JOEI33_proof ■ 3 April 2014 ■ 1/11

Journal of the Energy Institute xxx (2014) 1–11

Contents lists available at ScienceDirect

Journal of the Energy Institute

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

1 Design and study on performance of axial swirler for annular

2
3 combustor by changing different design parameters
4
5 Q4 Bhupendra Khandelwal, Dong Lili, Vishal Sethi*
6
Dept. of Power and Propulsion, School of Engineering, Cranfield University, MK43 0AL, UK
7
8
9
10 a r t i c l e i n f o a b s t r a c t
11
Article history: Recent technologies have been introduced for gas turbine engine to meet with stringent emission reg-
12 Received 15 January 2013 ulations. One of the technologies is to introduce recirculation in the combustion zone to control the
13 Accepted 23 July 2013 residence time and mixing by help of swirling flow. Effect of variation in geometric parameters and inlet
14 mass flow of swirler have been examined in this study by help of CFD. Detailed design methodologies
15 Keywords: have been proposed in this study to design a series of axial swirler with different vane angles and vane
16 Swirling flow
numbers. Substantial variation in swirler performance has been observed by changing vane angle, vane
Combustor
17 number and mass flow. Four different types of axial and radial velocity profiles have been observed.
Swirler
18 Design methodology Turbulence distribution pattern shows double peaks at all positions and reduces with increasing axial
19 Gas turbine combustor distance.
20 Q1 Combustion chamber Ó 2014 Published by Elsevier Ltd on behalf of Energy Institute.
21
22
23
24
1. Introduction
25
26
With the growth in aviation industry and protocols to reduce emissions, there is a need of continuous improvements in technology for
27
gas turbine engines. A modern combustor must satisfy various requirements for being able to deliver the desired performance. Combustor
28
should comply with the requirements including capability of easier ignition and stable operation over a wide range of work conditions,
29
complete combustion, appropriate temperature traverse at outlet and low pressure loss. It is of great importance that all necessary re-
30
quirements of combustor can be fulfilled while keeping minimum size, weight, cost, while maintaining operating life. Researchers around
31
the world are constantly developing new technologies and concepts to improve the performance of combustors and decrease the emissions
32
from the combustors [1,2].
33
The swirler is generally mounted in dome of combustors circumferentially and plays an important role in combustion design [3–6].
34
Swirlers have three basic functions. Firstly, the swirler can create a stable low pressure central recirculation zone that have merits of sta-
35
bilized and anchored flame close to swirler exit, enhancing mixing between fuel and air and acting as a continuous source of ignition.
36
Secondly, the swirling flow produced by swirler can reinforce the secondary holes recirculation, further increasing the turbulence, which
37
benefits stability of the secondary holes. Finally, the air through the swirler can form a film cooling layer to cool the first section of liner close
38
to the injectors. In order to fulfil all three functions, the swirler must impart a high radial component, since air which still has an axial
39
component reduces the secondary recirculation. In all, the swirler plays a vital role in gas turbine combustors to improve flame stabilization,
40
fuel air mixing and emissions. High turbulence intensities lead to better atomisation for the liquid fuel sprayed from the injectors, which
41
benefits the combustion stability of the combustor [7]. The length of recirculation zone is generally twice bigger as diameter of swirler if
42
operating on diffusion without any effects of secondary airflow [8,9].
43
Counter-rotating swirler can generally lead to fine atomization than single or co-rotating swirler. However co-rotating swirler can
44
generate less noisy flames than single and counter-rotating swirler. So for some civil application the co-rotating swirler are preferred.
45
Swirlers are also classified according to their vane types, flat vane and curve vane. The angles of swirler vane are generally between 45
and
46
70
[9]. However, for the curve vane swirler, the angles at inlet and outlet are different. Although curve vane swirler have much stronger
47
turbulence intensity than flat vane swirler, pressure losses are relatively greater and manufacturing is more difficult, which consequently
48
49
50
51 * Corresponding author. Tel.: þ44 01234 754765.
E-mail address: v.sethi@cranfield.ac.uk (V. Sethi).
52
53
http://dx.doi.org/10.1016/j.joei.2014.03.022
54 1743-9671/Ó 2014 Published by Elsevier Ltd on behalf of Energy Institute.

