Professional Documents
Culture Documents
Effect of Atomization Quality On Lean Blow Out Limits and Acoustic Oscillations in A Swirl Stabilized Burner - Published
Effect of Atomization Quality On Lean Blow Out Limits and Acoustic Oscillations in A Swirl Stabilized Burner - Published
Effect of Atomization Quality On Lean Blow Out Limits and Acoustic Oscillations in A Swirl Stabilized Burner - Published
Introduction
In spray combustion systems, atomization and vaporization processes exert significant
influence on the combustion characteristics and lean blow-out (LBO) limits of gas turbine
combustors Polymeropoulos and Das (1975); Ballal and Lefebvre (1980); Lefebvre (1985).
The LBO limit is an important performance parameter that needs to be considered for
flame stability in aero-engine combustors, over a wide range of operating conditions. For
heterogeneous fuel air mixtures, flame blow-out occurs when the heat released in the
combustion zone becomes insufficient to evaporate the fuel and to further heat the fresh
mixture to the required ignition temperature level to enable combustion reactions
(Lefebvre, 1985).
Several investigations have been carried out to understand the effects of fuel type,
operating conditions of reactants and exit conditions of a combustor on the LBO limits.
Baxter and Lefebvre (1992) performed experimental investigations to highlight the effect
of inlet temperature and velocity of mixture on weak extinction limits of a bluff body
stabilized burner using liquid hydrocarbon fuel. Sturgess et al. (1993) investigated the
effect of back pressure on LBO limits in a research combustor at various combustor
loadings. At greater exit blockage, significant improvement in the LBO limits is observed
at higher combustor loading conditions. At higher inlet mass flow rate, the residence time
of mixture decreases and therefore LBO equivalence ratio increases (Hejie et al. 2007).
Griebel et al. (2007) observed that premixing of hydrogen with methane air mixture
increases the flame speed, which shifts the blow-out condition to lower equivalence ratios.
Lieuwen et al. (2008) investigated the influence of chemical kinetic time of different fuel
compositions on blow-out in a lean premixed combustor.
Many authors have attempted to correlate and predict LBO limits in terms of operating
conditions of reactants and acoustic signals from combustor. Lefebvre (1985) developed
a correlation for heterogeneous fuel air mixtures using the experimental results of LBO
conditions of various gas turbine combustors. Attempts were also made to predict the
LBO limits using flame images or acoustic signals of the combustor. Tongxun and
Gutmark (2007b) proposed a method to predict the incipient LBO conditions in partially
premixed combustors using turpentine as fuel. They showed that the flame can be
stabilized at leaner conditions by increasing the swirl intensity of air flow. Xie et al.
(2012) related the LBO limit with the size of flame considering different combustor
configurations and developed an empirical correlation to predict LBO limits based on
the flame volume concept using these configurations.
Prior to LBO limit, when the operating condition of combustion system shifts from
stoichiometric to leaner conditions, significant increase in acoustic oscillations were
reported in both premixed and non-premixed combustors by many researchers Keith
et al. (1998); Venkataraman et al. (1999); Cohen et al. (1999); Tongxun and Gutmark
(2007b). Lieuwen et al. (2001) identified the mechanism responsible for combustion
instability in lean premixed systems and predicted conditions under which instability
occurs, in terms of the ratio of convective time to the period of oscillations. Lieuwen
(2005) plotted the pressure amplitude and damping coefficient with respect to mixture
velocity at different frequencies and observed positive damping coefficient at stable
condition and a zero damping coefficient value near unstable condition. Using the
damping coefficient values, a technique was developed to ascertain the stability limit.
The influence of convective time of mixture on the pressure fluctuations has been
investigated by Lieuwen et al. (2008) for different fuels. Samadhan et al. (2016) varied
the location of flame in a model spray combustor using ethanol as fuel at constant
equivalence ratio. Intermittent periodic pressure oscillations with higher peak and rms
values were observed before limit cycle oscillations.
Some authors have highlighted the effect of droplet diameter on combustion
instabilities associated with spray combustor, particularly at leaner operating condi-
tions. Levy and Bulzan (1995) and Nicoli et al. (2007) replaced gaseous fuel by liquid
fuel in the form of a spray, in a premixed burner and observed more oscillations in the
flame structure. The evaporation delay for large droplets in a spray combustor can give
rise to lower heat release rates, reduction in combustion efficiency and transition from
kinetic to diffusion-controlled combustion Bossard and Peck (1996) and Awasthi et al.
(2013). This vaporization delay leads to the formation of nonuniform mixture which
can result in equivalence ratio oscillations. The equivalence ratio oscillations at leaner
operating conditions trigger combustion instabilities. Yu and Wilson (2002) investi-
gated the effect of droplet size in secondary fuel injection stream on the pressure
oscillations of a dump combustor. With decrease in SMD (Sauter mean diameter),
the critical amount of fuel flux required to suppress the combustion generated pressure
1030 G. MUTHUSELVAN ET AL.
oscillations decreases. Lei and Turan (2010) demonstrated the significant impact of
evaporation process on the intensity of nonlinear pressure oscillations. Garcia et al.
(2009) investigated the influence of spray characteristics on thermo-acoustic instability
and observed a strong correlation between the frequency of spray fluctuation and
combustion oscillations. Chishty et al. (2011) investigated the effect of atomization
quality on the operational regime of a swirl stabilized combustor with kerosene. For
improved atomization quality (lower SMD), the transition from stable to unstable
combustion regime is delayed.
Duvvar et al. (1996) investigated the effect of droplet velocity relative to the velocity of
gas medium to determine the condition under which the vaporization process is capable of
driving a longitudinal mode of instability. When droplet moves slower than gas, a wake is
formed ahead of the droplet. This configuration affects the vaporization process such that it
supplies more energy to drive the combustion instability. But when droplet moves faster
than the surrounding gas, the vaporization process damps out the combustion oscillations.
