hub diameter to
the tip diameter and here is 0.5, which have better fall within the range of 1/32/3. The number of swirler vanes n commonly ranges from 8
to 16. The thickness of swirler vane is from 0.7 to 1.5 mm for the consideration of manufacturing. During designing, the thickness of vane for
all swirlers is considered as 1 mm and the number of swirler vanes is different from each swirler. After calculating these dimensions, the
swirl number SN need be checked by Eq. (7). If the swirler number SN is smaller than 1.0 then the number and the thickness need be
changed repeatedly until the swirl number of around 1.0 is achieved. This is due to the reason that the swirl number for gas turbine
combustors generally ranges from 1.0 to 1.5. For the annular swirler with constant vane angle q, the expression is given by Eq. (7) [5].

SN

2 1  Dh =Dt 3
tanq
3 1  Dh =Dt 2

(7)

Dhub/Dsw is dened as hub tip ratio of swirler and the second item in above equation generally have the value between 1 and 1.5.
For the curved vane the expression is different from the constant vane swirler due to the different angles at trailing and leading edge of
vane. Assuming the axial velocity component is uniform over cross section after swirler vane exit, the swirl number of curve vane swirler
with vane exit angle qm can be represented as Eq. (8) [13].

SN

D2t D2h
2D2t

tanqm

(8)

The vane space, s, is calculated by Eq. (9).

s pDt Dh =2n

(9)

The vane chord (c) can be obtained using Eq. (10) if assuming solidity of vanes is (sv)

c sv s

(10)

The length of swirler is decided by no see through rules. The NX Unigraphics software is used in this study to assist with the length
calculation. NX Unigraphics [19] is an advanced CAD/CAM/CAE software package developed by Siemens PLM Software.
3.4. Results of axial swirler design
The ve swirlers have been designed by procedure described above. One of them is designed for at vanes and other four were designed
for curve vanes, as illustrated in Fig. 2. Details of swirlers designed are mentioned in Table 2. The hub diameter of swirler is typically
calculated by the fuel injector. In this study a fuel injector used regularly in engineering is selected which has an outer diameter of 14 mm.
Therefore the hub diameter of swirler is 12 mm due to the thickness of wall is 1 mm. The outer diameter of swirler is calculated by the sizing
procedure and has a value of 28 mm, but the real outer diameter of swirler is 30 mm with wall thickness of 1 mm.

Fig. 2. 3-D curve vane (45
, 52
, 60
) swirler.

376

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

Table 2
Design results of axial swirler.
Name

Vane angle

Vane number

Vane type

Hub dia. (mm)

Tip dia. (mm)

Swirler number

Block factor

S45
S52
S60
S60-N
S45-T

45
52
60
60
45

8
8
8
10
8

Curve

14

28

0.625
0.81
1.08
0.81
1.34

0.257
0.295
0.364
0.455
0.257

Flat

Swirler number of swirler with helical prole can be calculated by the tip diameter, hub diameter and tip vane angle using the Eq. (8). The
block factor of swirler can be calculated after hub and tip diameters are calculated through using Eq. (11) [14], results are listed in Table 2.

j nd=2pRh cos q

(11)

