Applied Energy 211 (2018) 513–522

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Fostering distributed combustion in a swirl burner using prevaporized liquid T

fuels
⁎
Ahmed E.E. Khalil, Ashwani K. Gupta
Department of Mechanical Engineering, University of Maryland, College Park, MD 20742, USA

H I G H L I G H T S

• Examined the role of entrainment on Distributed Combustion with JP-8 and ethanol.
• NLowandO CO
2 are used to simulate entrained combustion gases from within the combustor.
2

• Reactionconcentration
2 in the mixture prior to ignition fostered distributed reaction.
• 95% NO distribution fostered at 9.2% O and lower, with almost invisible flame.
2

• reduction demonstrated at distributed combustion with minimal impact on CO.

x

A R T I C L E I N F O A B S T R A C T

Keywords: Colorless Distributed Combustion (CDC) has presented itself as an environmentally friendly combustion method
Colorless distributed combustion with significant benefits on ultra-low emissions, uniform thermal field (pattern factor), reduced noise, mitigation
Liquid fuels of instability and enhanced flame stability. CDC requisites include controlled entrainment of hot reactive species
Reaction distribution from within the combustor and their subsequent mixing with fresh reactants to form a low-oxygen concentration
Ultra low NOx
high-temperature oxidizer prior to ignition. This mixture results in distributed reaction over the entire com-
Gas turbine combustion
bustion volume. To date, most of the CDC research efforts have focused on gaseous fuels only. In this paper,
conditions fostering distributed combustion using JP-8 and alternative ethanol fuels. Data revealed that tran-
sition to CDC can be achieved at oxygen concertation of approximately 9.5% for both the fuels. This oxygen
concentration was determined based on reaction field uniformity as identified from OH∗ chemiluminescence
signatures. Under distributed combustion, emissions were substantially reduced by some 95% to result in NOx
emissions of less than 2 PPM with minimal impact on CO emission from both JP-8 and ethanol fuels at an
equivalence ratio of 0.9, with even lower emissions at lower equivalence ratios. Combining the data obtained
with liquid fuels with those from gaseous fuels revealed that, regardless of the fuel used, the oxygen con-
centration at which CDC prevailed can be predicted based on mixture temperature within a range of 0.75%. This
knowledge enables designing a combustor to achieve CDC regardless of the fuel used. The data are useful in the
design and development of fuel flexible CDC for high combustion intensity gas turbine applications.

1. Introduction
As energy awareness increase among societies, with desire for more
environmentally friendly energy conversion methods, research efforts
continue to focus on satisfying the energy needs with minimal impact
on the environment. Although significant achievements in renewable
energy have been accomplished, traditional power sources, including
fossil fuels, are expected to remain a major player in the future energy
conversion devices [1]. The Annual Energy Outlook predicts that the
share of petroleum and other liquid fuels will remain almost constant,
while natural gas’ share is expected to increase [2]. This presence of
fossil fuel in the worlds energy mix have motivated combustion en-
gineers to search for new and novel ways to convert the chemical en-
ergy stored in fuels into useful heat energy through new and novel
combustion techniques. Since the 1990s, different research groups have
proposed combustion methods that minimize combustion emissions
while maintaining overall performance. Among these methods are rich-
quench-lean burn (RQL) [3–5], flameless oxidation (FLOX) [6–8],
moderate or intense low oxygen dilution (MILD) [9–11], and colorless
distributed combustion (CDC) [12,13].
Colorless Distributed Combustion (CDC), which is the focus of this
paper, is based on the concept of HiTAC [14], wherein low oxygen

⁎
Corresponding author.
E-mail address: akgupta@umd.edu (A.K. Gupta).