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 2/11

2 B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11

1
2 Q2 Nomenclature q vane angle
3 s solidity
4 A area (unit: m2)
5 AF air to fuel ratio Subscripts
6 CFD computational fluid dynamics 3 compressor exit
7 C vane chord 4 turbine entry
8 D diameter (unit: m) c core
9 FCV fuel calorific value cd dome cooling
10 M mass flow (unit: kg/m3) cs swirling air
11 NOx oxides of nitrogen d dome
12 P pressure (unit: Pa) f fuel
13 q mass flow (unit: kg/m3) h hub
14 r radius (unit: m) hub hub
15 S vane spacing p primary zone
16 SN swirl number ref reference
17 TET turbine entry temperature (unit: K) s swirler
18 Z vane height sw swirler
19 t total
20 Greek symbols v vane
21 F equivalence ratio
22
23
24
25
26 increase the cost and are comprehensively applied in bigger engines. Therefore the flat vane swirler is usually preferable for small engines
27 due to simplicity and ease of manufacturing.
28 Several studies have been carried out on understanding working of a swirler [3,7,10–18]. To the author’s knowledge there is no study
29 available in literature which explains systematic design procedure of swirler and performance of varying designs. In this study a series of
30 axial swirler with different vane angles and vane numbers were designed by using pressure drop methods according to the procedure
31 developed and proposed in this study. A computational and analytical study is done to check the effect of the geometric parameters on flow
32 characteristics and the effect of mass flow on pressure drop coefficient. Comparison of different CFD models has also been done to suggest
33 the best model for computational analysis of swirler.
34
35 2. Design methodology of swirler
36
37 After a comprehensive literature review author’s found that there is no literature available on a sound theoretical method of designing an
38 axial swirler. Therefore the empirical methods are used in designing swirler in practice. Typical parameters of swirler need be determined as
39 shown in Fig. 1 when designing axial swirler.
40 Considering the geometrical parameters in Fig. 1 and performance of combustors, there are two basic methods to calculate the geometry
41 of axial swirler. One is flow drag coefficient method and another is pressure loss coefficient method. The two methods essentially differ in
42 calculating the effective area of swirler, in one method it is determined by flow drag coefficient and in the other one it is determined by
43 pressure loss coefficient.
44
45 3. Design of axial swirler
46
47 Design of swirler is based on preliminary design of combustors. Therefore design of axial swirler is carried out after finishing the pre-
48 liminary design of combustor, which provides combustor performance parameter. Model engine [15] gives total pressure at the compressor
49 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
50 flow (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
51 geometric dimensions of the combustors including reference area (Aref) as 0.8186 m2, reference diameter (Dref), chamber depth (H) as
52 57 mm, length (L) as 180 mm and so on.
53 The rules of designing the swirler can be summarized as no see through, suitable strong recirculation zone, required mass flow, matching
54 with fuel injectors. For combustors, the mass flow is more important than the recirculation zone, the design of swirler must be carried out
55 for the required mass flow for appropriate air distribution of combustors. Secondly the designed swirler must be suitable for the combustor
56 dimensions and at the same time it must match the fuel injectors.
57 In this section the axial swirler will be designed according to the given combustor’s [15] performance parameters and geometric di-
58 mensions. Firstly on the basis of preliminary design results of combustor the mass flow through the swirler need be calculated according to
59 the air distribution, and then the effective area is determined by the pressure loss coefficient method. At the end, sizing of the swirler is
60 carried out. Therefore the methods used in this section are both theoretical and empirical.
61
62 3.1. Procedure of calculating axial swirler mass flow
63
64 Firstly the air flowing through the core mass flow (Mc) is calculated according to the total mass flow (MT) and turbine cooling bleed factor
65 (Ccooling) using Eq. (1).