3.5. Validation of CFD model

Validation is carried out by simulating ow of an axial swirler in CFD to obtain axial and radial velocity distribution, which is then
compared with experiment results conducted by Fu et al. [13]. Fu has carried out an experimental investigation on aerodynamic characteristic of conned, non-reacting swirling ow induced by the swirler with different vane angles [14].
In this study, one of the experimental model is validated, namely curved vane axial swirler. The single swirler has six curved vanes with
inner diameter 9 mm and outer diameter 22 mm. The swirler cup has been designed with a tip vane angle of 60
and swirl number of 1. In
addition a convergentdivergent venturi is installed at the exit of swirler.
Calculation connement section is considered as a cuboid with section of 50.8 mm  50.8 mm and length of 304.8 mm. The Grid quality
has been examined carefully so as to ensure no skew cell exists to inuence numerical convergence. The skewed sized is less than 0.73 and
the aspect ratio is less than 3.67. Five turbulence models were validated to compare with the test data or each other. The inlet is setup to
Velocity inlet of 90 m/s that is calculated by given test condition. At the inlet turbulence strength is setup with default value of 10 and
turbulence scale length of 22 mm, which is the outer diameter of swirler. The Reynolds stress model (RSM) is a higher level, elaborate
turbulence model [20]. It is usually called a Second Order Closure. One of the more successful recent developments is the realizable K
Epsilon model developed by Shih et al. [21]. This model contains a new transport equation for the turbulent dissipation rate [22]. Also, a
critical coefcient of the model, Cm, is expressed as a function of mean ow and turbulence properties, rather than assumed to be constant as
in the standard model. This allows the model to satisfy certain mathematical constraints on the normal stresses consistent with the physics
of turbulence (realizability). The concept of a variable Cm is also consistent with experimental observations in boundary layers.
Fig. 3 shows axial velocity distribution along axial direction at z 120. The axial velocity distributions simulated by realizable turbulence
model are good agreement with those of experiment at positions z 3, z 12, z 48 and z 120 considered in this study.
Fig. 4 shows the radial velocity distribution along radial direction at z 48 mm. It can be observed from Fig. 4 that the k realizable
turbulence model once again presents the desirable simulation results. It is also observed that k standard turbulence model gives the
relative desirable computational results at positions of z 12 and 120 respectively while at the z 3 and z 48 positions the computational
errors are comparatively large. The RSM turbulence model gives the worst simulation results that the ow unsymmetrical occurs in each
position and have the irregular proles. In particular, at position of z 120, it gives the completely different trends from simulation results of
the other three models.
Four turbulence models provided in FLUENT software are investigated to check whether they are suitable for validated swirler model and
have enough precision to deal with the swirling ow through comparing the simulation results with the experimental data. The results
indicate that renormalization group (RNG) turbulence model gives worst simulation results, while k realizable turbulence model has best
simulation results. For axial velocity magnitude distribution along the axial direction, four turbulence models nearly have desirable proles
and values. The axial velocity and radial velocity distribution along radial direction at different positions illustrated RNG and RSM turbulence
models seems to be sensitive to the ow eld tangential variation and hence there are some deformation ow. However, the k realizable
turbulence model is possibly much easier to deal with the swirling ow especially at the time that the comparable effects are investigated.

Fig. 3. Axial velocity distribution at z 120 mm.

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

377

Fig. 4. Radial velocity distribution at z 48.

From above analysis it can be concluded that the incorrect selection of turbulence models may causes inaccurate simulation results.
Therefore model should be carefully selected when performing a CFD simulation.
4. Study of axial swirler ow
The effects of vane angle on aerodynamic characteristic of swirler are investigated by comparing the ow characteristic of swirler S45,
S52 and S60. A sketch of computational simulation domain is shown in Fig. 5. All the conditions of venturi and connement used in
computational simulation models are same, differences are there in swirler models only. For a real combustion system, a passage such as
venturi is usually combined with the swirler to form a swirler cup for better atomization and recirculation zone. This passage could be in
form of several proles that include at passage, divergent passage or convergencedivergence nozzle as well as real venturi. In study done
by Fu et al. [13] convergencedivergence nozzle is used to replace venturi. It is recommended that the optimum half angle of convergence
divergence nozzle should be about 30 and length ranges from one to two times of swirler exit diameter. Therefore the convergence
divergence nozzles used in simulation models have half angle of 30 and length of 14 mm.
The density of the air is set to be an incompressible ideal gas due to Mach number less than 0.3. The value of specic heat, thermal
conductivity and viscosity is obtained from the air properties temperature table according to the inlet temperature of Group Design Project
[15] combustor. All of items at solution controls are all selected to be second order and under-relaxation factors were adjusted down to a
smaller value of 0.6 except that energy factor is 0.9 and pressure relaxation factor is 0.1. In order to investigate the effect of mass ow on
swirler exit ow different inlet velocities were studied. Effects of vane number were also investigated through comparing ow characteristic
created swirler S60 and S60-N.
5. Results and discussion
5.1. The effect of vane angle on ow characteristic of swirler
Fig. 6 shows the variation of axial velocity as increasing axial distance for swirler S45, S52 and S60. S45 is the case with 45
S52 is with 52

and S60 is the case with 60
. It can be seen that the axial velocities initially reduce with the increase of vane angle and at the end tend to a
constant value of main ow. It is also observed that for most of the axial distance cases with 45
angle, high axial velocity is seen. This also
shows that by increasing the vane angle of swirl axial velocity decreases which would eventually increase recirculation.
Fig. 7 shows turbulence intensity characteristics for swirler with different vane angle along the axial direction. It is observed that for all
the three cases under consideration turbulent intensity decreases with axial distance. This is due to the fact that turbulence intensity
decreases with distance due to losses. Case with 45
angle is showing lowest turbulence intensity whereas; case with 60
is showing high
turbulence intensity at majority of the sections. Similar observations are also observed for turbulent kinetic energy and turbulent viscosity
ratio. They are all decreased as the axial distance increases and the increased as the increase of vane angle. That means that the larger vane
angle produces larger turbulence.

