You are on page 1of 11

Combustion and Flame 211 (2020) 406–416

Contents lists available at ScienceDirect

Combustion and Flame


journal homepage: www.elsevier.com/locate/combustflame

Control of NOx and other emissions in micro gas turbine combustors


fuelled with mixtures of methane and ammonia
Ekenechukwu C. Okafor a,∗, K.D. Kunkuma A. Somarathne b,
Rattanasupapornsak Ratthanan b, Akihiro Hayakawa b, Taku Kudo b, Osamu Kurata c,
Norihiko Iki c, Taku Tsujimura a, Hirohide Furutani a, Hideaki Kobayashi b
a
Fukushima Renewable Energy Institute, National Institute of Advanced Industrial Science and Technology (AIST), 2-2-9 Machiikedai, Koriyama, Fukushima
963-0298, Japan
b
Institute of Fluid Science, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan
c
Research Institute for Energy Conservation, National Institute of Advanced Industrial Science and Technology (AIST), Namiki 1-2, Tsukuba 305-8564,
Ibaraki, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Methane–ammonia mixtures have potentials as low-carbon fuels for gas turbines, however significantly
Received 3 June 2019 high fuel NOx production in their flames present challenges to their application. This study aims to pro-
Revised 31 July 2019
vide deep insight into the physical and chemical processes involved in the formation and control of emis-
Accepted 5 October 2019
sions from the combustion of CH4 –NH3 –air with up to 30% ammonia by heat fraction in gas turbine com-
bustors. Hence, laser diagnostics techniques such as Particle Image Velocimetry (PIV), and Planar Laser
Keywords: Induced Fluorescence (PLIF) imaging, in addition to Fourier Transform Infrared (FTIR) gas analysis were
Micro gas turbine employed to study the flow field, flame structure and emissions characteristics of a micro gas turbine
Ammonia swirl combustor fuelled with CH4 –NH3 –air mixtures. The control of emissions from the flames was fur-
Fuel NOx
ther studied using Large Eddy Simulation (LES) of a model swirl combustor. The results show that NOx
Rich-lean combustion
emissions from premixed CH4 –NH3 –air in single-stage combustion were more than 50 0 0 ppmv at equiv-
Emissions control
alence ratios,  = 0.8–1.1, which is about twice more than the values already reported for NH3 –air. Trends
in NOx emissions correspond with the trends in OH radicals concentration in the combustor owing to the
relevance of OH radicals in fuel NOx production. Emissions control leading to significantly low emissions
such as 49 ppmv of NOx, 2 ppmv of CO and approximately zero N2 O, HCN and NH3 emissions with a
99.8% combustion efficiency was achieved using rich-lean combustion. An optimum  of the primary
combustion zone for low NOx emission was identified, which varied from 1.30 to 1.35 depending on the
ammonia fraction. For  richer (leaner) than the optimum Φ , NOx emission increased due to an increase
in NOx production in the secondary (primary) combustion zone. Rich-lean combustion of CH4 –NH3 –air
emitted less NOx than that of NH3 -air because the higher flame speed of CH4 –NH3 –air mixtures ensured
lower NOx production in the secondary combustion zone.
© 2019 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction tion as an alternative fuel in thermal power generation may ensure


a substantial reduction in global CO2 emissions. Furthermore, the
Ammonia (NH3 ), with a well-established global production and introduction of ammonia as a gas turbine fuel may not demand
transportation infrastructure, a high hydrogen density, and boiling huge extra costs on infrastructures for ammonia transportation and
and condensation temperatures similar to propane, has recently at- storage because ammonia is already being used in most large-scale
tracted interest as an option for hydrogen transport and storage. thermal power plants for selective catalytic reduction of NOx [2].
On the other hand, there has been a rekindled interest in ammonia Until recently however, ammonia combustion applications had
as a fuel for combustion systems, especially micro gas turbines and been discouraged mainly by the relatively low flame speeds of
industrial furnaces [1,2]. Being a carbon-free fuel, its wide applica- ammonia–air mixtures and the high fuel NOx production in the
flames. A stoichiometric NH3 –air mixture has a laminar burning
velocity that is one-fifth that of CH4 –air mixture at 298 K and

Corresponding author.
0.10 MPa [3]. Previous studies by Pratt [4] reported that the low
E-mail address: ekenechukwu.okafor@aist.go.jp (E.C. Okafor). burning velocity of NH3 –air mixtures required the use of low inlet

https://doi.org/10.1016/j.combustflame.2019.10.012
0010-2180/© 2019 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416 407