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 3/11

B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11 3

1
2
3
4
5
6
7
8
9
10
11
12 Fig. 1. Geometric parameters of an axial swirler.
13
14
15

16 Mc ¼ MT  1  ccooling (1)
17
18 Secondly the air to fuel ratio (AF) at the design point, cruise condition of turboprop engine is estimated using Eq. (2), which reflects the
19 energy balance relationship of a typical gas turbine engine. In this equation, the specific heats Cpg and Cpa are found in thermodynamic
20 properties table of air and have the value of 1210 J/kg/K and 1093 J/kg/K respectively. Fuel Calorific Value (FCV) is 43 MJ here and hence the
21 air fuel ratio (AF) is 45.5.
22    
23 AF ¼ FCV  Cpg TET = Cpg TET  Cpa T3 (2)
24
Thirdly the fuel mass flow (Mf) needs to be again calculated by using Eq. (3), although the fuel mass flow has been given in overall
25
performance calculation, because here it is calculated from the prospective of combustion. The core mass flow (Mc) and air fuel ratio (AF)
26
have been obtained by above two steps and so the fuel mass flow is 0.21 kg/s.
27
28 Mf ¼ Mc =AF (3)
29
30 Fourthly some intermediate variables during calculation need be assumed because designing combustors cannot completely be theo-
31 retical, it is still said to be black art by many researchers [2,4]. These assumed variables include ratio of dome cooling mass flow (Mcd) to core
32 mass flow, Mcd/Mc, and ratio of air mass flow for atomizing fuel to fuel mass flow Mca/Mf if for air blast fuel jet. In order to ensure sufficient
33 and effective atomization the ratio of air mass flow for atomizing fuel to the fuel mass flow Mca/Mf has the value of 2–3, in this study value of
34 3 has been used. For annular combustors, the air for cooling dome is around 8% for older engine and 10–15% for modern engine, here the
35 value of 15% is chosen [16].
36 Fifthly the equivalence ratio in primary zone (Fp) should be assumed in order to calculate swirling air mass flow (Mcs). For a conventional
37 combustors, the equivalence ratio at primary zone ranges from 0.9 to 1.1. In this range there are relatively maximum flame temperatures and
38 consequently the emission pollutants such as NOx may be observed. If using rich burn quick quench technology, the equivalence ratio is
39 between 1.4 and 1.5. At this range the flame temperatures are lower than that of the conventional combustors and the level of emission are
40 lower. Therefore here the rich equivalence ratio is selected and has the value of 1.5.
41 Finally once the equivalence ratio is selected in primary zone, the swirling air mass flow can be calculated by using Eq. (4). AFth is the
42 Q3 stoichiometric air fuel ratio and the value is 14.7 (Table 1).
43
44 AFth
Mcs ¼  Mf  Mca  Mcd (4)
45 fP
46
47 3.2. Calculation of swirler area
48
49 The pressure loss coefficient method is used in this study to calculate the area of axial swirler. In general pressure loss of swirler should be
50 60–70% of total pressure loss through combustor. Therefore the pressure loss of swirler is here assumed as 65% of the total pressure loss
51 through the combustor. Pressure drop coefficient of swirler DPsw/qref can be obtained by pressure loss of swirler over reference dynamic
52 head. Pressure loss relationship of swirler is given in Eq. (5).
53 "    # 
54 DPs Aref 2 2 Aref 2 Msw
55 ¼ k sec q  (5)
qref As Al MT
56
57 Vane shape factor k is 1.3 for flat vane and 1.15 for curved vane, which indicates that pressure loss of curved vane is less than that of flat
58 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.
59
60
61 Table 1
62 Design results of air distribution (unit: kg/s).
63 MT Mf Md Mp Vp Mcs
64
9.53 0.21 2.45 7.08 1.25 1.02
65