Fig. 5. Sketch of swirler domain.

378

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

Fig. 6. Variation of axial velocity with vane angle.

Fig. 7. Variation of turbulent intensity with vane angle.

Fig. 8 shows the relationship between vane angle and pressure drop coefcient. It is observed that by increasing vane angle pressure drop
coefcient also increases. It can be easier to understand that increasing vane angle increases the chance to separate from the vane wall
surface, and hence the secondary loss increases and further the pressure drop coefcient increase. For a case with 45
vane angle pressure
drop coefcient is 7% whereas when the vane angle is increased to 60
pressure loss coefcient increases to 40%. This is a considerable
amount which would hamper performance of combustor substantially.
5.2. Effect of changing number of vanes on ow characteristics of swirler
Effect of changing the number of vanes has been studied by increasing the number of vanes in case S52 with 52
angle. Number of vanes
has been increased from 8 to 10 in the new case and it has been named as S5210. Fig. 9 shows variation in axial velocity by increasing the
number of vanes along the axial axis. It is observed that the axial velocity decreases with the increase of vane number.
Fig. 10 is showing variation in turbulent kinetic energy by increasing the number of vanes along the axial axis. It is observed that turbulent kinetic energy increases marginally with increase in number of vanes. Similar effect as of turbulent kinetic energy is also observed for

Fig. 8. Variation of pressure loss coefcient with vane angle.

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

379

Fig. 9. Variation of axial velocity with vane number.

turbulent intensity and turbulent viscosity ratio. Decrease in axial velocity and increase in turbulence with increase in number of vanes will
also lead to increased recirculation and better ammability limits.
5.3. Effect of inlet mass ow on pressure loss of swirler
One of the most important performance parameters for a curved vane type swirler is the pressure drop-mass ow characteristic, namely
what pressure drop has been occurred through the swirler at a certain mass ow. Here swirler S45 with 45
angle has been selected to
investigate pressure drop across swirler by changing mass ow. Pressure drop is also the static pressure difference between inlet and outlet
at swirler and has been normalized by swirler inlet dynamic head. The pressure drop coefcient can be consequently expressed as Eq. (12).

PDC

pi  po
0:5rV 2

(12)

Fig. 11 shows this characteristic of case S45, which is an example of strong swirl ow. It can be seen that the pressure drop decreases with
the increase of mass ow for given vane angle, vane number, tip and hub diameter. Pressure loss coefcient is 27% when the mass ow rate is
0.022 kg/s which decreased to 2.5% when the mass ow is increased to 0.039 kg/s. One of the possible reason for this is blockage in swirler
passage due to separation has been avoided and consequently reduces the frictional loss between vanes. It can be also seen that the pressure
loss coefcient seems to tend towards a constant value with increasing the mass ow. Therefore it appears to deduce that above a certain
mass ow the pressure drop coefcients are not dependent on the mass ow.
5.4. Velocity proles
The axial and radial velocities at different positions along radial direction have also been studied. Fig. 12 shows axial velocity prole at the
near exit or near eld decay part that approximately ranges from z 3 mm to z 9 mm. In this region the velocity proles of double peak
curves are observed and the velocities decay with the increase of axial distance.
The near exit or near eld similarity ow region is illustrated in Fig. 13, in which the axial velocities nearly dont change with the axial
positions and have the U shape proles. The peak velocity reduces from 38 m/s to25 m/s, approximately 0.66 times the peak velocity at z 3
position. The centreline reverse ow velocity is about 25 m/s.

Fig. 10. Variation of turbulent kinetic energy with vane number.

380

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

Fig. 11. Variation of pressure drop coefcient with mass ow.

Fig. 12. Axial velocity proles at near exit zones.

Fig. 13. Axial velocity proles at near exit similarity zones.

Fig. 14. Radial velocity proles at near exit zones.

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

381

For radial velocities along radial direction at different axial positions there are mainly four shapes, as illustrated in Figs. 14 to 16. It can be
seen that the radial positions of peak in the radial velocity nearly move outwards in downstream direction as a general trend in the ow
eld.
Fig. 14 shows that the radial velocity peaks along radial direction continuously decrease at near exit zone. It can also be seen that the
recirculation heights along radial direction increase gradually with the increase of axial distance. Fig. 15 describes the radial velocity proles
at near exit similarity zone. It can be seen in this region that the formation of free vortex appears to take place at the near wall. The peak
values of free vortex continuously decrease along the direction towards downstream but the radial positions of radial velocity peak nearly
keep unchanged.
Fig. 16 shows the radial velocity proles at the second decay zone, in which the radial peak velocities reduce slowly and the radial
position of radial velocity peak initially unchanged and then move inward over z 46 mm. This indicates that a reverse ow at radial
velocity direction is occurred at this position where is the maximum height of recirculation zone.
Fig. 17 gives the distributions of radial velocity at different axial positions. It can be seen that the wake ow region starts very early at
axial position. Although radial velocity still keeps the trend of second decay zone the radial velocity proles are irregular.