https://doi.org/10.1016/j.apenergy.2017.11.068
Received 1 July 2017; Received in revised form 10 November 2017; Accepted 13 November 2017
Available online 20 November 2017
0306-2619/ © 2017 Elsevier Ltd. All rights reserved.
A.E.E. Khalil, A.K. Gupta
concentration air preheated to high temperatures, is used for combus-
tion. The use of this hot diluted oxidizer has demonstrated that the
temperature of combustion gases in the furnace was only about
50–100 °C higher than that of the preheated low oxygen concentration
air just prior to ignition. In CDC, the low oxygen concentration high
temperature oxidizer is achieved through entrainment and recirculation
of hot reactive species from within the combustor and their subsequent
mixing with the freshly introduced air. Providing adequate entrainment
and mixing prior to ignition was shown to be critical to achieve dis-
tributed reaction combustion. As distributed reactions were achieved, a
lower reaction rate was attained over the entire volume of the com-
bustor, as opposed to the concentrated thin reaction flame front with
high reaction rates that also produce local hot spots. This distribution
over a larger volume results in the same fuel consumption, as that of
concentrated flame fronts, with lower temperature rise in the com-
bustor. The low reaction rates were achieved through lowering of the
oxygen concentration in the reactants, and increased temperature of the
reactants, which was achieved simultaneously through the entrainment
of hot reactive gases from within the combustor.
CDC not only helps avoid the formation of thin reaction zone but
also mitigates the formation of hot-spots in the flame that help mini-
mize thermal NOx formation and emission from the Zeldovich thermal
mechanism [15]. The benefits of distributed combustion in terms of
pollutants emission have been demonstrated using a variety of geo-
metries, temperatures, pressures, and injection methods [12,13,16,17].
Moreover, the conditions required to achieve CDC have been in-
vestigated in detail, with focus on the oxygen concentration required to
foster distributed combustion [18]. The role of various input and op-
erational parameters (oxygen concentration, temperature, and global
equivalence ratio) on thermal field and local equivalence ratio uni-
formity were examined [19,20], and the different flame behavior from
the flame-flow interaction along with the role of integral length scale
and flame thickness on distributed combustion [21] have been in-
vestigated. However, these investigations focused on methane [18–21],
propane and hydrogen enriched methane [22] as the fuel.
With the increased demand in power and propulsion for gas turbines
that operate on liquid fuels, we focus our efforts to investigate the
conditions that foster CDC with liquid fuels. Achieving low emissions
with liquid fuels have been examined by different groups that has been
called MILD and Flameless Combustion [23–25]. In their work, Ye et al.
examined prevaporized ethanol, acetone, and heptane in a reverse flow
combustor, wherein they found that ethanol burns well under MILD
conditions while acetone and heptane demonstrated some instabilities
[24]. Reddy et al. showed that achieving flameless combustion with
liquid fuels at ambient conditions is possible in a preliminary study
using a two-stage combustor [25]. However, sometimes operating with
premixed and prevaporized fuel can induce instabilities. These studies
used phase-averaged particle image velocimetry (PIV) that outlined a
range of equivalence ratio in the TAPS (twin annular premixed swirler)
combustor that could operate without entering a “growl” regime [26].
To enhance liquid fuel stability, it was suggested to add hydrogen to JP-
8 flame, leading to stabilizing the flame, and reducing the flame lift off.
The authors suggested hydrogen addition as an effective way to reduce
unburned hydrocarbons as well [27]. Focus on JP-8 combustion was
not only limited to combustion in gas turbines, but extended to include
laminar burning speeds at high temperature and pressures as well as
with diluents [28,29]. JP-8 performance with diluents is critical to
understand the possibility of achieving CDC with liquid fuels. The au-
thors concluded that laminar burning speeds of JP-8 increased with
increase in temperature and decreased with increase in pressure [28].
They also found that addition of diluents did not had a major effect on
stability but decreased the laminar burning speeds due to decreased
adiabatic flame temperature [29]. This behavior of lower flame speeds
upon dilution is favorable to achieve CDC [21].
Liquid fuel interests are not only limited to JP-8. Bio-fuels are also of
interest as they offer a good viable solution for the continuous decrease
514
Applied Energy 211 (2018) 513–522
in fossil-fuel reserves as well as the environmental concerns about
carbon emission. The term bio-fuel is referred to fuels produced from
bio-materials, such as, biomass and wastes. Such fuels include, but not
limited to, bio-methanol, bio-ethanol, bio-diesel pyrolysis oil, biogas,
synthetic gas, and hydrogen [30].
Ethanol has been examined for gas turbines applications, and the
results show favorable properties as a sustainable and renewable fuel.
However, due to its lower lubricity and lower heating value, some
changes are required in the fuel handling system for retrofitting in gas
turbines [31]. Using biofuels, as opposed to traditional fossil fuels, is
carbon neutral so that its use reduces the net amount of carbon dioxide
emitted into the atmosphere. Experimental and numerical studies on
the combustion and oxidation of ethanol has been performed that
provided numerous new data to support modeling studies on the oxi-
dation and combustion of ethanol [32]. Ethanol has been also in-
vestigated as an alternate fuel for diesel engines [33]. Ethanol has also
been studied in detail with focus on laminar burning speed and ex-
tinction properties and their dependency on temperature [34–36].
In this paper, the conditions fostering distributed combustion have
been examined using liquid fuels as opposed to previous efforts that
focused on gaseous fuels [22]. Focus is on the oxygen concentration at
which the reaction regime transforms into volumetric distributed
combustion mode along with ultra-low emissions. It is also desired to
compare the performance of liquid fuels with gaseous fuels to assess the
fuel flexibility in distributed combustion. The primary focus is on de-
termining if the role of fuel has any impact on the oxygen concentration
at which distributed combustion occurs.
2. Experimental facility
The experiments were performed using a swirl burner. Details of
this swirl burner can be found elsewhere [37]. Different amounts of
nitrogen and carbon dioxide (mixed in a 90%-10% ratio by volume)
were mixed with air upstream of the burner, thus simulating product
gas entrainment with lower oxygen concentration in the mixture prior
to ignition. To start the burner, methane was injected in the air line.
Once the burner was stabilized, the fuel was gradually switched to li-
quid fuel (as vapors) until the burner completely operated with the li-
quid fuel. To control the gases, differential pressure laminar flow con-
trollers were used with an uncertainty of 0.2% of full scale +0.8% of
set point for air and Nitrogen. Thermo-gravimetric flow controllers
were used for other gases, with uncertainty of 1% of full scale (methane
for startup), and 1.5% of full scale (carbon dioxide). For liquid fuels, a
liquid fuel positive displacement pump (VICI M50) was used to furnish
the required flowrates with a precision of ± 0.5%.
It is to be noted that both fuels examined, JP-8 and ethanol, have a
much higher laminar flame velocity as compared to that of methane