The survey of literature presented above indicates that the LBO limit in a spray
combustor is a very complex phenomenon, influenced by many factors. Although many
studies are available on LBO limits and acoustic phenomena in a non-premixed combus-
tion system, most of these deal with gaseous fuels. Some researchers have discussed about
the LBO limits and acoustic phenomena in spray combustors, with specific atomizer
configurations. In the present study, different configurations of injector (representing
small gas turbine applications) are used to represent variation in atomization quality.
The acoustic oscillations and LBO limits are measured for different values of initial mean
drop size (SMD) of spray. Correlations have been developed for LBO condition in terms
of different time scales. In addition to the three time scales (evaporation time, mixing time
and reaction time) used by Lefebvre (1985), three more time scales (droplet residence
time, duct acoustics and residence time of gas flow) are considered in the development of
present correlation. The study identifies a critical droplet size range for mode change in
lean blowout and also relates the overall sound pressure level (OASPL) in the combustor
to the initial mean droplet size (SMD).
Methodology
The range of flow number (FN ¼ pffiffiffiffiffiffiffi
mf ffi
) of atomizer used in the present investigation
ΔPρl
varies from 2.18 × 10–7 to 5.01 × 10−7m2. The SMD of the fuel spray for the five different
injector configurations is evaluated using a Malvern spray analyzer with 5 mW helium–
neon laser beam of 10 mm diameter and 300 mm optical lens. The SMD is measured at
30 mm from the orifice exit in an unconfined ambience under isothermal (T = 300 K)
conditions using jet-A fuel. The measured values of SMD for the five different atomizer
configurations are shown in Table 1, for a range of fuel flow rates and atomizer constant
values. The accuracy of SMD data presented in Table 1 is ± 3%. In Table 1, for a particular
fuel flow rate, SMD of the spray increases with increase in atomizer constant K (= total
port area/(swirl chamber diameter x orifice diameter)). In other words, using five different
atomizer configurations corresponding to different K values in the swirl stabilized burner,
the atomization quality (initial average droplet diameter) is varied. Hot flow combustion
experiments have been carried out in a swirl stabilized burner using these five different
COMBUSTION SCIENCE AND TECHNOLOGY 1031
Table 1. SMD for different simplex injectors at various fuel flow rates (Muthuselvan et al. (2018)).
Fuel flow Sauter mean diameter (SMD) (µm)
Sl. no. rate (kg/h) K = 0.132 K = 0.177 K = 0.368 K = 0.530 K = 0.655
1 9 78 85 138 205 344
2 10 74 79 125 181 289
3 11 67 74 111 179 241
4 12 63 69 101 142 202
5 13 59 63 98 129 177
6 14 53 61 90 118 158
7 15 50 59 80 111 144
atomizer configurations for fuel flow rate varying in the range of 9–15 kg/h. The influence
of atomization quality on lean blowout limits and acoustic oscillations is quantified.
An axial air swirler with swirl angle of 60° at tip radius is used and the corresponding
swirl number is estimated as 1.4. Figure 1(a) shows the atomizer with air swirler employed
in the present study and Figure 1(b) shows the assembly of atomizer, swirler and dome.
A rectangular combustion chamber (120 mm × 100 mm) having a length of 570 mm
(Figure 2) is used downstream of the burner assembly. The schematic diagram of experi-
mental setup used to measure the LBO limits and the sound spectra in a swirl stabilized
burner, is shown in Figure 3.
The sound spectra from the combustor system are obtained using unsteady micro-
phone measurements for different injector geometries and equivalence ratio conditions.
A PCB make microphone (Model no. 103B02) with a sensitivity of 217.5 mV/kPa is placed
at a distance of 300 mm from the side plate of the swirl stabilized burner. A four-channel
OROS analyzer is used to acquire the output signal of the microphone. Before using the
microphone for actual measurements, it is calibrated using a GRAS – make piston phone
(type 42AC) which generates 134 dB noise at 250 Hz. The corresponding output of the
microphone was observed to be around 133.1 dB at a frequency of 250 Hz. The data
analysis was performed using PROSIG DATS V7.0.26 software.
An optical view port is located on the side plate of the burner, to view the flame at different
operating conditions. Coriolis mass flow meters are used to measure the fuel and air mass flow
rates through the atomizer and axial swirler, respectively, with an accuracy of ± 0.125% of full
range. In the present experimental investigation, initially the fuel flow rate is maintained
constant and the air flow rate is gradually increased in steps of 10 kg/h, in general. While
approaching the lean blowout condition, air flow rate is increased in steps of 1 kg/h and
Figure 1. Simplex atomizer with axial swirler and dome: (a) atomizer with air swirler and (b) assembly
with dome.
1032 G. MUTHUSELVAN ET AL.
maintained constant for 2 min to capture LBO accurately. The flame features are observed
through both the exit section and the side view port at all air mass flow rate conditions. By
following this procedure, the air mass flow rates at both lift-off and LBO equivalence ratios are
noted. Subsequently, the air mass flow rate is maintained constant and the fuel flow rate is
gradually decreased to simulate lean operating conditions and both lift-off and LBO equiva-
lence ratios are noted. The errors in fuel and air flow rate measurements are within ± 1%.
drop size (measured at cold flow condition, as listed in Table 1). The rate of droplet
evaporation (as evaluated through d2 law) and the associated rate of fuel burning are
strongly dependent on droplet size (Awasthi et al. (2013)). Therefore, the initial mean
droplet size is taken as a measure of spray atomization quality in the parametric study.
This approach has also been followed by Lefebvre (1985) for studying lean blowout
conditions. The variation in atomization quality for different injector configurations
and operating conditions have been comprehensively described by Muthuselvan et al.
(2018).
Figure 4. Variation of acoustic oscillations with respect to equivalence ratio, at initial mean droplet diameter of 205 µm.