velocities in ammonia gas turbine combustors to ensure sufficient micro gas turbine demonstrated that CH4 –NH3 –air mixtures re-
fluid residence time for ammonia consumption. However, low inlet sult in higher combustion and thermal efficiencies than NH3 –air
velocities inhibited turbulent mixing and flame stabilization. Thus, mixtures [5].
unacceptably low combustion efficiencies were recorded in the gas However, CH4 –NH3 –air flames may produce significantly higher
turbine combustor. Given that ammonia is a toxic substance [1], NOx emissions than NH3 –air flames [5,26,28]. NOx production
unburned ammonia emission from gas turbines is undesirable. On from ammonia containing flames increases with an increase in the
the other hand, NO emissions from ammonia combustion could be concentration of O, H and OH radicals in the flames because these
several thousand ppmv owing to high NO production from the N O/H radicals promote the oxidation of NH2 and NH to form NO
atom in ammonia molecule [5,6]. via a HNO intermediate route [31]. The combustion of hydrocar-
Recent numerical and experimental studies [2,5–11] have how- bon species in ammonia-containing flames generates high amounts
ever shown that NH3 -air flames can be stabilized over a wide of O/H radicals leading to high NO production [32]. 1D numerical
range of conditions by employing swirling flows in combustors. simulation of adiabatic and unstretched CH4 –NH3 –air flames indi-
Three-dimensional Large Eddy Simulations (LES) by Somarathne et cated that CH4 –NH3 –air flames with more than 10% of NH3 in the
al. [8–10] using detailed chemistry indicate that stable NH3 –air fuel by volume may produce more NO than NH3 –air flames [26].
flames can be achieved using swirl combustors even for high in- Furthermore, studies on the oxidation of mixtures of ammonia
let velocities of 39.1 m/s. The flow field of combustors with strong and carbon-containing species in staged combustion indicate
swirls are characterized by regions of recirculating fluid motion that HCN and consequently, N2 O production may be important
which circulates hot products and active radicals from downstream in hydrocarbon–ammonia–air combustion [33,34,35]. Kristensen
to upstream of the combustor. Consequently, mixing and preheat- et al. [35] studied the nitrogen chemistry in the burnout zone of
ing of the fresh unburned mixture is enhanced thereby enhanc- staged combustion containing a mixture of CO, NO, HCN, NH3 and
ing the stability of the flame [12–14]. In addition, the recirculating O2 . They found that HCN oxidation leads to high N2 O production
motion increases the fluid residence time, allowing more time for and recorded up to 100 ppmv of N2 O emission at lean burn out
consumption of the fuel. [15]. conditions. Wargadalam et al. [32] investigated the homogenous
Kurata et al. [5] employed a high-swirl non-premixed combus- formation of NO and N2 O from the combustion of HCN and NH3 at
tor (swirl number = 0.88) in which the fuel was injected parallel 873–1273 K and noted that HCN oxidation promotes NO and N2 O
to the combustor central axis, and reported successful generation formation. Takagi et al. [33] suggests that HCN contributes to
of power from a 50 kWe-class regenerative cycle micro gas turbine the higher NOx production from hydrocarbon-ammonia flames in
fuelled with NH3 –air mixtures. They reported combustion efficien- comparison to ammonia-containing flames with no hydrocarbon
cies of up to 96% for Combustor Inlet Temperatures (CIT) above content. This is because HCN has a higher tendency than NH3 to
700 K. Lower CIT however was characterized by high level of un- be oxidized to NO, being consumed at leaner combustion zones
burned NH3 emissions. A more recent study by Okafor et al. [6] us- with higher OH concentrations. Note that HCN is a very toxic gas
ing a similar combustor in a laboratory combustor test rig showed and 1 min exposure to ca. 300 ppmv of HCN is lethal [36]. On
that the combustion efficiency of non-premixed NH3 –air flames the other hand, N2 O is a greenhouse gas whose Global Warming
could be improved by the use of inclined fuel injection as opposed Potential (GWP) is 298 times larger than that of CO2 [5]. Therefore,
to fuel injection parallel to the combustor central axis. This has it is desirable to achieve a complete consumption of these species
been demonstrated to improve the combustion efficiency of non- in the combustor.
premixed NH3 –air mixtures in a micro gas turbine even at low CIT Most of the previous experimental studies on the control of
[11]. emissions from ammonia-blended hydrocarbon combustion con-
Furthermore, two-stage rich-lean combustion has been demon- sidered only small quantities of ammonia in the order of few
strated to control NOx emissions from combustors fuelled with thousand ppmv in the fuel. Because ammonia was seldom consid-
NH3 –air mixtures [2,6,9,10]. Rich-lean combustion takes advantage ered a fuel until recently, the characteristics and control of emis-
of the low NOx production from the oxidation of ammonia at fuel- sions from combustors fuelled with hydrocarbon-ammonia mix-
rich conditions. The primary combustion zone is maintained at a tures with high ammonia concentration in the fuel has seldom
fuel-rich condition to ensure low NOx production while air is in- been investigated. Jójka and Ślefarski [37] studied the combustion
jected in the secondary combustion zone to burn off the unburned of CH4 –NH3 –air mixtures with up to 5% volume fraction of am-
NH3 and H2 from the fuel-rich primary combustion zone. Okafor monia in the fuel in a nozzle burner and found that NOx emis-
et al. [6] showed that NOx emission from two-stage combustors sion, which was in the order of 20 0 0–30 0 0 ppmv increased with
fuelled with NH3 –air mixtures varied non-monotonically with the the ammonia content in the fuel. Valera-Medina et al. [38] stud-
equivalence ratio of the primary combustion zone, Φ pri , and had ied the stability and emissions characteristics of CH4 –NH3 –air mix-
a minimum value at Φ pri = 1.10. At this optimum low NOx equiv- tures with 69% ammonia volume fraction in the fuel using a single-
alence ratio and an ambient pressure of 0.30 MPa, they reported stage tangential swirl combustor and showed that NOx emissions
NOx emission from premixed NH3 –air combustion of 42 ppmv at at fuel-lean conditions ( = 0.8) was more than 30 0 0 ppmv. They
16% O2 concentration with a combustion efficiency of 99.5%. concluded that a stratified injection of methane and ammonia
On the other hand, there is growing interest in the combus- as opposed to fully premixed injection is required to encourage
tion applications of blends of ammonia and other fuels such as the production of active radicals for NOx reduction. Ramos et al.
hydrogen [16–21], diesel [22,23], methane [5,24–30] and pulver- [29] measured emissions from premixed CH4 –NH3 –air flames with
ized coal [2]. The present study focuses on mixtures of methane up to 70% NH3 by mole fraction stabilized on a laminar flame
and ammonia. An important advantage of ammonia blending in burner. They reported that NOx emission increased as NH3 mole
hydrocarbon combustion is the reduction in CO2 emission due fraction increased up to 50% and then decreased afterwards. Li et
to the replacement of some part of the hydrocarbon by ammo- al. [28] numerically analysed the combustion of methane-ammonia
nia. On the other hand, such blending offers potentials for an mixtures in a model combustor consisting of perfectly stirred reac-
improvement of the combustion characteristics of ammonia mix- tors and plug flow reactors at 2.3 MPa. The 1D calculations indicate
tures. It has already been reported that mixtures of CH4 –NH3 – the potential of two-stage combustion in controlling NOx emis-
air with up to 30% of ammonia by heat fraction (52% by volume sions from the combustion of CH4 –NH3 –air mixtures with 40% am-
fraction) in the fuel have burning velocities twice more than that monia volume fraction when the equivalence ratio of the primary
of NH3 –air mixtures at the same conditions [25,26]. Studies in a combustion zone is 1.5. However, there is yet no experimental
408 E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416