[38]. This can cause instabilities as the swirl burner used in the ex-
periments was designed for gaseous fuels. The liquid fuel was pre-
vaporized and this fuel will have a large density as compared to the
gaseous fuel leading to lower flow velocities across the injection holes.
This low injection velocity, coupled with the higher burning velocity of
the liquid fuels at higher temperatures as they were injected pre-
vaporized, resulted in more propensity to flashback towards the injec-
tion location. This scenario will lead to flame oscillation between the
injection point and the swirler downstream which was to be avoided.
This instability was mitigated by adding carbon dioxide to the liquid
fuel line that effectively increased the flow velocity through the liquid
fuel injection ports, and helped stabilize the flame. Carbon dioxide
addition also helped to mitigate any coking or cracking of the liquid
fuels. The added carbon dioxide to the fuel line increased the flow
velocity significantly in the fuel feed lines, lowering the residence time
in the lines, to less than 1 s. This minimized any thermal decomposition.
Widegren et al. showed that for 1% of the fuel (Jet A, which is similar to
JP-8) to decompose at 450 °C, the residence time shall be less than 1
min [39], with similar results shown by others at 375 °C [40].
A.E.E. Khalil, A.K. Gupta
Injection Temperature
Quartz Thermocouple
Confinement
Liquid Fuel
0
Injection 45 Swirler
Air + N2/CO2 Air + N2/CO2
A was conducted via OH∗ chemiluminescence intensity distribution across
Prevaporized liquid fuel + CO2
Fig. 1. Schematic of the swirl burner.
For pre-vaporization, the liquid fuel line (liquid fuel + carbon di-
oxide) was heated using heating tapes, while air, nitrogen, and carbon
dioxide mixture were heated using an air preheater. The burner body
was also heated using heating tape. This prevented the liquid fuel from
condensing inside the burner after it was vaporized in the liquid fuel
line. The heating tapes and air preheater were controlled to keep the
temperature at the injection point at 700 K ± 10 K. The added CO2 also
helped in reducing any thermal decomposition as it increased the flow
velocity in the fuel line leading to a residence time of about 1 s, which
resulted in almost no decomposition at the pre-vaporization tempera-
ture of about 700 K [39].
The burner was confined in the quartz tube having an internal
diameter of 3 inches and length of 12 inches. Thermocouples were
fitted along the reactor walls to confirm the existence of reaction zone
at low oxygen concentration as the flame was almost invisible (color-
less) with almost no combustion noise. Fig. 1 shows a schematic of the
swirl burner.
Detailed investigations on the overall emissions from the combustor
have been performed for the various experimental conditions. The
concentration of NO was measured using a NO–NOx chemiluminescent
gas analyzer; CO concentration was measured using the non-dispersive
infrared method and O2 concentration (used to correct the NO and CO
emissions at standard 15% oxygen concentration) was measured using
galvanic cell method. The uncertainty of the sampling system is esti-
mated to be about ± 0.5 PPM for NO and ± 10% for CO emission. For
imaging the OH∗ chemiluminescent intensity distribution, an ICCD
camera coupled to a narrow band filter, centered at wavelength 307 nm
wavelength, was used. The obtained images were then processed using
MATLAB to obtain different distribution characteristics. Fig. 2 shows
the experimental facility along with diagnostics.
3. Experimental conditions
The experimental investigations reported here were focused on the
reaction zone behavior using OH∗ chemiluminescence, as well as the
resultant pollutants emission. For each experiment, the fuel flow rate,
along with carbon dioxide in the fuel line, was kept constant, to result
in a constant thermal load. The air flow rate was changed to change the
515
Applied Energy 211 (2018) 513–522
equivalence ratio, and for each equivalence ratio, the diluents flow
rates were changed to decrease the oxygen concentration. The heat load
was kept constant at 3.25 KW to give a heat (thermal) release intensity
of 2.4 MW/m3-atm. The experiments were performed at an injection
temperature of 700 K ± 10 K to insure complete fuel vaporization and
no liquid fuel condensation. The percentage values given here are based
on volumetric basis. Two fuels were examined, JP-8 and Ethanol.
Table 1 gives the fuel properties while Tables 2 and 3 give the flow
conditions for each fuel.
The amount of carbon dioxide added to the fuel was the minimum
amount required for stable operation of the burner. Experiments were
performed at higher amounts (up to 125% of the minimum amount
needed). These experiments provided no impact of the additional
amount of carbon dioxide in the fuel line on overall flame behavior and
emissions.
4. Results and discussion
4.1. OH∗ chemiluminescence with JP-8
The initial investigations focused on quantifying the reaction zone
distribution under different oxygen concentrations using JP-8 fuel. This
the burner. Fig. 3 shows characteristic distribution for JP-8.
From the figure, one can note that as the oxygen concentration is
decreased, due to N2 and CO2 addition, the OH∗ intensity distribution
changes from being concentration at the swirl lobes to a more uniform
distribution all over the examined combustion volume. This trend is
similar to that obtained before using gaseous fuels (methane, propane,
and hydrogen enriched methane) [18,19]. To further distinguish the
behavior at lower oxygen concentrations, Fig. 4 shows the OH∗ che-
miluminescence intensity for the low oxygen concentration at a scale
one quarter of that shown in Fig. 3.
From Fig. 4, one can see that the intensity gradients drop sig-
nificantly at oxygen concentration of 9.92%, 10.54%, 11.6%, 12.88%
for equivalence ratios of 0.9, 0.8, 0.7, and 0.6, respectively. The OH∗
intensity for the above figures was also analyzed to examine the signal
distribution across the examined burner area (viewed area of the flame
zone). For each image, the mean and standard deviation were calcu-
lated and scaled by the maximum and mean, respectively. The mean to
the maximum shows how close the mean is to the maximum so that
higher value represents a more distributed reaction zone. In addition,
the standard deviation to the mean shows how well this mean is re-
presentative of the data; a smaller standard deviation indicates that the
values are all closer to the mean, as opposed to a large standard de-
viation. In addition, this scaling allows comparisons across different
equivalence ratio (as they have different maxima). Fig. 5 shows the
mean to maximum ratio, while Fig. 6 shows the standard deviation to
mean ratio.
The results show a good similarity trend between what was ob-
served on OH∗ intensity images (Fig. 4) and that shown in Figs. 5 and 6.
The reaction zone seemed distributed once the mean to maximum ratio
was higher than approximately 35% (see Fig. 5, where O2% of 9.92,
10.54, and 11.6 seemed to reveal good uniformity in the reaction zone
that represent equivalence ratios of 0.9, 0.8, and 0.7, respectively). At
equivalence ratio of 0.6, this value was higher, reaching a value of 40%.
As for the standard deviation to mean ratio, the data shows that a ratio
of approximately less than 0.7 seems to be the differentiating edge for
the aforementioned oxygen concentrations at said equivalence ratios,
except for phi = 0.6, wherein that value was closer to 0.6.
4.2. OH∗ chemiluminescence with ethanol
The reaction zone distribution at different oxygen concentrations
using ethanol fuel is now presented. The OH∗ chemiluminescence in-
tensity distribution were obtained across the burner. Fig. 7 shows this
A.E.E. Khalil, A.K. Gupta Applied Energy 211 (2018) 513–522