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 4/11

4 B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11

1 In typical combustors, swirl induced by recirculation is augmented by the airflow through the primary hole; a certain amount of air
2 should be involved in mass flow of swirler for calculating area of swirler. Out of the quantity of airflow that enters in the primary zone of
3 combustor through primary holes, 30–70% takes part in the recirculation [8]. Consequently the area of swirler that can be calculated from
4 Eq. (5) need be corrected with augmented mass flow. Here allowance of 50% more airflow has been used.
5
6
3.3. Swirler sizing
7
8
The flow area of swirler (Asw) is calculated by both pressure loss method and geometric structure. The area of swirler can be calculated by
9
Eq. (6) if using the geometric dimensions.
10
11 p

Asw ¼ D2t  D2h  0:5ndðDt  Dh Þ (6)
12 4
13
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
14
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
15
the tip diameter and here is 0.5, which have better fall within the range of 1/3–2/3. The number of swirler vanes n commonly ranges from 8
16
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
17
all swirlers is considered as 1 mm and the number of swirler vanes is different from each swirler. After calculating these dimensions, the
18
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
19
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
20
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].
21
22
2 1  ðDh =Dt Þ3
23 SN ¼ tanðqÞ (7)
3 1  ðDh =Dt Þ2
24
25 Dhub/Dsw is defined as hub tip ratio of swirler and the second item in above equation generally have the value between 1 and 1.5.
26 For the curved vane the expression is different from the constant vane swirler due to the different angles at trailing and leading edge of
27 vane. Assuming the axial velocity component is uniform over cross section after swirler vane exit, the swirl number of curve vane swirler
28 with vane exit angle qm can be represented as Eq. (8) [13].
29
30 D2t þ D2h
31 SN ¼ tanðqm Þ (8)
2D2t
32
33 The vane space, s, is calculated by Eq. (9).
34
35 s ¼ pðDt þ Dh Þ=ð2nÞ (9)
36
The vane chord (c) can be obtained using Eq. (10) if assuming solidity of vanes is (sv)
37
38 c ¼ sv s (10)
39
40 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
41 calculation. NX Unigraphics [19] is an advanced CAD/CAM/CAE software package developed by Siemens PLM Software.
42
43 3.4. Results of axial swirler design
44
45 The five swirlers have been designed by procedure described above. One of them is designed for flat vanes and other four were designed
46 for curve vanes, as illustrated in Fig. 2. Details of swirlers designed are mentioned in Table 2. The hub diameter of swirler is typically
47 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.
48 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
49 procedure and has a value of 28 mm, but the real outer diameter of swirler is 30 mm with wall thickness of 1 mm.
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65 Fig. 2. 3-D curve vane (45
, 52
, 60
) swirler.