Fig. 1. The swirl combustor and liners. The model liner shown in (c) has similar dimensions as the original gas turbine liner shown in (a).

evidence of control of NOx and other emissions from CH4 –NH3 – the swirler/injector flow ratio in the corresponding non-premixed
air flames with high ammonia contents. case.
In this study, experiments and 3D computations which comple- Two combustor liners with similar dimensions were employed,
ment each other were employed to study the control of NOx, in- namely; the real gas turbine combustor liner shown in Fig. 1a, and
cluding NO2 and N2 O, and other emissions such as HCN and CO a model liner shown in Fig. 1c. The real gas turbine liner had six
from combustors fuelled with mixtures of methane and ammo- holes in the secondary combustion zone with diameter of 23 mm,
nia with up to 30% of ammonia by heat fraction (52% by volume). which allowed for secondary air injection. On the other hand, the
In the experiments, laser diagnostics techniques such as PIV, OH- model liner had no holes for secondary air injection, instead the
PLIF and simultaneous OH/NO-PLIF imaging were first employed to cylindrical sections were made of quartz glass to allow for optical
characterize the flow field and the flame structure in the com- measurements. The regions enclosed by the dash lines in Fig. 1c
bustor. Then the emission characteristics of CH4 –NH3 –air flames indicate the measurement areas for PIV and PLIF. The methodology
in single stage and two-stage combustion were investigated, em- and results of the PIV measurement are provided as supplementary
ploying Fourier Transform Infrared (FTIR) gas analysis. The effects materials to this article.
of ammonia fraction, equivalence ratio, mixture homogeneity, and The combustor was installed inside a laboratory combustor test
ambient pressure on emissions were examined. On the other hand, rig described in our previous studies [6] and in the present supple-
the computational approach involved the application of Large Eddy mentary material. Gaseous ammonia (> 99.9% purity) and methane
Simulation (LES), which provides a compromise between cost and (99.0% purity) were supplied from their respective pressurized
accuracy, to study the process of emissions formation and control cylinders, and mixed at a point 5 m upstream of the combustor in-
in a two-stage combustor in order to advance the understanding of let ports. The concentration of ammonia in the fuel was expressed
the experimental observations. in terms of the heat fraction, ENH3 , and was varied from 0 to 0.30.
The objective of this study is thus to provide deep insight into
the physical and chemical processes involved in emissions forma- xNH 3 LHVNH 3
tion and control in the combustion of mixtures of methane and
ENH3 = (1)
xNH 3 LHVNH 3 + xCH 4 LHVCH 4
ammonia with high ammonia concentration, as well as demon-
strate the potential of achievement of low levels of NOx and other Here, xNH3 and xCH4 are mole fractions of NH3 and CH4 in the bi-
emissions in the application of such mixtures as gas turbine fuels. nary fuel and LHV stands for the lower heating value (LHVCH 4 =
802.30 kJ/mol, LHVNH 3 = 316.84 kJ/mol).
2. Experimental and numerical methods The air flowrate was measured with Brooks mass flow con-
trollers with an accuracy of ±0.9% of the set point. On the other
2.1. The combustor hand, the flowrates of ammonia and methane were measured with
Kofloc mass flow meters with accuracies ranging from ± 1% to
The combustor, shown schematically in Fig. 1a, was that of a ± 1.5% of full scale (F.S.) depending on the value of F.S. The esti-
50 kWe type micro gas turbine employed by Kurata et al. [5,11]. mated absolute uncertainties in measuring the equivalence ratio
The combustor had a swirler, shown in Fig. 1b, with a swirl num- and fuel fraction increased with ENH3 and the maximum values
ber of 0.88 surrounding a 12-hole fuel injector. The injection an- were ± 0.04 and ± 0.015, respectively. The primary combustion air
gle of each hole was 45° to the combustor central axis. In this was separately measured and supplied directly to the swirler while
study, premixed and non-premixed combustion were investigated. the secondary combustion air, whose flowrate was also separately
In non-premixed combustion, the fuel was introduced through the controlled, was supplied to the base of the combustor test rig and
injector while the swirler was used for air. To ensure comparable made to flow through the holes on the combustor liner. The com-
fluid residence times in both non-premixed and premixed com- bustor inlet temperature (CIT) was kept at 298 K at all measure-
bustion cases, about equal flow rates were allowed through the ments in the present study. The chamber pressure was manually
injector and swirler in both cases at any given condition. In the controlled using valves mounted on the exhaust pipe. Three K-type
premixed case, the mixture flow rate was divided between the thermocouples were placed at the liner exit to measure the com-
swirler and injector in a proportion approximately equivalent to bustor outlet temperature, (COT).
E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416 409

multiple scales were employed in analyzing species such as NO,


NH3 , and CO, whose mole fractions may vary from several thou-
sand ppmv to few ppmv. On the other hand, single scales were
employed for species such as NO2 , N2 O, and HCN. The maximum
uncertainty was less than ±8 ppmv for most measurements. How-
ever, for NO and NH3 mole fractions more than 500 ppmv the un-
certainties were about ±15 ppmv and ±30 ppmv, respectively. For
all experiments on emission measurement, the swirler mean inlet
velocity was kept at ∼2.70 m/s in order to ensure an approximately
constant fluid residence time in the combustor at all conditions.