Fig. 2. Schematic of the experimental facility.

Table 1 OH∗ distribution using ethanol as the fuel.

Fuel Properties. From Fig. 7, one can note that decrease in the oxygen concentration,
due to N2 and CO2 addition, changed the OH∗ intensity distribution
Fuel Calorific value Density [kg/ Volume Flow A/F Stoichiometric
[kJ/kg] m3] rate
from being concentrated at the swirl lobes to a more uniform dis-
tribution all over the examined volume. This trend was similar to that
JP-8 ∼42800 ∼807 5.64 15 was observed before using JP-8; however, stable operation with ethanol
Ethanol ∼29700 ∼790 8.31 8.95 was possible at lower oxygen concentration than that observed with JP-
8 fuel. For instance, at equivalence ratio of 0.9, stable operation of JP-8
was observed till an oxygen concentration of 9.18%. This was extended
to 8.4% with ethanol. This can be attributed to the fuel oxygen content
Table 2 which helps stabilize the flame at lower oxygen concentration. Similar
Experimental parameters for JP-8.
to what was shown with JP-8, better differentiation between flame
JP-8 CO2 in Equivalence Air [l/ N2 [l/ CO2 [l/ Oxygen Con.
behavior at low oxygen concentration can be observed at a scale one
flow fuel line Ratio min] min] min] in oxidizer quarter of that shown in Fig. 7. This behavior can be seen from the data
[ml/ [l/min] [%] presented in Fig. 8.
min] Fig. 8 shows that the intensity gradients drop significantly at oxygen
5.64 4 0.9 64.75 0–75 0–8.32 21–9.18
concentration of 9.54%, 10.15%, 10.8%, 12.11% for equivalence ratios
0.8 72.8 0–75 0–8.32 21–9.8 of 0.9, 0.8, 0.7, and 0.6, respectively, which can be viewed as an in-
0.7 83.25 0–65 0–7.22 21–11.25 dicator of achieving distributed combustion. Similar to the analysis
0.6 97.1 0–60 0–6.66 21–12.45 performed on JP-8, Fig. 9 shows the mean to maximum ratio, while
Fig. 10 shows the standard deviation to mean ratio.
The results support a similarity in trend between what was observed
on OH∗ intensity images (Fig. 8) to that shown in Figs. 9 and 10; this
was also demonstrated for JP-8 fuel. The reaction zone seemed dis-
Table 3 tributed once the mean to maximum ratio was higher than 35% (see
Experimental parameters for Ethanol. Fig. 9, where O2% of 9.54, 10.15 and 10.8 represent equivalence ratios
Ethanol CO2 in Equivalence ratio Air N2 [l/ CO2 [l/ Oxygen
of 0.9, 0.8, and 0.7, respectively). As for equivalence ratio of 0.6, dis-
flow [ml/ fuel [l/ min] min] Con. in tributed combustion was not achieved until the higher ratio of 50%. In
min] line [l/ min] oxidizer general, the curve for equivalence ratio of 0.6 is seen to exhibit higher
min] [%] values as compared to that at other equivalence ratios that are some-
8.31 4 0.9 55.5 0–75 0–8.32 21–8.4
what close to each other, see Fig. 9. The same trend for the curve re-
0.8 62.4 0–75 0–8.32 21–9 presenting phi = 0.6 was observed with JP-8 fuel, see Fig. 5. The
0.7 71.4 0–70 0–7.77 21–10 standard deviation to mean ratio data shows that a ratio of approxi-
0.6 83.3 0–60 0–6.66 21–11.66 mately less than 0.6 seems to be differentiating edge for the