COMBUSTION SCIENCE AND TECHNOLOGY 1035
130
140
120
135
110
130 100
125 90
Amplitude (dB)
80
120
0 50 100 150 200 250 300 350 400
115
110
105
100
Figure 5. Variation of sound spectra with respect to equivalence ratios (initial mean droplet diameter of
205 μm, mf = 9 kg/h, K = 0.53).
135
130 Ф = 0.888
125
120
Amplitude (dB)
115
110
105
100
95
SMD = 85
90 SMD = 138
85 SMD = 205
SMD = 344
80
0 100 200 300 400 500 600 700 800 900 1000
Frequency (Hz)
Figure 6. Variation of sound spectra for different initial mean droplet sizes at equivalence ratio of 0.888.
1036 G. MUTHUSELVAN ET AL.
13.0
12.0 SMD - 85
11.0
SMD - 138
Figure 7. Comparison of average drop size distributions for different average droplet diameter at mf
= 9 kg/h and ma = 0 (Muthuselvan et al. 2018).
the droplet size distributions for four different atomizer geometries under non-reacting cold
flow conditions are shown (Muthuselvan et al., 2018), in an unconfined ambience (K varying
from 0.177 to 0.655). It is noted that the volume fraction of larger droplets increases with
increase in the mean droplet diameter. As higher fractions of larger droplets are generated,
evaporation delay increases for the diffusion mode of combustion prevailing in the spray
combustor. Bossard and Peck (1996) observed reduction in combustion efficiency for sprays
with wider drop size distributions. Thus, if a large volume of vapor accumulates due to
evaporation delay, large amplitude oscillations may be generated due to the rapid combustion
of the accumulated vapor and the subsequent expansion of the hot product gas. This can be
thought of as being akin to the “diesel knock” phenomenon which occurs due to vapor
accumulation in diesel engines. While the frequency of acoustic oscillations is locked-on to
the longitudinal quarter wave mode, the overall sound level seems to be related to the quantity of
vapor accumulated. Therefore, as seen in Figure 6, the sound level is significantly higher when
initial drop diameters are larger.
Effect of fuel atomization quality and equivalence ratio on overall sound pressure
level (OASPL)
In Figure 8, the OASPL variations with respect to equivalence ratio for different initial mean
droplet diameters of spray are shown at the fuel flow rate of 15 kg/h. The transition equivalence
ratios at which shift in first mode frequency occurs are circled in Figure 8, for the sake of
comparison. It is observed that with decrease in equivalence ratio, the OASPL gradually increases
up to the transition equivalence ratio, after which the OASPL increases steeply as the LBO
condition is approached. With decrease in initial mean droplet diameter, the transition equiva-
lence ratio continuously decreases. Smaller droplets evaporate relatively faster, and hence eva-
poration delay reduces. In turn, the high amplitude acoustic oscillations (Gireesh Kumaran and
Sujith (2015)) are postponed to higher air flow rates. Therefore, with decrease in initial mean
droplet diameter of the spray, the transition equivalence ratio shifts toward leaner operating
conditions. The reason for significant decrease in the OASPL at equivalence ratio of 0.55 for initial
mean droplet diameter of 50 µm is discussed in the following section.
COMBUSTION SCIENCE AND TECHNOLOGY 1037
144
140
136
132
OASPL (dB)
128
124
120
116
50 µm 80 µm
112
108 111 µm 144 µm
104
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1
Equivalence ratio
Figure 8. Variation of OASPL with respect to equivalence ratio for different initial mean droplet
diameters, mf = 15 kg/h.
145 140
140 130
120
135 110
130 100
125 90
Amplitude (dB) 80
120
0 40 80 120 160 200 240 280 320 360 400
115
110
105
100
95
90
85
80 Φ = 1.0 Φ = 0.67 Φ = 0.6 Φ = 0.57 Φ = 0.55
75
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Frequency (Hz)
Figure 9. Variation of sound spectra with respect to equivalence ratio (initial mean droplet dia-
meter = 50 μm, mf = 15 kg/h, K = 0.132).
increase in air flow rate for a lifted flame condition can cause LBO when initial mean droplet
diameter is more than 80 µm. When initial mean droplet diameter is higher than 140 µm, the flame
blows-out immediately after lifting from the swirler exit. The smaller droplets (<80 µm) issuing
from the injector can vaporize faster and mix with air flow rapidly in the recirculation zone; so, the
partially premixed flame may be stable even at lifted condition. But in the case of a spray with larger
droplets, fuel droplets are not able to vaporize completely within the recirculation zone to stabilize
the flame. Hence, LBO occurs immediately after the lifted condition. This highlights that atomiza-
tion quality significantly affects the mode of LBO as well as the OASPL.
In Figure 11, the equivalence ratio at LBO condition is plotted with respect to the initial mean
droplet diameter of spray. Numerous experiments are carried out at each condition of initial mean
droplet diameter, and the average value of blow-out equivalence ratio is plotted with error bar.
The spread of LBO limit within the error bar could be attributed to minor variations in the droplet
size distribution at nominally same combustor operating conditions. Since the time required for
complete evaporation of fuel droplet increases with initial mean droplet diameter, the LBO
equivalence ratio also drastically increases with drop size. In Figure 12, the equivalence ratio of
a swirl stabilized burner at LBO condition is plotted with respect to fuel flow rate for five different
injector configurations. Various diagonal lines shown in Figure 12 represent different constant air
mass flow rates. For a given fuel flow rate, the LBO equivalence ratio increases with increase in
atomizer constant. Similarly, the LBO equivalence ratio is observed to decrease with decrease in
atomizer constant (i.e., poorer atomization quality) at a fixed air flow rate also.