2.4. Large eddy simulation (LES)

Two-stage combustion of CH4 –NH3 –air mixtures in a gas


turbine-like combustor was modeled using OpenFOAM [39]. De-
tails of the computational domain and the numerical methods are
as described in the previous modeling studies on NH3 –air com-
bustion by Somarathne et al. [10] and also provided in the present
Supplementary materials. Few relevant points are mentioned here.
The unstructured 3D computational domain was a cylindrical non-
premixed combustor with two annulus swirlers and eight square-
shaped secondary combustion air injection ducts.
Fig. 2. Setup for simultaneous OH/NO-PLIF imaging. The LES model employed spatially filtered mass, momentum,
energy and species equations [40], in which scales smaller than
the grid size were not resolved but were accounted for through
2.2. Planar laser induced fluorescence (PLIF) imaging
the sub-grid scale tensor modeled based on an eddy-viscosity as-
sumption. The wall adaptive local eddy-viscosity (WALE) model
The OH-PLIF and simultaneous OH/NO-PLIF imaging were con-
[41] was employed for modeling the sub-grid scale viscosity. Un-
ducted in the flames at 0.10 MPa. The OH-PLIF system consisted
resolved fluxes of species and sensible enthalpy in the spatially fil-
of a frequency doubled dye laser (Lumonics, 13 mJ/pulse at
tered mass fraction and sensible enthalpy equations were adopted
282.929 nm after SHG) pumped by an Nd:YAG laser (Spectral
as gradient assumptions. Furthermore, turbulence was modeled
physics, 300 mJ/pulse at 532 nm), a digital delay generator (Stan-
by solving a model transport equation for the turbulent kinetic
ford Research Systems Inc.), photomultiplier and an ICCD camera
energy [42]. The filtered reaction rate terms were modeled us-
(Andor technology). The dye laser was tuned to 282.929 nm to ex-
ing partially stirred reactor (PaSR) model [43], which is a multi-
cite the (1, 0) band of the OH (X2  ← A2  + ) system. The OH flu-
scale model accounting for the turbulent–chemistry interaction.
orescence was detected at around 310 nm using the ICCD camera
The LES-PaSR model has been extensively used and validated for
mounted with a combination of a Schott UG11 optical filter cen-
modeling turbulent–chemistry interactions [44]. The walls of the
tered at 325 nm (FWHM = 110 nm) and LX0313 filter centered at
combustor were non-slip and adiabatic, and the outlet was speci-
313 nm (FWHM =10 nm). All OH-PLIF images in the present study
fied by the continuity boundary condition. An in-situ adaptive tab-
were taken with the same laser excitation power and image acqui-
ulation algorithm library [45] was combined with the main com-
sition settings of the ICCD camera.
putational program in order to speed up the calculations. A con-
The setup for simultaneous OH/NO-PLIF imaging is as shown
stant total air flow rate of 0.022 m3 /s was maintained for all
schematically in Fig. 2. For NO excitation, an Nd:YAG laser was
computations to ensure an approximately constant fluid residence
used to pump a wavelength tunable dye laser with a frequency
time, while the ratio of the primary to the total air flow rates was
doubler (Sirah Precision Scan). The laser wavelength was tuned to
constant at 0.306. The computations were performed at an ambi-
226.298 nm in order to excite the Q1(29.5) transition for NO with
ent pressure of 0.50 MPa, air inlet temperature of 500 K and fuel
a power of 5 mJ/pulse. The OH-PLIF and NO-PLIF were separately
inlet temperature of 300 K.
detected with two intensified CCD (ICCD) cameras (Andor technol-
In the previous numerical work by Somarathne et al. [10] on
ogy) mounted with the appropriate optical filters. A band pass fil-
NH3 –air combustion, the chemical reaction mechanism by Miller
ter centered at 236 nm was employed in the case of NO-PLIF.
et al. [46] was employed. The present LES however employed a
reduced reaction mechanism by Okafor et al. [26] with 42 species
2.3. Exhaust gas analysis and 130 reactions. The mechanism was developed and optimized
with measured values of the unstretched laminar burning velocity
The exhaust gases from the exit of the combustor were ana- of CH4 –NH3 –air, and NH3 –air flames and species concentrations in
lyzed using a Fourier Transform Infrared (FTIR) gas analyzer (BOB- the flames over a wide range of conditions including equivalence
20 0 0FT) containing O2 and H2 analyzers with a scanning frequency ratios = 0.7–1.3, ENH3 = 0–1.0, and mixture pressure = 0.10 MPa–
of 5 Hz. The temperature of the sampling line was maintained at 0.50 MPa. On the other hand, a detailed mechanism by Okafor et al.
464 K to avoid condensation of water vapor along the line. Before [25] validated for modeling the burning velocity and speciation
each series of measurements, the gas analyzer was calibrated using in CH4 –NH3 –air flames [25,26,47] was used in 1D analysis of the
span gases, while for each measurement, the species concentration chemistry of pollutants formation in the unstretched and adiabatic
in the exhaust gases were allowed to attain approximately steady flames using ANSYS Chemkin-PRO software [48]
values before data was logged for about 60 sec. The emission val- We recognize that the equivalence ratio of the non-premixed
ues reported here are the mean. The standard deviation from the flames varies locally in the combustor. Therefore, the term “global
mean at all conditions and the repeatability of the measurements equivalence ratio, Φ ” is henceforth used to refer to the average
are less than 5% of the mean values. Because the accuracy of the equivalence ratio. To avoid confusion that may arise due to the use
measurements was dependent on the range of the measuring scale, of numerous symbols for equivalence ratios, Φ is used for both
410 E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416

Fig. 3. (a) Single shot OH-PLIF images and (b) time-average OH-PLIF images of 200 Fig. 4. (a) Single shot OH-PLIF images and (b) time-average OH-PLIF images of 200
single shots in premixed CH4 –NH3 –air flames with ENH3 = 0.30. single shots in non- premixed CH4 –NH3 –air flames with ENH3 = 0.30.