Fig. 15. Radial velocity proles at near exit similarity zone.

Fig. 16. Radial velocity proles at second decay zone.

Fig. 17. Radial velocity proles at far eld.

382

B. Khandelwal et al. / Journal of the Energy Institute 87 (2014) 372382

6. Conclusions
The present works in this paper mainly focus on the procedure of designing axial swirler, and further investigation on ow characteristic
of swirling ow generated by vane axial swirler by help of computational methods. Moreover the effects of geometric parameters, namely
vane angle and vane number and ow are also reported in the study.
Swirler design is carried out on the basis of semi-empirical equations. The design procedure using pressure drop method include three
steps namely, mass ow calculation through air distribution, effective area calculation through pressure drop and sizing from the geometric
relationship. The non-reacting swirling ow can be simulated very well through using k realizable turbulence model provided in FLUENT
that has been validated in this study.
Axial reverse ow velocity, turbulence along axial direction and pressure drop increases with the increase in angle of vanes. Namely the
better fuel/air mixing and recirculation structure are obtained if the vane angle of swirler is larger but at the penalty of the pressure drop of
combustors. Increasing the vane number increases the turbulence energy and decreases the axial ow velocity. For a given swirler,
increasing mass ow at inlet decreases the pressure drop through the swirler and the pressure drop seems to keep constant if the mass ow
is increased above a critical value. Thus the mass ow through the swirler should be adjusted carefully when designing swirler. The results
show that the axial velocity and radial velocity proles have four different types of proles.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]

B. Khandelwal, Y. Li, P. Murthy, V. Sethi, R. Singh: Proc. of ASME Turbo Expo 2011, GT2011-46845, pp. 293298.
B. Khandelwal, M. Yan, G. Hegde, V. Sethi, R. Singh, SAE Aerotech 2011, Toulouse, France, October 2011. SAE paper 2011-01-2562.
M. Abd-Rabbou, Craneld Thesis, Craneld, Bedfordshire, UK, 1977.
J.N. Murthy, PHD Thesis, Craneld University, 1988.
P.J. Stuttaford, P.A. Rubini, ASME J. Eng. Gas. Turbines Power 119 (3) (1997) 546552.
H. Mongia, in: 39th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, 2023 July 2003. Huntsville, Alabama, AIAA 2003-4495.
E. Kilik, PHD Thesis, Craneld University, 1976.
Sawyers Gas Turbine Engineering Handbook, third ed., vol. 1, 1985.
J.M. Beer, Combustion Aerodynamics, Applied Science Publishers Ltd, 1972.
S.A. Beltagui, N.R.L. Maccallum, J. Inst. Fuel 49 (1967) 183193.
Y.Q. Fu, S.M. Jeng, in: 44th Aerospace Sciences Meeting & Exhibit, Reno, 2006. AIAA 2006-545.
E. Kilik, in: 21st AIAA/SAE/ASME/ASEE Joint Propulsion Conference, Monterey, 1985. AIAA Paper 1985-1103.
Y.O. Fu, S.M. Jeng, T. Robert, in: ASME Turbo Expo 2005, Reno, Nevada, USA, June 2005, pp. GT200568728.
E. Kilik, in: 23rd Aerospace Sciences Meeting, Reno, NV, 1985. AIAA Paper 1985-1087.
MSc Thermal Power 2010: Group Design Project, Craneld University, UK, March 2010.
A.H. Lefebvre, Gas turbine combustion, second ed., Taylor & Francis Group, Great Britain, 1998.
David G. Lilley, J. Propuls. Power 15 (2) (1999).
G.B. Hegde, B. Khandelwal, V. Sethi, R. Singh, in: ASME Turbo Expo 2012, Copenhagen, Denmark, June 2012. GT2012-69215.
http://www.plm.automation.siemens.com (accessed 28.06.13).
B.E. Launder, G.J. Reece, W. Rodi, J. Fluid Mech. 68 (3) (1975) 537566.
L.H. Shih, J.R. Koseff, G.N. Ivey, J.H. Ferziger, J. Fluid Mech. 525 (2005) 193214.
http://www.kxcad.net/star-ccm/online/139-kepsilonturbulence-07.html (accessed 28.06.13).