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 5/11

B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11 5

1 Table 2
2 Design results of axial swirler.

3 Name Vane angle Vane number Vane type Hub dia. (mm) Tip dia. (mm) Swirler number Block factor
4 S45 45 8 Curve 14 28 0.625 0.257
5 S52 52 8 0.81 0.295
6 S60 60 8 1.08 0.364
S60-N 60 10 0.81 0.455
7
S45-T 45 8 Flat 1.34 0.257
8
9
10
11 Swirler number of swirler with helical profile can be calculated by the tip diameter, hub diameter and tip vane angle using the Eq. (8). The
12 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.
13
14 j ¼ nd=2pRh cos q (11)
15
16 3.5. Validation of CFD model
17
18 Validation is carried out by simulating flow of an axial swirler in CFD to obtain axial and radial velocity distribution, which is then
19 compared with experiment results conducted by Fu et al. [13]. Fu has carried out an experimental investigation on aerodynamic charac-
20 teristic of confined, non-reacting swirling flow induced by the swirler with different vane angles [14].
21 In this study, one of the experimental model is validated, namely curved vane axial swirler. The single swirler has six curved vanes with
22 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
23 addition a convergent–divergent venturi is installed at the exit of swirler.
24 Calculation confinement section is considered as a cuboid with section of 50.8 mm  50.8 mm and length of 304.8 mm. The Grid quality
25 has been examined carefully so as to ensure no skew cell exists to influence numerical convergence. The skewed sized is less than 0.73 and
26 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
27 ‘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
28 turbulence scale length of 22 mm, which is the outer diameter of swirler. The Reynold’s stress model (RSM) is a higher level, elaborate
29 turbulence model [20]. It is usually called a Second Order Closure. One of the more successful recent developments is the realizable K–
30 Epsilon model developed by Shih et al. [21]. This model contains a new transport equation for the turbulent dissipation rate ε [22]. Also, a
31 critical coefficient of the model, Cm, is expressed as a function of mean flow and turbulence properties, rather than assumed to be constant as
32 in the standard model. This allows the model to satisfy certain mathematical constraints on the normal stresses consistent with the physics
33 of turbulence (realizability). The concept of a variable Cm is also consistent with experimental observations in boundary layers.
34 Fig. 3 shows axial velocity distribution along axial direction at z ¼ 120. The axial velocity distributions simulated by realizable turbulence
35 model are good agreement with those of experiment at positions z ¼ 3, z ¼ 12, z ¼ 48 and z ¼ 120 considered in this study.
36 Fig. 4 shows the radial velocity distribution along radial direction at z ¼ 48 mm. It can be observed from Fig. 4 that the k–ε realizable
37 turbulence model once again presents the desirable simulation results. It is also observed that k–ε standard turbulence model gives the
38 relative desirable computational results at positions of z ¼ 12 and 120 respectively while at the z ¼ 3 and z ¼ 48 positions the computational
39 errors are comparatively large. The RSM turbulence model gives the worst simulation results that the flow unsymmetrical occurs in each
40 position and have the irregular profiles. In particular, at position of z ¼ 120, it gives the completely different trends from simulation results of
41 the other three models.
42 Four turbulence models provided in FLUENT software are investigated to check whether they are suitable for validated swirler model and
43 have enough precision to deal with the swirling flow through comparing the simulation results with the experimental data. The results
44 indicate that renormalization group (RNG) turbulence model gives worst simulation results, while k–ε realizable turbulence model has best
45 simulation results. For axial velocity magnitude distribution along the axial direction, four turbulence models nearly have desirable profiles
46 and values. The axial velocity and radial velocity distribution along radial direction at different positions illustrated RNG and RSM turbulence
47 models seems to be sensitive to the flow field tangential variation and hence there are some deformation flow. However, the k–ε realizable
48 turbulence model is possibly much easier to deal with the swirling flow especially at the time that the comparable effects are investigated.
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65 Fig. 3. Axial velocity distribution at z ¼ 120 mm.

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 6/11

6 B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16 Fig. 4. Radial velocity distribution at z ¼ 48.
17
18
19 From above analysis it can be concluded that the incorrect selection of turbulence models may causes inaccurate simulation results.
20 Therefore model should be carefully selected when performing a CFD simulation.
21
22 4. Study of axial swirler flow
23
24 The effects of vane angle on aerodynamic characteristic of swirler are investigated by comparing the flow characteristic of swirler S45,
25 S52 and S60. A sketch of computational simulation domain is shown in Fig. 5. All the conditions of venturi and confinement used in
26 computational simulation models are same, differences are there in swirler models only. For a real combustion system, a passage such as
27 venturi is usually combined with the swirler to form a swirler cup for better atomization and recirculation zone. This passage could be in
28 form of several profiles that include flat passage, divergent passage or convergence–divergence nozzle as well as real venturi. In study done
29 by Fu et al. [13] convergence–divergence nozzle is used to replace venturi. It is recommended that the optimum half angle of convergence–
30 divergence nozzle should be about 30 and length ranges from one to two times of swirler exit diameter. Therefore the convergence–
31 divergence nozzles used in simulation models have half angle of 30 and length of 14 mm.
32 The density of the air is set to be an incompressible ideal gas due to Mach number less than 0.3. The value of specific heat, thermal
33 conductivity and viscosity is obtained from the air properties temperature table according to the inlet temperature of Group Design Project
34 [15] combustor. All of items at solution controls are all selected to be second order and under-relaxation factors were adjusted down to a
35 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 flow on
36 swirler exit flow different inlet velocities were studied. Effects of vane number were also investigated through comparing flow characteristic
37 created swirler S60 and S60-N.
38
39 5. Results and discussion
40
41 5.1. The effect of vane angle on flow characteristic of swirler
42
43 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