non-premixed and premixed flames in this study. Where staged Figure 4 shows the single shot and the time-averaged OH-PLIF
combustion (such as rich-lean) strategy is employed, Φ overall is images for the non-premixed flames. Two important differences
used to refer to the equivalence ratio obtained based on the to- can be observed between the OH-PLIF images of the premixed
tal air and fuel supplied to the combustor while Φ pri is used to flames and those of the non-premixed flames. As the equivalence
refer to the equivalence ratio of the primary combustion zone. ratio increased, the reaction zone tended to move closer to the
combustor walls in the non-premixed flames. The regions close to
the combustor wall may be leaner than regions around the center
3. Results and discussion of the combustor because the air swirler surrounded the fuel in-
jector in this study. In addition, unlike in the premixed case, the
3.1. Flame structure OH-PLIF intensity did not decrease appreciably with an increase
in equivalence ratio. The OH-PLIF intensity was high enough even
Figure 3a shows single shot OH-PLIF images of premixed at Φ = 1.4 for the flame structure to be clearly observed in non-
CH4 –NH3 –air flames of ENH3 = 0.3 at different equivalence ratios. premixed combustion.
Note that the laser sheet passed from right to left hence the de- The plane-integrated time-averaged OH-PLIF intensities of the
crease in OH-PLIF intensity from right to left on the images is premixed and the non-premixed flames at various equivalence ra-
due to laser absorption. A time-averaged image of 200 single shots tios are shown in Fig. 5. The premixed flame had a significantly
shown in Fig. 3b highlights the mean flame shape and may also higher OH-PLIF intensity close to stoichiometric equivalence ra-
represent a probability density map of OH-PLIF intensity in the tios and clearly showed a peak at  = 0.9. On the other hand,
combustor. Across the reaction zone, the concentration of OH rad- the non-premixed flame had higher OH-PLIF intensities in fuel-lean
icals rises sharply from zero (in the unburned mixture) to super (Φ ≤ 0.7) and fuel-rich (Φ ≥ 1.2) conditions. It is considered that
equilibrium concentrations, represented by the brightest regions in in the fuel-lean and the fuel-rich non-premixed flames there may
Fig. 3a. Note that OH-PLIF intensity is proportional to the concen- exit local regions with equivalence ratios close to unity where the
tration of OH radicals in a flame [51,52]. Therefore, the approxi- OH radical concentrations are relatively high. Consequently, the
mate location of the reaction zone can be identified as the position OH-PLIF intensity for the non-premixed flame at Φ = 1.4 was com-
of relatively steep gradients of the OH-PLIF intensity [49,50,53]. parable to that at Φ = 1.0.
The intensity of the OH-PLIF was found to be highest around The qualitative trends in OH-PLIF intensities shown in
equivalence ratio of 0.9, which corresponds to the equivalence ra- Fig. 5 may represent the trends in the concentration of OH radi-
tio where the OH concentration in 1D unstretched and adiabatic cals in the combustor since all the measurements were completed
CH4 –NH3 –air flames peaks. As the flame became richer, the inten- using the same laser excitation power and image acquisition set-
sity of the OH-PLIF decreased. The OH-PLIF intensity at Φ = 1.2 can tings. It is worth noting here that these trends in Fig. 5 corre-
be seen to be relatively low. At Φ = 1.4, the OH-PLIF intensity was spond with the trends in NO emission from the flames discussed in
too low to allow identification of the flame structure (not shown Section 3.3 (See Fig. 7) because NO production from CH4 –NH3 –air
in this manuscript). flames is promoted by the production of OH radicals. Simultane-
E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416 411

Fig. 5. Variation of the plane-integrated time-averaged OH-PLIF intensities with


equivalence ratio for CH4 –NH3 –air flames with ENH3 = 0.30. The error bars repre-
sent the standard deviation from the mean of plane-integrated intensities of 400
single shots.

Fig. 7. Measured emissions of (a) NO, (b) NO2 and (c) N2 O from single-stage com-
bustion of CH4 –air and CH4 –NH3 –air mixtures.

3.2. Measured emissions from single stage combustion

An understanding of the emission characteristics of single stage


combustion is relevant to the control of emissions using two stage
combustion because the concentration of species from the primary
combustion zone of two-stage combustors is related to the emis-
sions from single-stage combustion. In addition, the optimum Φ pri
for low NOx emission in two-stage combustion may correspond to
the equivalence ratio in single stage combustion where the sum of
NO and unburned NH3 emissions have a minimum value, as dis-
cussed in Section 3.4.
Figure 7 shows that CH4 –NH3 –air premixed flames emitted
several times more NO than CH4 –air flames. For Φ = 0.8–1.0,
the emissions were larger than 50 0 0 ppmv which is the maxi-
mum measurable range of NO in this study. It can be seen that
NO emissions decreased as the equivalence ratio became leaner
(richer) than 0.7 (1.1). However even at Φ 0.6, NO emission from
CH4 –NH3 –air combustion was more than 20 0 0 ppmv. The leaner
conditions also had high N2 O emission of up to 120 ppmv for
ENH3 = 0.3 at Φ = 0.6. Note that these measured values of NOx
emissions from CH4 –NH3 –air flames are about twice more than
the already reported values of NOx emissions from NH3 -air flames
at similar conditions by Okafor et al. [6].
NO formation/reduction in ammonia containing flames is
mainly through the fuel NOx route, which is strongly dependent on
the concentration of O, H and OH radicals in the flame. N2 O is pro-
duced mainly through the reactions of NO with NH and with HO2 .
Fig. 6. Pairs of simultaneously captured instantaneous OH-PLIF and NO-PLIF images It is then rapidly consumed mainly through reaction with H rad-
in fuel-rich ( = 1.4) non-premixed CH4 –NH3 –air flames with ENH3 = 0.20. The top
icals and through thermal dissociation. Hence, low temperatures
and bottom pairs are separated only by time.
and low concentration of H radicals, which are typical conditions
in far-lean equivalence ratios, may hinder N2 O consumption and
consequently lead to significant emissions of N2 O as observed in
ous imaging of OH-PLIF and NO-PLIF performed in the fuel-rich the experiment.
(Φ = 1.4) non-premixed flame indicates, as shown in Fig. 6, that Because of the dependence of NO production on OH concentra-
the local regions of relatively high OH-PLIF intensity in the fuel- tion, NO emissions in Fig. 7 has a similar trend with the OH-PLIF
rich flame correspond with regions of relatively high NO-PLIF in- intensities in Fig. 5. It can be seen that the high OH-PLIF intensity
tensity. This observation, which is important in NOx control using in the fuel-lean premixed flames conditions corresponds with the
two-stage combustion, suggests that the fuel-rich non-premixed high NO emissions at this condition. Furthermore, the difference
flames with Φ > 1.2 may produce more NO in the primary com- between the measured NO emissions from the premixed flames
bustion zone than the fuel-rich premixed flames as discussed in and the non-premixed flames corresponds with the disparity in the
Section 3.4 (see also Fig. 12). OH-PLIF intensities of the two modes of combustion. The produc-
412 E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416