516
A.E.E. Khalil, A.K. Gupta Applied Energy 211 (2018) 513–522

Fig. 3. OH* Chemiluminescence at different O2 concentration and equivalence ratios with JP-8 fuel. Note the different scale for each equivalence ratio.

aforementioned oxygen concentrations at the equivalence ratios ex- A comparison of the results from JP-8 and Ethanol fuels reveals that
amined. However, this value was found to be lower than 0.4 for distributed combustion was established at a lower oxygen concentra-
phi = 0.6. tion for ethanol as compared to JP-8. This is attributed to presence of

Fig. 4. OH* Chemiluminescence at low O2 con-

centration and equivalence ratios with JP-8 fuel.
Note the different scale for each equivalence
ratio.

517
A.E.E. Khalil, A.K. Gupta Applied Energy 211 (2018) 513–522

55 Fig. 5. Percentage ratio of mean to maximum OH*

intensity for JP-8 at different oxygen concentrations
50 Phi=0.9 and equivalence ratios examined.
Mean/Maximum Signal [%]

45 Phi=0.8
Phi=0.7
40
Phi=0.6
35
30
25
20
15
10
8 10 12 14 16 18 20 22
Oxygen Concentration [%]

oxygen in the fuel. In addition, the reaction did not seem distributed
until the standard deviation to the mean ratio was below 0.6, as com-
pared to 0.7 for JP-8. On the other hand, the oxygen content in ethanol
allowed for stable operation at lower oxygen concentrations for ethanol
as compared to JP-8 (8.4% vs. 9.18% at equivalence ratio of 0.9).
4.3. Pollutants emission
relatively constant with values around 6 PPM. When the fuel was
switched to ethanol rather than JP-8, CO emissions hovered around 10
PPM at equivalence ratio of 0.9 with slight increase with oxygen con-
centration reduction. For equivalence ratio of 0.7, CO emissions re-
mained around 4 PPM for all oxygen concentration examined. These
measurements indicate that CDC does not have a significant impact on
CO emission. Similar trend was reported before for gaseous fuels [22].

Reaction zone distribution from within the combustor volume is
beneficial in terms of thermal field uniformity. However, one of the
main benefits of CDC is pollutants emission reduction, with up to 90%
reduction in NO and CO have been reported for gaseous fuels. To de-
monstrate this emission reduction with the two different liquid fuels,
pollutants emission measurements were performed for both the fuels
examined and the results obtained are shown in Figs. 11 and 12.
From Fig. 11, one can see that NO emission diminishes rapidly with
decrease in oxygen concentration for all the equivalence ratios. When
distributed combustion was established, NO emissions were reduced by
some 95%. For instance, for JP-8, the NO emissions were reduced from
64 PPM to 2 PPM at equivalence ratio of 0.9. For ethanol, the corre-
sponding value was 1.5 PPM (down from 49 PPM). NOx emissions were
also measured and the data obtained (open symbols in Fig. 11) followed
the same trend as NO emission. It is noteworthy that across the different
experimental points measured, NO emissions were ∼92% of the total
NOx emission (with a minimum of 89% at the high end of emission to
96% at low end of emission). This shows that NO is dominant in the
NOx.
As for CO, the reaction zone distribution had minimal effect on CO
emission since the measured CO emission did not reveal a clear trend.
For JP-8 at an equivalence ratio of 0.9, the CO emission increased with
decrease in oxygen concentration. This trend continued until reaching a
maximum value of 270 PPM at about 12% O2. Afterwards, CO emission
dropped slightly. At equivalence ratio of 0.7, CO emission remained
4.4. Design for distributed combustion
In our previous investigation on CDC using gaseous fuel, a corre-
lation was proposed to predict the oxygen concentration at which CDC
occurs as a function of the entrained reactive gases from within the
combustor. This correlation was modified to be a function of the mix-
ture temperature rather than the entrained gases alone. This change
allows one to compare this correlation for previously examined gaseous
fuels as well as the experiments performed herein using liquid fuels,
wherein the inlet gases mixture temperature was kept at 700 K ± 10 K.
The modified data, as a function of inlet mixture temperature, for
methane, propane, and hydrogen enriched methane are shown in
Fig. 13. Based on these data, the correlation takes the form of:
O2 = 63.873 × T∧ (−0.278)
where O2 is the oxygen concentration in percentage and T is mixture
temperature in degrees Kelvin. The black line in Fig. 13 represents this
equation, with the dotted lines showing a ± 0.75% margin. Also shown
in the plot are the points of transition to CDC for JP-8 and ethanol at
equivalence ratio of 0.9 (for consistency with the plotted data for gas-
eous fuels). From the figure, one can see that both fuels fall within the
predicted values. In addition, the slight deviation for ethanol can be
related to the oxygen concentration in the fuel, which shifted the dis-
tributed combustion to occur at lower oxygen concentration.
This outlines the fuel flexibility of distributed combustion, as one