Development of empirical correlations for the LBO limit of a swirl stabilized burner
In order to investigate the effect of atomization quality on LBO limits, using five
different atomizer configurations in the range of fuel flow rate (9–15 kg/h), 35 different
operating conditions are considered (refer to Table 1). For each operating condition,
LBO limits are measured and repeated about six times to get better repeatability. Using
the experimentally measured LBO limits, an empirical correlation is developed using
COMBUSTION SCIENCE AND TECHNOLOGY 1039
0.7
0.675
0.65
0.6
0.575
0.5
50 60 70 80 90 100 110 120 130 140
Initial droplet diameter (micron)
(a) Lift-off and blow-out data
0.7
0.675
0.65
Equivalance ratio
0.625
0.6
0.575
Lean blowout
0.55
Lifted flame
0.525 Lifted flame
0.5
50 60 70 80 90 100 110 120 130 140
Initial droplet diameter (micron)
(b) Classification of stable and unstable zones
Figure 10. Variations of equivalence ratio at lift-off and blowout condition with respect to initial mean
droplet diameter of spray: (a) lift-off and blow-out data and (b) classification of stable and unstable
zones.
the data obtained from 206 different experimental runs. The correlation for the
equivalence ratio at LBO condition (EquLBO) is expressed in terms of six different
time scales, as shown in Eq. (1). In addition to the three time scales (evaporation time,
mixing time and reaction time) used by Lefebvre (1985), three more time scales (for
droplet residence, duct acoustics and gas flow residence) are considered in the devel-
opment of present correlation. Employing only the first four time scales (droplet
evaporation, droplet residence, reaction and gas flow residence), a modified correlation
is also presented in Eq. (2). Table 2 shows the values of six different time scales
1040 G. MUTHUSELVAN ET AL.
0.800
0.750
0.650 K = 0.132
0.600 K = 0.177
K = 0.368
0.550
K = 0.53
0.500
K = 0.655
0.450
40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360
Initial droplet diameter (micron)
Figure 11. Variation of equivalence ratio at LBO condition with respect to initial mean droplet diameter
of spray, for different injector configurations.
0.8
0.76
Lean blowout equivalence ratio
0.72
0.68
0.64
0.6
0.56
0.52
0.48
8 9 10 11 12 13 14 15 16
Fuel flow rate (kg/hr)
Figure 12. Variation of equivalence ratio at LBO condition with respect to fuel flow rate for different
atomizer configurations (□ – K = 0.132, X – K = 0.177, Δ – K = 0.368, ◊ – K = 0.53, ○ – K = 0.655).
calculated, covering five different atomizer configurations in the range fuel flow rates
(9–15 kg/h)) with increasing initial mean droplet diameter conditions. The details of
the calculation of six different time scales are described in Appendix – A. The
coefficient values are calculated through regression for both Eqs. (1) and (2) and
COMBUSTION SCIENCE AND TECHNOLOGY 1041
Table 2. Six different time scales calculated at LBO condition for different initial mean droplet diameter.
Fuel LBO_
flow Equiv Droplet Droplet Gas Reaction Acoustic
Sl. Atomizer rate alence residence time evaporation residence time Mixing time
no. SMD constant (K) (mf) ratio (ms) time (ms) time (ms) (ms) time (ms) (ms)
1 50 0.132 15 0.521 5.096 3.239 1.821 3.323 0.620 2.368
2 53 0.132 14 0.524 5.460 3.652 1.961 3.405 0.669 2.398
3 59 0.132 13 0.525 5.888 4.504 2.115 3.575 0.724 2.415
4 59 0.177 15 0.553 5.612 4.405 1.929 2.445 0.659 2.526
5 61 0.177 14 0.549 6.014 4.723 2.051 2.676 0.702 2.519
6 63 0.132 12 0.524 6.388 5.141 2.286 3.853 0.784 2.420
7 63 0.177 13 0.549 6.485 5.070 2.209 2.829 0.758 2.529
8 67 0.132 11 0.541 6.978 5.901 2.573 3.454 0.888 2.521
9 69 0.177 12 0.555 7.034 6.090 2.416 2.867 0.832 2.557
10 74 0.132 10 0.540 7.686 7.189 2.826 3.727 0.978 2.533
11 74 0.177 11 0.574 7.679 7.105 2.725 2.623 0.944 2.613
12 78 0.132 9 0.535 8.564 8.019 3.113 4.208 1.082 2.530
13 79 0.177 10 0.574 8.457 8.129 2.994 2.817 1.041 2.624
14 80 0.368 15 0.642 7.083 7.774 2.226 1.407 0.769 2.923
15 85 0.177 9 0.568 9.399 9.427 3.298 3.148 1.152 2.627
16 90 0.368 14 0.631 7.597 9.694 2.345 1.562 0.811 2.937
17 98 0.368 13 0.620 8.189 11.398 2.482 1.746 0.860 2.951
18 101 0.368 12 0.631 8.880 12.286 2.734 1.751 0.951 2.937
19 111 0.368 11 0.645 9.705 14.929 3.046 1.742 1.065 2.920
20 111 0.53 15 0.664 7.937 13.988 2.298 1.285 0.796 2.898
21 118 0.53 14 0.650 8.517 15.705 2.412 1.434 0.836 2.914
22 125 0.368 10 0.635 10.686 18.748 3.301 1.949 1.158 2.932
23 129 0.53 13 0.629 9.181 18.491 2.516 1.670 0.872 2.940
24 138 0.368 9 0.620 11.894 22.681 3.584 2.258 1.261 2.951
25 142 0.53 12 0.657 9.951 22.682 2.841 1.559 0.992 2.907
26 144 0.655 15 0.725 8.476 22.837 2.500 1.055 0.872 2.838
27 158 0.655 14 0.710 9.089 27.128 2.624 1.158 0.916 2.852
28 177 0.655 13 0.696 9.789 33.514 2.774 1.273 0.970 2.865
29 179 0.53 11 0.692 10.865 35.707 3.259 1.450 1.147 2.870
30 181 0.53 10 0.712 11.970 37.521 3.686 1.447 1.307 2.850
31 202 0.655 12 0.693 10.621 43.082 2.993 1.360 1.049 2.868
32 205 0.53 9 0.721 13.321 48.167 4.145 1.507 1.479 2.842
33 241 0.655 11 0.710 11.600 60.716 3.342 1.367 1.179 2.852
34 289 0.655 10 0.754 12.779 87.785 3.890 1.296 1.386 2.814
35 344 0.655 9 0.765 14.215 123.255 4.382 1.349 1.573 2.805
they are listed in Tables 3 and 4, respectively. The predicted equivalence ratio from Eq.