of NHi (i = 1,2,3) with H radicals [2]. On the other hand, H2 pro-


duction from the methane oxidation chemistry at fuel-rich condi-
tion is predominantly through the reaction of CH4 and CH2 O with
H radicals. As the ammonia content in the CH4 –NH3 –air flames in-
creases, a decrease in H2 production from the methane chemistry
tends to be compensated by an increase in production from the
ammonia chemistry. Contrary to the trend in Fig. 8, the equilibrium
mole fraction of H2 in rich CH4 –NH3 –air combustion may increase
with an increase in ENH3 owing to increased H2 production from
the NH3 chemistry. However, an increase in residence time is re-
quired for an increase in the production of H2 when ENH3 increases
because the reactivity of the mixture decreases. Therefore, the con-
stant finite residence time in the present combustor may limit the
production of H2 as ENH3 increases leading to the observed trend
in H2 emission.
The COT and the emissions of CO2 , H2 O, CO and CH4 are pro-
vided as supplementary materials to this article.

3.3. LES of two-stage combustion


Fig. 8. Measured emissions for (a) NH3 , (b) HCN and (c) H2 from single-stage com-
bustion of CH4 –air and CH4 –NH3 –air mixtures. The primary target of NOx control in combustors fuelled with
ammonia mixtures is to control the oxidation of NH2 and NH by
the O/H radicals. To this effect, a control of the equivalence ra-
tion of OH and NO in the non-premixed flame may be lower than tio is critical because it ensures a control of the distribution of
that in the premixed flame due to less efficient combustion owing the radicals. Unlike in thermal NOx control, temperature tends to
to insufficient mixing. This conclusion is supported by the signifi- play a secondary role in promoting fuel NOx formation through
cantly higher emissions of unburned fuels from the non-premixed its influence on the production rate of the O/H radicals. Figure 9
flame (see Fig. 8). shows the profiles of temperature and NO in the modeled two-
Observe however that even though Figs. 5 and 6 suggest that stage combustor for four different conditions. The temperature in
the fuel-rich non-premixed flames of Φ > 1.2 may produce higher the primary combustion zone is lowest for the fuel-lean case of
NO due to the high OH concentration from the local lean regions Φ pri = 0.8, however the NO emission is highest for that condition.
in the combustor, Fig. 7 shows that NO emissions from the non- Observe also that the cases with stoichiometric and fuel-rich pri-
premixed flames was not higher than that of premixed flames at mary combustion zones show relatively higher NO production near
Φ > 1.2. This is because the NO produced in single-stage com- the combustor wall because the near-wall regions are leaner than
bustion at fuel-rich conditions has sufficient residence time to be the central regions of the combustor.
reduced by the large concentration of unburned NH3 in the com- A comparison of the space time average NO concentration at
bustor. However, in two-stage rich-lean combustion where the un- 75 mm from the base of the combustor and at the combustor
burned NH3 is consumed completely in the secondary combustion exit indicates that NO concentration decreased after the injection
zone, NO produced in the primary combustion zone may not be of secondary combustion air. This apparent decrease is primar-
sufficiently reduced due to the shorter residence time for its reduc- ily due to the dilution effect of the injected air rather than a
tion. Consequently, this NO contributes to the higher NO emission chemical reduction of NO. It is reasonable to expect a selective
from the non-premixed flames in two stage rich-lean combustion non-catalytic reduction (SNCR) of NO in the secondary combustion
when the equivalence ratio of the primary zone is larger than 1.20 zone, as suggested by Kurata et al. [5], given that the temperature
as discussed in Section 3.4 (see Fig. 12.) is largely within the thermal DeNOx temperature window (1100–
Figure 8 shows the emissions of NH3 , HCN and H2 from the 1400 K) [54]. However, Fig. 10 shows that for Φ pri < 1.2, HCN and
flames. The emissions increased rapidly with equivalence ratio in unburned NH3 concentration from the primary combustion zone
the fuel-rich conditions, with values exceeding the measurable were too low to support any significant NO reduction. On the other
range for the species in this study at some conditions (as indi- hand, the relatively high concentration of HCN and unburned NH3
cated by the arrows). HCN emissions of more than 500 ppmv was from the primary combustion zone of the richer flames favors NO
recorded in the non-premixed CH4 –NH3 –air flame at Φ ≥ 1.25. formation more than its reduction in the secondary combustion
HCN is formed in CH4 –NH3 –air flames through reaction of NO or N zone due to the high concentration of O2 , CO and H2 as discussed
with CH3 , CH2 and CH. The production of HCN from the reaction of in the next paragraph.
NO with hydrocarbon radicals is dominant in fuel-lean conditions Figure 11 shows that the high concentration of H2 , CO, OH,
while its production from the reaction of N atom with the hy- and HCN from the fuel-rich primary combustion zone are effec-
drocarbon radicals is dominant in fuel-rich conditions. The formed tively consumed in the secondary combustion zone with the injec-
HCN is then oxidized mainly through reactions with O and OH rad- tion of fresh air. An inspection of the species profiles shows that
icals in the postreaction zone. The abundance O and OH radicals in H2 and CO consumption leads to the production of OH radicals.
fuel-lean conditions facilitate rapid consumption of HCN to even- H2 and CO are consumed mainly via chain-propagating reactions
tually produce NHi radicals and, then, NO. However, in fuel-rich H2 +OH=H + H2 O and CO+OH=CO2 +H, respectively, which pro-
conditions HCN oxidation tends to be hindered by the relatively mote O and OH radicals production through the chain-branching
low concentration of O and OH radicals, consequently leading to reaction H + O2 =OH+O. This contributes to an increase in O/H rad-
the high HCN emissions as observed in the experiments. ical concentration in the secondary combustion zone thereby en-
Figure 8 shows that H2 emissions was insensitive to the fuel abling the oxidation of HCN and the unburned fuels. Owing to the
composition at a constant Φ . H2 production from the NH3 oxida- high concentration of O/H radicals, HCN and NH3 are converted
tion at fuel-rich conditions is predominantly through the reaction predominantly to NO in the secondary combustion zone.
E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416 413