1.7 Fig. 6. Ratio of standard deviation to mean OH* in-

1.6 tensity for JP-8 at different oxygen concentrations and
1.5 equivalence ratios examined.
Standard Deviation/Mean

1.4
1.3
1.2
1.1
1
0.9
0.8
Phi=0.9
0.7
0.6 Phi=0.8
0.5 Phi=0.7
0.4 Phi=0.6
0.3
8 10 12 14 16 18 20 22
Oxygen Concentration [%]

518
A.E.E. Khalil, A.K. Gupta Applied Energy 211 (2018) 513–522

Fig. 7. OH* Chemiluminescence for different O2 concentration and equivalence ratios with ethanol fuel. Note the different scale for each equivalence ratio.

can predict the oxygen concentration at which CDC can occur effects of fuel atomization and vaporization on the combustion beha-
within ± 0.75% regardless of the fuel used for a given mixture tem- vior. For liquid fuel operations, injectors will be needed to adequately
perature. furnish mixing with the freshly introduced air and the entrained re-
The liquid fuels examined here were prevaporized to decouple the active hot gasses prior to ignition. If such adequate spray performance

Fig. 8. OH* Chemiluminescence for the low O2

concentration and equivalence ratios with ethanol
fuel. Note the different scale for each equivalence
ratio.

519
A.E.E. Khalil, A.K. Gupta Applied Energy 211 (2018) 513–522

70 Fig. 9. Percentage ratio of mean to maximum OH*

intensity for ethanol at different oxygen concentra-
65 Phi=0.9
Mean/Maximum Signal [%]

tions and equivalence ratios.

60 Phi=0.8
55
Phi=0.7
50
Phi=0.6
45
40
35
30
25
20
15
10
8 10 12 14 16 18 20 22
Oxygen Concentration [%]

1.6 Fig. 10. Ratio of standard deviation to mean OH*

1.5 intensity for ethanol at different oxygen concentra-
1.4 tions and equivalence ratios examined.
Standard Deviation/Mean

1.3
1.2
1.1
1
0.9
0.8
0.7
0.6 Phi=0.9
0.5
Phi=0.8
0.4
0.3 Phi=0.7
0.2 Phi=0.6
0.1
8 10 12 14 16 18 20 22
Oxygen Concentration [%]

JP-8, Phi=0.9 JP-8, Phi=0.7 Ethanol, Phi=0.9 Ethanol , Phi=0.7 Fig. 11. NO and NOx emissions for JP-8 and ethanol
at equivalence ratios of 0.9 and 0.7 at different
80 oxygen concentrations examined. NO: Closed symbols
Emissions corrected to 15% O2 [PPM]

with dashed lines, NOX: open symbols.

70 NO
NO x
60

50

40

30

20

10

0
8 10 12 14 16 18 20 22
Oxygen Concentration [%]

is established, distributed combustion shall occur at the same oxygen
concentrations shown using prevaporized liquid fuels.
5. Conclusions
Distributed combustion conditions were examined for liquid fuels,
namely JP-8 and ethanol, with focus on pollutants emission and transition
to distributed combustion condition. OH∗ chemiluminescence was used to
determine the global flame structure at different conditions that outline
the transition to distributed combustion conditions. OH∗ chemilumines-
cence imaging showed a significant reduction in signal intensity with
oxygen concentration reduction along with the disappearance of the
traditional swirl lobes associated with swirl combustion. Transition to
distributed combustion, with minimal intensity gradients, was achieved at
9.92% and 9.54% oxygen concentration for JP-8 and ethanol, respectively
at an equivalence ratio of 0.9. Below these concentration, distributed
combustion was attained until the flame blowout. Similar results were
shown at equivalence ratios between 0.6 and 0.8 albeit at slightly higher
oxygen concentration.
For pollutants emission, significant reductions in NO and NOx
emission were demonstrated upon transition to CDC, with up to 95%
reduction in emissions. Less than 2 PPM of total NOx was demonstrated
at transition to CDC for the different equivalence ratios examined for
both the liquid fuels examined. On the other hand, there was no clear
trend for CO emissions once distributed combustion was achieved. For
JP-8, CO emissions remained relatively high, averaging around