44 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
45 constant value of main flow. It is also observed that for most of the axial distance cases with 45
angle, high axial velocity is seen. This also
46 shows that by increasing the vane angle of swirl axial velocity decreases which would eventually increase recirculation.
47 Fig. 7 shows turbulence intensity characteristics for swirler with different vane angle along the axial direction. It is observed that for all
48 the three cases under consideration turbulent intensity decreases with axial distance. This is due to the fact that turbulence intensity
49 decreases with distance due to losses. Case with 45
angle is showing lowest turbulence intensity whereas; case with 60
is showing high
50 turbulence intensity at majority of the sections. Similar observations are also observed for turbulent kinetic energy and turbulent viscosity
51 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
52 angle produces larger turbulence.
53
54
55
56
57
58
59
60
61
62
63
64
65 Fig. 5. Sketch of swirler domain.

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 7/11

B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11 7

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Fig. 6. Variation of axial velocity with vane angle.
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33 Fig. 7. Variation of turbulent intensity with vane angle.
34
35
36 Fig. 8 shows the relationship between vane angle and pressure drop coefficient. It is observed that by increasing vane angle pressure drop
37 coefficient also increases. It can be easier to understand that increasing vane angle increases the chance to separate from the vane wall
38 surface, and hence the secondary loss increases and further the pressure drop coefficient increase. For a case with 45
vane angle pressure
39 drop coefficient is 7% whereas when the vane angle is increased to 60
pressure loss coefficient increases to 40%. This is a considerable
40 amount which would hamper performance of combustor substantially.
41
42
43 5.2. Effect of changing number of vanes on flow characteristics of swirler
44
45 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
46 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
47 number of vanes along the axial axis. It is observed that the axial velocity decreases with the increase of vane number.
48 Fig. 10 is showing variation in turbulent kinetic energy by increasing the number of vanes along the axial axis. It is observed that tur-
49 bulent kinetic energy increases marginally with increase in number of vanes. Similar effect as of turbulent kinetic energy is also observed for
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65 Fig. 8. Variation of pressure loss coefficient with vane angle.