Fig. 9. Computed 2D profiles of temperature and mole fraction of NO in the two-stage non-premixed combustion of CH4 –NH3 –air mixtures with ENH3 = 0.20. The space-
time-average NO mole fractions at the exit and at 75 mm downstream of the combustor are indicated. (xNO represents the mole fraction of NO).

combustion. The average temperature at the combustor exit de-


creased with the injection of fresh air into the secondary zone. The
COT was however around 10 0 0 K for the conditions with Φ pri ≥
1.3. Figure 12b shows that the CO2 emissions at 16% O2 concen-
tration decreased with an increase in ENH3 and did not vary with
Φ pri . On the other hand, NO emissions varied non-monotonically
with Φ pri as shown in Fig. 12c. For the premixed flames, the
optimum low-NOx Φ pri , i.e. Φ opt , was found to be around 1.30
for ENH3 = 0.20 and 0.30, and 1.35 for ENH3 = 0.10. Φ opt tends to
decrease with an increase in ENH3 and has been reported to be
around 1.10 for NH3 -air flames [6]. For Φ pri > Φ opt , NO emission
increased with Φ pri due to an increase in NO production in the
secondary zone as explained in Section 3.3.
NO emissions from the non-premixed combustion was less
sensitive to Φ pri and had Φ opt = 1.2. Non-premixed combustion
resulted in lower NO emissions than the premixed combustion
for Φ pri ≤ 1.25. However, higher NO emissions were recorded
Fig. 10. Computed space-time-average mole fractions of NH3 , HCN, H2 , and CO at
75 mm from the base of the combustor which gives an estimate of the mole fraction
from non-premixed combustion for Φ pri > 1.25. Notice also
of the species from the primary combustion zone. that at the respective Φ opt for the two modes of combustions,
non-premixed combustion emitted more than three times more
It can be deduced from the above discussion that even though NO than premixed combustion. These higher NO emissions from
that with an increase in Φ pri , NO production in the fuel-rich pri- non-premixed combustion is partly due to higher NO formation
mary combustion zone decreases, the corresponding increase in from the higher concentration of HCN and unburned NH3 in the
H2 , CO, HCN and unburned NH3 concentrations from the primary secondary combustion zone, as suggested by Fig. 8. In addition,
combustion zone leads to an increase in NO production in the sec- the non-premixed combustion may have higher NO production
ondary combustion zone. in the primary combustion zone than the premixed combustion
when Φ pri > 1.25 due to the presence of local relatively lean
3.4. Measured emissions from two-stage combustion regions with high NO production.
On the other hand, Fig. 13 shows that NO emission from pre-
Figure 12 shows the dependence of the COT, and emissions mixed CH4 –NH3 –air combustion was insensitive to Φ overall for a
of CO2 and NO on Φ pri for a constant Φ pri = 0.38 in two-stage range of Φ overall from about 0.3 to 0.9 at a constant Φ pri = Φ opt .
414 E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416

Fig. 11. Computed 2D profiles of the mole fractions of H2 , CO, OH and HCN in rich-lean non-premixed combustion of CH4 –NH3 –air mixtures with ENH3 = 0.20 for Φ pri = 1.4.
The space-time-average concentration of H2 , CO and HCN at the exit of the combustor are indicated.

Fig. 12. Measured values of (a) combustor outlet temperature, COT (b) CO2 emis- Fig. 13. Measured variation of emissions from two-stage premixed combustion of
sions, and (c) NO emissions from two-stage combustion of CH4 –NH3 –air mixtures CH4 –NH3 –air and NH3 –air mixtures with Φ overall at Φ opt . The data for NH3 –air mix-
at a constant overall equivalence ratio of Φ overall = 0.38. tures were adopted from [6].

The figure also shows that unburned NH3 and HCN emissions were
approximately zero, while N2 O emission was less than 2 ppmv.
CO emission was less than 100 ppmv for Φ overall = 0.3 to 0.6 but
was about 20 0 0 ppmv at Φ overall = 0.9 for the flame of ENH3 = 0.10.
The figure also shows the corresponding emissions from NH3 –air
flames [6] in the combustor recorded at similar conditions as the
present study. It can be seen that at Φ opt for CH4 –NH3 –air and
NH3 –air combustion, NO emission from CH4 –NH3 –air combustion
was lower than that from NH3 –air flames in rich-lean combus-
tion. Since in rich-lean combustion, efficient consumption of the
fuel in the primary combustion zone ensures low NO production
in the secondary combustion zone, fuels with high flame speeds
may produce lower fuel NO emissions than slower burning fuels
when Φ pri = Φ opt .
Figure 14 shows the dependence of the COT, the combustion
efficiency, NOx and CO emissions on ambient pressure. The com-
bustion efficiency was obtained based on the heating values of the
fuels, considering CH4 , NH3, CO and H2 as the unburned fuels in
the exhaust gases. The figure shows that the COT and the com- Fig. 14. Dependence of the combustor outlet temperature (COT), the combustion
bustion efficiency increased with ambient pressure. The maximum efficiency(ƞcombust ), NOx and CO emissions on ambient pressure for CH4 –NH3 –air
combustion efficiency recorded in the present study was 99.8% at mixtures (ENH3 = 0.20) at constant Φ overall = 0.56 and Φ pri = 1.30.