520
A.E.E. Khalil, A.K. Gupta Applied Energy 211 (2018) 513–522

JP-8, Phi=0.9 JP-8, Phi=0.7 Ethanol, Phi=0.9 Ethanol , Phi=0.7 Fig. 12. CO emission for JP-8 and ethanol at equiva-
1000 lence ratios of 0.9 and 0.7 for the different oxygen
CO emission corrected to 15% O2 [PPM]

concentration examined.

100

10

1
8 10 12 14 16 18 20 22
Oxygen Concentration [%]

Methane Propane Hydrogen Enriched Methane JP8 Ethanol Fig. 13. Oxygen concentration at transition to dis-
15 tributed combustion for mixture temperatures and
fuels examined.
Oxygen at Transition to CDC [%]

14

13

12

11

10

8
300 400 500 600 700 800 900 1000
Mixture Temperature [K]

200 PPM regardless of the oxygen concentration at an equivalence ratio References

of 0.9. Much lower CO was recorded at equivalence ratio of 0.7
(∼6 PPM). Ethanol, on the other hand, emitted significantly low CO
with values lower than 15 PPM and 5 PPM, at equivalence ratios of 0.9
and 0.7, respectively, regardless of the oxygen concentration.
The data from the examined fuels regarding oxygen concentrations
at which transition to CDC occurs was combined with previously ob-
tained data on gaseous fuels to reveal that the transition occurred
within the range previously reported for gaseous fuels. This indicates
the global features of CDC, as it does not significantly dependent on the
fuel type, provided that the fuel is pre-vaporized as revealed from the
experimental results reported herein. Further studies are required to
assess the dependency using direct fuel injection.
The results demonstrate the benefit of colorless distributed combus-
tion as a method to achieve ultra-low emissions with good fuel flexibility,
including both gaseous and prevaporized liquid fuels. Results showed
that, distributed combustion can provide emissions as low as 2 PPM of
NOx. These emissions, combined with the distributed OH∗ chemilumi-
nescence images shown under distributed combustion conditions, high-
lights the beneficial aspects of distributed combustion conditions and its
ability to handle different fuels with ultra-low emissions.
Acknowledgments
This research was supported by Office of Naval Research (ONR) –
United States and is gratefully acknowledged.
[1] U.S. Energy Information Administration. International Energy Outlook 2016. Report
Number: DOE/EIA-0484(2016). Washington DC, USA: U.S. Department of Energy.
[2] U.S. Energy Information Administration. Annual Energy Outlook 2017, AEO 2017.
Washington DC, USA: U.S. Department of Energy.
[3] Tacina RR. Low NOx Potential of Gas Turbine Engines. NASA Technical
Memorandum. In: 28th AIAA aerospace sciences meeting, Reno, Nevada, January
8–11; 1990. Paper # AIAA-90-0550.
[4] Rizk NK, Mongia HC. Low NOx Rich-Lean Combustion Concept Application. In: 27th
AIAA joint propulsion conference, Sacramento, California, June 24–26; 1991. Paper
# AIAA-91-1962.
[5] Feitelberg AS, Lacey MA. The GE rich-quench-lean gas turbine combustion. J Eng
Gas Turb Power 1998;120(3):502–8.
[6] Wunning JA, Wunning JG. Flameless oxidation to reduce thermal NO formation.
Progr Energy Combust Sci 1997;23:81–94.
[7] Lammel O, Schutz H, Schmitz G, Luckerath R, Stohr M, Noll B, Aigner M, Hase M,
Krebs W. FLOX combustion at high power density and high flame temperature. J
Eng Gas Turb Power 2010;132(12):121503.
[8] Zornek T, Monz T, Aigner M. Performance analysis of the micro gas turbine turbec
T100 with a new FLOX-combustion system for low calorific fuels. Appl Energy
2015;159:276–84.
[9] Cavaliere A, de Joannon M. MILD combustion. Progr Energy Combust Sci
2004;30(4):329–66.
[10] Weber R, Smart JP, Vd Kamp W. On the (MILD) combustion of gaseous, liquid, and
solid fuels in high temperature preheated air. Proc Combust Inst
2005;30(2):2623–39.
[11] Ozdemir IB, Peters N. Characteristics of the reaction zone in a combustor operating
at MILD combustion. Exp Fluids 2001;30:683–95.
[12] Arghode VK, Gupta AK. Effect of flowfield for colorless distributed combustion
(CDC) for gas turbine combustion. Appl Energy 2010;78:1631–40.
[13] Khalil AEE, Gupta AK. Swirling distributed combustion for clean energy conversion
in gas turbine applications. Appl Energy 2011;88:3685–93.