(1) is compared with the measured equivalence ratio at LBO condition as a parity plot
in Figure 13. A good match between the predicted and measured experimental results
is observed. In Figure 14, the predicted equivalence ratio from Eq. (2) is compared
with the measured equivalence ratio at LBO condition:.
1042 G. MUTHUSELVAN ET AL.
Tevp;drop Treaction Tmix;gas
EquLBO ¼ C0 C1 C2 C3 (1)
Tres;drop Tres;drop Tacoustic
Tevp;drop Treaction
EquLBO ¼ D0 D1 D2 (2)
Tres;drop Tres;drop
The predicted equivalence ratio at LBO condition of the correlation using six time scales
(Figure 13) shows a similar trend with the correlation using four time scales (Figure 14).
Further, the corresponding values of coefficients and R2 value of Eqs (1) and (2) are closer.
0.80
Equivalence ratio_LBO_Predicted
0.75
0.70
0.65
0.60
0.55
0.50
0.50 0.55 0.60 0.65 0.70 0.75 0.80
Equivalence ratio_LBO_Measured
Figure 13. Comparison of equivalence ratio at LBO condition from the present experimental data and
predicted by Eq. (1), R2 = 0.9827.
0.80
Equivalence ratio_LBO_Predicted
0.75
0.70
0.65
0.60
0.55
0.50
0.50 0.55 0.60 0.65 0.70 0.75 0.80
Equivalence ratio_LBO_Measured
Figure 14. Comparison of equivalence ratio at LBO condition from the present experimental data and
predicted by Eq. (2), R2 = 0.9811.
COMBUSTION SCIENCE AND TECHNOLOGY 1043
Thus, it is evident that the mixing time of gas flow and duct acoustic time scale play only
minor roles with respect to the prediction of LBO limits, as compared to the other four time
scales. Since the swirler used in the present investigation has a good swirl strength (swirl
number of 1.4), mixing rates are high and mixing time does not significantly vary with respect
to different initial droplet diameter values at LBO condition. As already stated, the amplitude
of acoustic oscillations is related to the volume of vapor accumulated (proportional to initial
mean droplet diameter) but the acoustic time does not significantly affect the LBO condition
of the flame. In other words, for the small scale combustors considered in the present work,
acoustic oscillations appear to be effects occurring near the LBO condition, rather than the
causes for the LBO.
Variation in the ratio of droplet evaporation time to droplet residence time with
respect to increase in the initial mean droplet diameter is shown in Figure 15. Similarly,
the variation in the ratio of gas phase reaction time to gas residence time is shown in
Figure 16. For smaller initial mean droplet diameters (<80 µm), the droplet evaporation
time is less than the droplet residence time (as shown in Table 2) but reaction time is
higher than gas residence time. So for small initial mean droplet diameters (<80 µm),
combustion is chemically controlled and the lean blowout of flame occurs because gas
residence time is less than the time required to complete the chemical reaction. This
could be the reason for the occurrence of lifted flame and subsequent lean blowout in
the case of smaller initial mean droplet diameters as observed in Figure 10. For higher
initial mean droplet diameters (>80 µm), the gas residence time is adequately higher
than the reaction time; however, the droplet evaporation time is higher than the
corresponding droplet residence time. So, in the case of higher initial mean droplet
diameters (>80 µm), the lean blowout of flame occurs due to droplet residence time
being less than the time required for the evaporation of droplet (evaporation-controlled
lean blowout). This shift in the mode of lean blowout from reaction to evaporation
10
Droplet evaporation time/ Droplet residence
6
time
Figure 15. Variation of ratio of droplet evaporation time to droplet residence time with respect to
initial mean droplet diameter at LBO condition.
1044 G. MUTHUSELVAN ET AL.
2.75
2.5
Large
reaction
Figure 16. Variation of ratio of reaction time to gas residence time with respect to initial mean droplet
diameter at LBO condition.
controlled regime with increase in initial mean droplet diameter is already observed in
Figure 10 as occurrence of lifted flame before blowout in the case of small mean
droplet sizes. Such a shift in the mode of lean blowout is similar to the computational
results of Awasthi et al. (2013). Variation in the ratio of gas mixing time to the acoustic
time with respect to initial mean droplet diameter is shown in Figure 17. Since the
acoustic time does not significantly influence the condition of lean blowout, this ratio
does not change monotonically with respect to the initial mean droplet diameter. It is
evident that while the sound level is significantly influenced by the average droplet size,
especially close to the LBO limit, the acoustic time scale does not affect the LBO
phenomena in the cases studied here. In the present small scale combustor facility, for
the operating conditions studied, the mode change from evaporation delay controlled
lean blowout to gas phase reaction rate based LBO is seen to occur around the drop
size of 80 µm.
The LBO data measured in the present study corresponds to the ambient operating
conditions (T3 = 300 K, P3 = 1bar). In order to estimate the evaporation delay based
critical droplet diameter at typical gas turbine operating conditions, the droplet evapora-
tion time and droplet residence time are calculated at higher pressure and temperature
conditions (T3 = 500 K, P3 = 5.5 bar) using Eq. (2) for different initial mean droplet
diameters. Variations in the ratio of droplet evaporation time to droplet residence time
with respect to increase in the initial mean droplet diameter are shown in Figure 18. For
initial mean droplet diameters less than 150 µm, the droplet evaporation time is less than
the droplet residence time. The critical droplet diameter for evaporation controlled regime
which was evaluated as 80 µm at ambient operating conditions, now increases to 150 µm
at higher operating conditions. However, in order to fully understand the LBO phenom-
enon in large-scale gas turbine combustors, mixing processes in the presence of recircula-
tion zones, gas phase turbulence and other phenomena may have to be taken into account,
in a more comprehensive way.