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 8/11

8 B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17 Fig. 9. Variation of axial velocity with vane number.
18
19
20
turbulent intensity and turbulent viscosity ratio. Decrease in axial velocity and increase in turbulence with increase in number of vanes will
21
also lead to increased recirculation and better flammability limits.
22
23
24 5.3. Effect of inlet mass flow on pressure loss of swirler
25
26 One of the most important performance parameters for a curved vane type swirler is the pressure drop-mass flow characteristic, namely
27 what pressure drop has been occurred through the swirler at a certain mass flow. Here swirler S45 with 45
angle has been selected to
28 investigate pressure drop across swirler by changing mass flow. Pressure drop is also the static pressure difference between inlet and outlet
29 at swirler and has been normalized by swirler inlet dynamic head. The pressure drop coefficient can be consequently expressed as Eq. (12).
30 pi  po
31 PDC ¼ (12)
0:5rV 2
32
33 Fig. 11 shows this characteristic of case S45, which is an example of strong swirl flow. It can be seen that the pressure drop decreases with
34 the increase of mass flow for given vane angle, vane number, tip and hub diameter. Pressure loss coefficient is 27% when the mass flow rate is
35 0.022 kg/s which decreased to 2.5% when the mass flow is increased to 0.039 kg/s. One of the possible reason for this is blockage in swirler
36 passage due to separation has been avoided and consequently reduces the frictional loss between vanes. It can be also seen that the pressure
37 loss coefficient seems to tend towards a constant value with increasing the mass flow. Therefore it appears to deduce that above a certain
38 mass flow the pressure drop coefficients are not dependent on the mass flow.
39
40 5.4. Velocity profiles
41
42 The axial and radial velocities at different positions along radial direction have also been studied. Fig. 12 shows axial velocity profile at the
43 near exit or near field decay part that approximately ranges from z ¼ 3 mm to z ¼ 9 mm. In this region the velocity profiles of double peak
44 curves are observed and the velocities decay with the increase of axial distance.
45 The near exit or near field similarity flow region is illustrated in Fig. 13, in which the axial velocities nearly don’t change with the axial
46 positions and have the U shape profiles. The peak velocity reduces from 38 m/s to25 m/s, approximately 0.66 times the peak velocity at z ¼ 3
47 position. The centreline reverse flow velocity is about 25 m/s.
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65 Fig. 10. Variation of turbulent kinetic energy with vane number.

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 9/11

B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11 9

1
2
3
4
5
6
7
8
9
10
11
12
13
14 Fig. 11. Variation of pressure drop coefficient with mass flow.
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31 Fig. 12. Axial velocity profiles at near exit zones. Q5
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Fig. 13. Axial velocity profiles at near exit similarity zones.
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65 Fig. 14. Radial velocity profiles at near exit zones.

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 10/11

10 B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11

1 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
2 seen that the radial positions of peak in the radial velocity nearly move outwards in downstream direction as a general trend in the flow
3 field.
4 Fig. 14 shows that the radial velocity peaks along radial direction continuously decrease at near exit zone. It can also be seen that the
5 recirculation heights along radial direction increase gradually with the increase of axial distance. Fig. 15 describes the radial velocity profiles
6 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
7 values of free vortex continuously decrease along the direction towards downstream but the radial positions of radial velocity peak nearly
8 keep unchanged.
9 Fig. 16 shows the radial velocity profiles at the second decay zone, in which the radial peak velocities reduce slowly and the radial
10 position of radial velocity peak initially unchanged and then move inward over z ¼ 46 mm. This indicates that a reverse flow at radial
11 velocity direction is occurred at this position where is the maximum height of recirculation zone.
12 Fig. 17 gives the distributions of radial velocity at different axial positions. It can be seen that the wake flow region starts very early at
13 axial position. Although radial velocity still keeps the trend of second decay zone the radial velocity profiles are irregular.
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31 Fig. 15. Radial velocity profiles at near exit similarity zone.
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48 Fig. 16. Radial velocity profiles at second decay zone.
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65 Fig. 17. Radial velocity profiles at far field.

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022
 JOEI33_proof ■ 3 April 2014 ■ 11/11