521
A.E.E. Khalil, A.K. Gupta
[14] Tsuji H, Gupta AK, Hasegawa T, Katsuki M, Kishimoto K, Morita M. High tem-
perature air combustion: from energy conservation to pollution reduction. Boca
Raton: CRC Press; 2003.
[15] Correa SM. A review of NOx formation under gas-turbine combustion conditions.
Combust Sci Technol 1992;87:329–62.
[16] Khalil AEE, Gupta AK. Distributed swirl combustion for gas turbine application. J
Appl Energy 2011;88:4898–907.
[17] Khalil AEE, Gupta AK, Bryden MK, Lee SC. Mixture preparation effects on dis-
tributed combustion for gas turbine applications. J Energy Resour Technol
2012;134(3):032201.
[18] Khalil AEE, Gupta AK. Impact of internal entrainment on high intensity distributed
combustion. J Appl Energy 2015;156:241–50.
[19] Khalil AEE, Gupta AK. Internal entrainment effects on high intensity distributed
combustion using non-intrusive diagnostics. J Appl Energy 2015;160:467–76.
[20] Khalil AEE, Gupta AK. Thermal field investigation under distributed combustion
conditions. J Appl Energy 2015;160:477–88.
[21] Khalil AEE, Gupta AK. Towards colorless distributed combustion regime. Fuel
2017;195:113–22.
[22] Khalil AEE, Gupta AK. Fuel property effects on distributed combustion. Fuel
2016;171:116–24.
[23] Xing F, Kumar A, Huang Y, Chan S, Ruan C, Gu S, Fan X. Flameless combustion with
liquid fuel: a review focusing on fundamentals and gas turbine applications. J Appl
Energy 2017;193:28–51.
[24] Ye J, Medwell PR, Varea E, Kruse S, Dally BB, Pitsch HG. An experimental study on
MILD combustion of prevaporized liquid fuels. J Appl Energy 2015;151:93–101.
[25] Reddy VM, Sawant D, Trivedi D, Kumar S. Studied on a liquid fuel based two stage
flameless combustor. Proc Combust Inst 2013;34:3319–26.
[26] Temme JE, Allison PM, Driscoll JF. Combustion instability of a lean premixed
prevaporized gas turbine combustor studied using phase-averaged PIV. Combust
Flame 2014;161:958–70.
[27] Seibert M, Nieh S. Measurements of hydrogen-enriched combustion of JP-8 in open
flame. J Energy Resour Technol 2017;139:012205.
522
Applied Energy 211 (2018) 513–522
[28] Far KE, Parsinejad F, Metghalchi H. Flame structure and laminar burning speeds of
JP-8/air premixed mixtures at high temperatures and pressures. Fuel
2010;89:1041–9.
[29] Far KE, Moghaddas A, Metghalchi H, Keck JC. The effect of diluent on flame
structure and laminar burning speeds of JP-8/oxidizer/diluents premixed flames.
Fuel 2011;90:1476–86.
[30] Gupta KK, Rehman A, Sarviya RM. Bio-fuels for the gas turbine: a review. Renew
Sustain Energy Rev 2010;14(9):2946–55.
[31] Schornick J, Farr R, DiCampli J. Liquid biofuels in the aeroderivative gas turbine.
In: World energy 21st congress, 11–16 September 2010, Montreal, Canada; 2010.
[32] Leplat N, Dagaut P, Togbé C, Vandooren J. Numerical and experimental study of
ethanol combustion and oxidation in laminar premixed flames and in jet-stirred
reactor. Combust Flame 2011;158(4):705–25.
[33] Cetin M, Yuksel F, Kuş H. Emission characteristics of a converted diesel engine using
ethanol as fuel. Energy Sustain Develop 2009;13(4):250–4.
[34] Saxena P, Williams FA. Numerical and experimental studies of ethanol flames. Proc
Combust Inst 2007;31:1149–56.
[35] Konnov AA, Meuwissen RJ, De Goey LPH. The temperature dependence of the la-
minar burning velocity of ethanol flames. Proc Combust Inst 2011;33:1011–19.
[36] Liao SY, Jiang DM, Huang ZH, Zeng K, Cheng Q. Determination of the laminar
burning velocities for mixtures of ethanol and air at elevated temperatures. Appl
Therm Eng 2007;27:374–80.
[37] Khalil AEE, Brooks JM, Gupta AK. Impact of confinement on flowfield of swirl flow
burners. Fuel 2016;184:1–9.
[38] Vagelopoulos CM, Egolfopoulos FN, Law CK. Further considerations on the de-
termination of laminar flame speeds with the counterflow twin flame technique.
Proc Combust Inst 1994;25:1341–7.
[39] Widegren JA, Burno TJ. Thermal decomposition kinetics of the aviation turbine fuel
jet A. Ind Eng Chem Res 2008;47:4342–8.
[40] Wu G, Katsumura Y, Matsuura C, Ishigure K, Kubo J. Comparison of liquid-phase
and gas-phase pure thermal cracking of n-hexadecane. Ind Eng Chem Res
1996;35:4747–54.