COMBUSTION SCIENCE AND TECHNOLOGY 1045
0.7
0.6
0.4
0.3
0.2
0.1
0
40 80 120 160 200 240 280 320 360
Initial droplet diameter (micron)
Figure 17. Variation of ratio of gas mixing time to acoustic time with respect to initial mean droplet
diameter at LBO condition.
4
Droplet evaporation time/ Droplet residence time
3.5
2.5
1.5
1 High
evaporation
0.5 delay
0
50 75 100 125 150 175 200 225 250 275 300 325 350
Initial droplet diameter (micron)
Figure 18. Variation of ratio of droplet evaporation time to droplet residence time with respect to
initial mean droplet diameter at LBO condition at ambient conditions of T3 = 500 K and P3 = 5.5 bar.
mean droplet diameter and useful correlations for the LBO limits have been derived based
on evaporation delay and gas phase reaction delay concepts.
The frequency of acoustic oscillations locks-in with the quarter wave frequency of the
combustor duct for all initial mean droplet diameters considered. Although the acoustic lock-
in frequency is nearly constant in the regime of sustained combustion, the frequency reduces
close to the lean blowout limit, due to lower combustion gas temperature at leaner mixture
conditions.
The results indicate that the atomization quality greatly influences the mode of lean
blowout. The higher evaporation time of larger droplets results in accumulation of large
volume of vapor prior to combustion. Such vapor accumulation followed by rapid
combustion results in high amplitude acoustic oscillations. Hence, the amplitude of
oscillations increases with the initial mean droplet diameter. The lean blowout of flame
occurs due to droplet residence time being lower than the time required to complete the
evaporation of droplet, in the case of larger droplets, giving rise to the evaporation
controlled lean blowout.
However, in the case of small initial mean droplet diameters (<80 μm), the droplets evaporate
faster and form a partially premixed reactant mixture with the recirculating air. Here, a partially
premixed lifted flame occurs which sustains combustion even at relatively lower equivalence
ratios. With further increase in air flow rate, the lean blowout of flame occurs, which is
attributed to gas residence time being less than gas phase reaction time for combustion.
Six time scales corresponding to droplet evaporation, droplet residence, gas phase reac-
tion, duct acoustics, gas residence and gas mixing, are correlated to the LBO limits. Among
these, gas mixing time and duct acoustic time do not significantly affect the condition of LBO
and a simplified correlation with the other four time scale is also obtained. The correlation
results match with the experimental results with R2 value close to unity. Thus, the experi-
mental results of the present work clearly illustrate the major role played by atomization
quality on the lean blowout limits and acoustic oscillations for a swirl spray burner.
Acknowledgments
The authors are thankful to the Director, CSIR-NAL for permitting to publish this work and the
staff members of the combustion laboratory, propulsion division, CSIR-NAL.
Nomenclature
AFR – Air fuel ratio
FN – Flow number
K – Atomizer constant
ma – Air mass flow rate (kg/h)
SMD – Sauter mean diameter (µm) (the diameter of a drop whose ratio of volume to surface
area is the same as that of the spray)
Tg – Temperature of gas
Tevp,drop – Evaporation time of initial droplet diameter
Tres,drop – Residence time of droplet in the burner
Treaction – Reaction of time of fuel air mixture
Tacoustic – Acoustic time of the burner
Tres,air – Residence time of air in the burner
Tmix,air – Mixing time of air
COMBUSTION SCIENCE AND TECHNOLOGY 1047
ORCID
Govindaraj Muthuselvan http://orcid.org/0000-0003-2027-5207
References
Awasthi, I., Gogos, G., and Sundararajan, T. 2013. Effects of size on combustion of isolated
methonal droplets. J. Combust. Flame, 160, 1789–1802. doi:10.1016/j.
combustflame.2013.03.023
Ballal, D.R., and Lefebvre, A.H. 1980. Weak extinction limits of turbulent heterogeneous fuel/air
mixtures. J. Eng. Gas Turbine Power, 102, April.
Baxter, M.R., and Lefebvre, A.H. 1992. Weak extinction limits of large scale flame holders. J. Eng.
Gas Turbine Power, 114, 776–782. doi:10.1115/1.2906656
Bossard, J.A., and Peck, R.E. 1996. Droplet size distribution effects in spray combustion.26th
symposium on combustion, The combustion institute, pp. 1671–1677.
Chin, J.S., and Lefebvre, A.H. 1982. Effective values of evaporation constant for hydrocarbon fuel
drops. Proceedings of 20th Automotive technology development contractors meetings, p.
120,325–331, Society of automotive engineers, Warrendale PA.
Chishty, W.A., Lepera, S.,D., and Vandsburger, U. 2011. Spray combustion dynamics under
thermoacoustic oscillations. Int. J. Mech. Aerosp. Ind. Mechatron. Manuf. Eng., 5(1).