B. Khandelwal et al. / Journal of the Energy Institute xxx (2014) 1–11 11

1 6. Conclusions
2
3 The present works in this paper mainly focus on the procedure of designing axial swirler, and further investigation on flow characteristic
4 of swirling flow generated by vane axial swirler by help of computational methods. Moreover the effects of geometric parameters, namely
5 vane angle and vane number and flow are also reported in the study.
6 Swirler design is carried out on the basis of semi-empirical equations. The design procedure using pressure drop method include three
7 steps namely, mass flow calculation through air distribution, effective area calculation through pressure drop and sizing from the geometric
8 relationship. The non-reacting swirling flow can be simulated very well through using k–ε realizable turbulence model provided in FLUENT
9 that has been validated in this study.
10 Axial reverse flow velocity, turbulence along axial direction and pressure drop increases with the increase in angle of vanes. Namely the
11 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
12 combustors. Increasing the vane number increases the turbulence energy and decreases the axial flow velocity. For a given swirler,
13 increasing mass flow at inlet decreases the pressure drop through the swirler and the pressure drop seems to keep constant if the mass flow
14 is increased above a critical value. Thus the mass flow through the swirler should be adjusted carefully when designing swirler. The results
15 show that the axial velocity and radial velocity profiles have four different types of profiles.
16
17 References
18
19 [1] B. Khandelwal, Y. Li, P. Murthy, V. Sethi, R. Singh: Proc. of ‘ASME Turbo Expo 2011’, GT2011-46845, pp. 293–298.
[2] B. Khandelwal, M. Yan, G. Hegde, V. Sethi, R. Singh, SAE Aerotech 2011, Toulouse, France, October 2011. SAE paper 2011-01-2562.
20 [3] M. Abd-Rabbou, Cranfield Thesis, Cranfield, Bedfordshire, UK, 1977.
21 [4] J.N. Murthy, PHD Thesis, Cranfield University, 1988.
22 [5] P.J. Stuttaford, P.A. Rubini, ASME J. Eng. Gas. Turbines Power 119 (3) (1997) 546–552.
[6] H. Mongia, in: 39th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, 20–23 July 2003. Huntsville, Alabama, AIAA 2003-4495.
23 [7] E. Kilik, PHD Thesis, Cranfield University, 1976.
24 [8] Sawyer’s Gas Turbine Engineering Handbook, third ed., vol. 1, 1985.
25 [9] J.M. Beer, Combustion Aerodynamics, Applied Science Publishers Ltd, 1972.
[10] S.A. Beltagui, N.R.L. Maccallum, J. Inst. Fuel 49 (1967) 183–193.
26
[11] Y.Q. Fu, S.M. Jeng, in: 44th Aerospace Sciences Meeting & Exhibit, Reno, 2006. AIAA 2006-545.
27 [12] E. Kilik, in: 21st AIAA/SAE/ASME/ASEE Joint Propulsion Conference, Monterey, 1985. AIAA Paper 1985-1103.
28 [13] Y.O. Fu, S.M. Jeng, T. Robert, in: ASME Turbo Expo 2005, Reno, Nevada, USA, June 2005, pp. GT2005–68728.
[14] E. Kilik, in: 23rd Aerospace Sciences Meeting, Reno, NV, 1985. AIAA Paper 1985-1087.
29
[15] MSc Thermal Power 2010: ‘Group Design Project’, Cranfield University, UK, March 2010.
30 [16] A.H. Lefebvre, Gas turbine combustion, second ed., Taylor & Francis Group, Great Britain, 1998.
31 [17] David G. Lilley, J. Propuls. Power 15 (2) (1999).
32 [18] G.B. Hegde, B. Khandelwal, V. Sethi, R. Singh, in: ASME Turbo Expo 2012, Copenhagen, Denmark, June 2012. GT2012-69215.
[19] http://www.plm.automation.siemens.com (accessed 28.06.13).
33 [20] B.E. Launder, G.J. Reece, W. Rodi, J. Fluid Mech. 68 (3) (1975) 537–566.
34 [21] L.H. Shih, J.R. Koseff, G.N. Ivey, J.H. Ferziger, J. Fluid Mech. 525 (2005) 193–214.
35 [22] http://www.kxcad.net/star-ccm/online/139-kepsilonturbulence-07.html (accessed 28.06.13).

36
37

Please cite this article in press as: B. Khandelwal, et al., Design and study on performance of axial swirler for annular combustor by changing
different design parameters, Journal of the Energy Institute (2014), http://dx.doi.org/10.1016/j.joei.2014.03.022