Cohen, J.M., Rey, N.M., Jacobson, C.A., and Anderson, T.J. 1999. Active control of combustion
instability in a liquid – fuelled low NOx combustor. J. Eng. Gas Turbine Power, 121, April, 1999,
281. doi:10.1115/1.2817118
de la Cruz Garcia, M., Mastorakos, E., and Dowling, A.P. 2009. Investigations on the self-excited
oscillations in a kerosene spray flame. J. Combust. Flame, 156, 374–384. doi:10.1016/j.
combustflame.2008.11.018
Duvvur, A., Chaing, C.H., and Sirignano, W.A. 1996. Oscillatory fuel droplet vaporization: driving
mechanism for combustion instability. J. Propul. Power, 12(2), March-April, 358–365.
doi:10.2514/3.24036
Gireesh Kumaran, T., and Sujith, R.I. 2015. Intermittent burst oscillations: signature prior to flame
blowout in a turbulent swirl-stabilized combustor. J. Propul. Power, 31(6), 1661–1671.
doi:10.2514/1.B35526
Griebel, P., Boschek, E., and Jansohn, P. 2007. Lean blowout limits and NOx emissions of turbulent,
lean premixed, hydrogen-enriched methane/air flames at high pressure. J. Eng. Gas Turbine
Power, 129, April, 404–410. doi:10.1115/1.2436568
Jones, A.R. 1982. Design optimization of a large pressure-jet atomizer for power plant. Proceedings of the
2nd International conference on liquid atomization and spray systems, Madison, Wis., pp. 181–185.
Lefebvre, A.H. 1985. Fuel effects on gas turbine combustion, ignition, stability and combustion
efficiency. J. Eng. Gas Turbine Power, 107, January, 1985, 24. doi:10.1115/1.3239693
Lefebvre, A.H. 1989. Atomization and Sprays, CRC press, Hemisphere Washington, DC.
Lei, S., and Turan, A. 2010. Chaotic modelling and control of combustion instabilities due to
vaporization. Int. J. Heat Mass Transf., 53, 21–22, October 2010, 4482–4494. doi:10.1016/j.
ijheatmasstransfer.2010.06.045
Levy, Y., and Bulzan, D.L. 1995. On the oscillation of combustion of a laminar spray. J. Combust.
Flame, 100, 543–549. doi:10.1016/0010-2180(94)00135-F
Li, H., Zhou, X., Jeffries, J.B., and Hanson, R.K. 2007. Sensing and control of combustion instabil-
ities in swirl-stabilized combustors using diode-laser absorption. AIAA J., 45(2), Feb, 390–398.
doi:10.2514/1.24774
1048 G. MUTHUSELVAN ET AL.
Lieuwen, T. 2005. Online combustor stability margin assessment using dynamic pressure data.
J. Eng. Gas Turbine Power, 127, July, 2005, 478. doi:10.1115/1.1850493
Lieuwen, T., McDonell, V., Petersen, E., and Santavicca, D. 2008. Fuel flexibility influences on
premixed combustor blow-out, flashback, auto ignition and stability. J. Eng. Gas Turbine Power,
130, January, 2008, 011506. doi:10.1115/1.2771243
Lieuwen, T., Torres, H., Johnson, C., and Zinn, B.T. 2001. A mechanism of combustion instability
in lean premixed gas turbine combustor. J. Eng. Gas Turbine Power, 123, January, 2001, 182.
doi:10.1115/1.1339002
Mcmanus, K.R., Magill, J., and Michael, F.M. 1998.Combustion instability suppression in liquid-fueled
combustors. AIAA 98-0642, 36th AIAA Aerospace sciences meeting and exhibt.
Muthuselvan, G., Ghate, K.D., Muralidhara, H.S., Venkat, S.I., Sundararajan, T., and Srinivasan, K.
2018. Experimental study of spray breakup phenomena in small scale simplex atomizer with and
without air swirl. J. Atomization Sprays, 28(4), 299–320. doi:10.1615/AtomizSpr.2018021190
Nicoli, C., Haldenwang, P., and Suard, S. 2007. Effect of substituting fuel spray for fuel gas on flame
stability in lean pre-mixtures. J. Combust. Flame, 149, 295–313. doi:10.1016/j.
combustflame.2006.12.018
Polymeropoulos, C.E., and Das, S. 1975. The effect of droplet size on the burning velocity of
kerosene-air sprays. J. Combust. Flame, 25, 247–257. doi:10.1016/0010-2180(75)90091-7
Samadhan, A.P., Vishnu, R., Vadivukkarasan, M., Panchagnula, M.V., and Sujith, R.I. 2016.
Intermittency route to combustion instability in a laboratory spray combustor. J. Eng. Gas
Turbine Power, 138, April, 2016.
Sturgess, G.J., Heneghan, S.P., Vangsness, M.D., Ballal, D.R., Lesmerises, A.L., and Shouse, D. 1993.
Effects of back –pressure in a lean blow-out research combustor. J. Eng. Gas Turbine Power, 115,
486–498. doi:10.1115/1.2906735
Tongxun, Y., and Ephraim, J.G. 2007a. Combustion instabilities and control of a multiswirl atmo-
spheric combustor. J. Eng. Gas Turbine Power, 129, 31–37. doi: 10.1115/1.2181595
Tongxun, Y., and Gutmark, E.J. 2007b. Real-time prediction of incipient lean blowout in a partially
premixed, liquid-fueled gas turbine combustor. AIAA J., 45(7), July, 1734–1739. doi:10.2514/1.25847
Venkataraman, K.K., Preston, L.H., Simons, D.W., Lee, B.J., and Santavicca, D.A. 1999. Mechanism
of combustion instability in a lean premixed dump combustor. J. Propul. Power, 15, 6, Nov-Dec.
doi:10.2514/2.5515
Xie, F., Huang, Y., Hu, B., and Wang, F. 2012. Improved semi-empirical correlation to predict lean
blowout limits for gas turbine combustors. J. Propul. Power, 28(1), 197–203. doi:10.2514/1.B34296
Yu, K.H., and Wilson, K.J. 2002. Scale-up experiments on liquid-fueled active combustion control.
J. Propul. Power, 18(1), Jan-Feb, 53–60. doi:10.2514/2.5897
APPENDIX – A
This appendix presents the derivation of six different time scales used in the development of
empirical correlation for the condition of lean blowout.
Due to tangential entry of fuel into swirl chamber of simplex atomizer, low-pressure zone is created
near axis of atomizer and as a consequence, an air core is formed along the center line. The air core
size, in turn, determines the effective flow area for liquid stream and the coefficient of discharge.
Ratio of the air core area and nozzle orifice exit area is expressed as
Aa
X ¼ (A1)
Ao
where
Aa is the area of the air core present along center line of atomizer,
Ao is the nozzle orifice exit area, and
COMBUSTION SCIENCE AND TECHNOLOGY 1049
X is the ratio of the air core area and nozzle orifice exit area.
Velocity of liquid film emanating from orifice exit of atomizer is derived from mass continuity and
is expressed as
mf
U¼ (A2)
ρl ðAo Aa Þ
where
U – velocity of liquid film emanating from orifice exit of atomizer,
mf – fuel flow rate, and
ρ1 – liquid density.
Ratio of the air core area and nozzle orifice exit area (X) and the coefficient of discharge is related
(Lefebvre. A.H 1989)
!
ð1 XÞ3
Cd ¼ 0:5 (A3)
ð1 þ X Þ
where
where
Table A1. Variation of droplet velocity for a range of fuel flow rates and different atomizer
configurations.
Velocity of droplet Ud (m/s)
Fuel flow rate (kg/h) K = 0.132 K = 0.1769 K = 0.3681 K = 0.53 K = 0.6545
9 13.312 12.129 9.585 8.558 8.020
10 14.831 13.480 10.668 9.524 8.921
11 16.337 14.846 11.747 10.492 9.828
12 17.846 16.208 12.837 11.456 10.734
13 19.360 17.580 13.921 12.418 11.645
14 20.878 18.956 15.006 13.384 12.543
15 22.369 20.315 16.094 14.363 13.450
1050 G. MUTHUSELVAN ET AL.
from the atomizer orifice. Table A1 shows the analytically calculated velocity of droplet for a range of
fuel flow rates and five different atomizer configurations representing different values of K.
Droplet residence time is defined as the time available for droplet to stay in the burner, which is
related to length of burner by
Lb 15
Tres; drop ¼ (A5)
Ud
where
Tres, drop
– droplet residence time and
Lb 15 – 1/5 of length of the burner (estimated as the combustion zone length within the
burner).
2. Droplet evaporation time (Tevp)
Droplet evaporation time is defined as the time required for droplet to completely evaporate in the
burner, which is expressed by
Do 2
Tevp ¼ (A6)
λeff
where
Tevp – evaporation time of droplet,
Do – initial diameter of droplet (= SMD), and
λeff – evaporation constant.
Empirical Eq. (A7) (Chin, J.S et al. 1982) used to calculate the evaporation constant in the present
analysis as shown below
K ð1 þ 0:22 Redo 0:5 Þ
λeff ¼ 8 glnð1 þ BÞ (A7)
Cp ρl
The transfer number (B) and Reynolds number in Eq. (A7) are mentioned in the form of Eqs. (A8)
and (A9), respectively,
Tg
3 Tb Cp g
B¼ (A8)
hfg
Ur Do ρg
Redo ¼ (A9)
μg
where
Kg – thermal conductivity of gas,
Cp g – specific heat of gas,
B – transfer number,
Redo – Reynolds number of droplet,
Tg – adiabatic flame temperature,
Tb – boiling temperature of fuel,
hfg – fuel property calculated for jet A, and
ρg – density of gas.
The gas phase velocity used in Eq. (A10) is calculated using Eq. (A11), which relates gas phase
velocity with air mass flow rate:
COMBUSTION SCIENCE AND TECHNOLOGY 1051
ma LBO
Ug ¼ (A11)
ρg Ab
where
½Xf
Treaction ¼ (A13)
r
where
r – reaction rate,
A – pre-exponential factor = 3.8 X 1011,
Ru – activation temperature = 15098 K,
Ea
m – 0.25,
n – 1.5, and
Xf – molar concentration of fuel.
4. Residence time of gas flow (Tres, gas)
Residence time of gas flow is defined as time available for the gas flow to stay in the burner, which is
related to length of burner by the following equation:
1.00E-06
Evaporation constant (m2/s)
9.50E-07
9.00E-07
8.50E-07
8.00E-07
7.50E-07
40 80 120 160 200 240 280 320 360
Initial droplet diameter (micron)
Figure A1. Variation of evaporation constant with respect to different initial mean droplet diameter.
1052 G. MUTHUSELVAN ET AL.
Tg
3 Tb Cp g
B¼ (A14)
hfg
where
Vg – volume flow rate of gas and
Volb ð1Þ – 1/5 of volume of the burner (= volume of the combustion zone).
5
The volume flow rate of gas is related to the mass flow rate of gas and density of gas by Eq. (A15).
mg
Vg ¼ (A15)
ρg
where
mg – mass flow rate of gas = mf+ ma.
5. Acoustic time (Tacoustic)
The acoustic time of the burner is calculated for both attached and lifted flame as follows:
Lb
Tacoustic ¼ ðattachedflamesÞ (A16)
C
Lb Ug Treaction
Tacoustic ¼ ðliftedflamesÞ (A17)
C
where
Mixing time of air flow (Lefebvre, 1985) is defined as the time required for the air flow to mix with the
gas in the burner for flame stabilization, which is related to the mixing rate of burner by Eq. (A18).
The mixing rate between air and surrounding gas is related to the density gradient in Eq. (A19)
ρg Vb ð1Þ
Tmix; air ¼ 5
(A18)
Mixingrate
Mixing rate ¼ UJ;sw Asw ρa ρg (A19)
ma LBO
Uj;sw ¼ (A20)
Cd;sw ρa Asw
where
Lb ðUg Treaction Þ – geometric flow area of swirler,
Tacoustic ¼ C
ASW – geometric flow area of swirler,
UJ,sw – velocity of jet discharge from swirler, and
Cd,sw – discharge coefficient of swirler.