0.25 MPa. On the other hand, CO and NOx emissions decreased


E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416 415

with an increase in ambient pressure with values of 2 ppmv and Acknowledgment


49 ppmv at 0.25 MPa, respectively. N2 O and HCN emissions at
0.25 MPa were 0.5 ppmv and 0.1 ppmv, respectively, at 16% O2 Part of this research was supported by the Council for Science,
concentration. It has been reported that fuel NO production may Technology and Innovation (CSTI), the Cross-ministerial Strategic
decrease with pressure as a result of a decrease in O/H radical Innovation Promotion Program (SIP), “Energy Carriers” (Funding
pool, and an increase in the combination reaction of NHi radi- Agency: The Japan Science and Technology Agency (JST)). We also
cals [2,6,10,26]. Furthermore, the increase in combustion efficiency acknowledge the support from the collaborative research project,
with pressure suggests an increase in the rate of consumption of Institute of Fluid Science, Tohoku University.
H2 , CO, HCN and unburned NH3 in the primary combustion re-
sulting to lower concentrations of these species in the secondary Supplementary materials
combustion zone. Consequently, NO production in the secondary
combustion zone and the overall NO emission may decrease with Supplementary material associated with this article can be
an increase pressure [6]. found, in the online version, at doi:10.1016/j.combustflame.2019.10.
The lowest NOx emission level recorded in this study is well 012.
below the 70 ppmv regulatory limit imposed by the Japanese gov-
ernment for gas turbines, however it is still well above the more References
stringent limits of about 10 ppmv enforced by various prefectural
governments in Japan [2]. Therefore, it may still be necessary to [1] A. Valera-Medina, H. Xiao, M. Owen-Jones, W.I.F. David, P.J. Bowen, Ammonia
employ exhaust gas aftertreatments for NOx reduction. Low NOx for power, Prog. Energy Combust. Sci. 69 (2018) 63–102.
[2] H. Kobayashi, A. Hayakawa, K.D.K.A Somarathne, E.C. Okafor, Science and tech-
emission from the combustor such as demonstrated in this work nology of ammonia combustion, Proc. Combust. Inst. 37 (2019) 109–133.
would encourage a significant reduction in the cost and size of SCR [3] A. Hayakawa, T. Goto, R. Mimoto, Y. Arakawa, T. Kudo, H. Kobayashi, Laminar
systems needed to achieve below-regulatory-limits NOx values. burning velocity and Markstein length of ammonia/air premixed flames at var-
ious pressures, Fuel 159 (2015) 98–106.
[4] D.T. Pratt, Performance of Ammonia-Fired Gas- Turbine Combustors, Technical
Report No.9, DA- 04-200-AMC-791(x), Berkley University of California, 1967,
available at http://www.dtic.mil/dtic/tr/fulltext/u2/657585.pdf.
4. Conclusions [5] O. Kurata, N. Iki, T. Matsunuma, T. Inoue, T. Tsujimura, H. Furutani,
H. Kobayashi, A. Hayakawa, Performances and emission characteristics of
This work demonstrates the application of rich-lean combustion NH3 -air and NH3 -CH4 -air combustion gas-turbine power generations, Proc.
Combust. Inst. 36 (2017) 3351–3359.
strategy in the control of NOx and other emissions from CH4 –NH3 – [6] E.C. Okafor, K.D.K.A Somarathne, A. Hayakawa, T. Kudo, O. Kurata, N. Iki,
air combustion in a micro gas turbine combustor with up to 30% H. Kobayashi, Towards the development of an efficient low-NOx ammonia
of ammonia in the fuel in terms of heat fraction. The following is combustor for a micro gas turbine, Proc. Combust. Inst. 37 (2019) 4597–4606.
[7] A. Hayakawa, Y. Arakawa, R. Mimoto, K.D.K.A. Somarathne, T. Kudo,
a summary of major findings in this study. H. Kobayashi, Int. J. Hydrogen Energy 42 (2017) 14010–14018.
[8] K.D.K.A. Somarathne, A. Hayakawa, H. Kobayashi, Numerical investigation on
the combustion characteristics of turbulent premixed ammonia/air flames sta-
• OH-PLIF and simultaneous OH/NO-PLIF measurements showed
bilized by a swirl burner, J. Fluid Sci. Technol. 11 (2016) JFST0026.
that local lean regions with high OH and NO concentrations [9] K.D.K.A. Somarathne, S. Hatakeyama, A. Hayakawa, H. Kobayashi, Numerical
in the fuel-rich primary combustion zone of the rich-lean non- study of a low emission gas turbine like combustor for turbulent ammonia/air
premixed combustor may contribute to the higher NOx emis- premixed swirl flames with a secondary air injection at high pressure, Int. J.
Hydrogen Energy 42 (2017) 27388–27399.
sions from the combustor in comparison to the rich-lean pre- [10] K.D.K.A. Somarathne, S. Colson, A. Hayakawa, H. Kobayashi, Modelling of am-
mixed combustor. monia/air non-premixed turbulent swirling flames in a gas turbine-like com-
• LES of two stage combustion of CH4 –NH3 –air mixtures shows bustor at various pressures, Combust. Theory Model. 22 (2018) 973–997.
[11] O. Kurata, N. Iki, T. Inoue, T. Matsunuma, T. Tsujimura, H. Furutani, M. Kawano,
that a fuel-rich primary combustion zone ensures low NO pro- K. Arai, E.C. Okafor, A. Hayakawa, H. Kobayashi, Development of a wide range–
duction but with relatively high H2 , CO, HCN and unburned operable, rich-lean low-NOx combustor for NH3 fuel gas-turbine power gener-
fuels concentration. The combustion of CO and H2 in the sec- ation, Proc. Combust. Inst. 37 (2019) 4587–4595.
[12] T. Terasaki, S. Hayashi, The effects of fuel-air mixing on NOx formation in
ondary combustion zone produces high concentration of O/H non-premixed swirl burners, Proc. Combust. Inst. 26 (1996) 2733–2739.
radicals that facilitate the conversion of HCN and NH3 mainly [13] N. Syred, A review of oscillation mechanisms and the role of the precessing
to NO. vortex core (PVC) in swirl combustion systems, Prog. Energy Combust. Syst. 32
(2) (2006) 93–161.
• Although single stage combustion of CH4 –NH3 –air resulted in
[14] A. Valera-Medina, N. Syred, A. Griffiths, Visualization of isothermal large co-
about twice more NOx emissions than that of NH3 -air, rich-lean herent structures in a swirl burner, Combust. Flame 156 (2009) 1723–1734.
combustion of CH4 –NH3 –air emitted less NOx than NH3 –air be- [15] V.M. Reddy, A. Katoch, W.L. Roberts, S. Kumar, Experimental and numerical
analysis for high intensity swirl based ultra-low emission flameless combustor
cause the higher flame speed of CH4 –NH3 –air mixtures ensured
operating with liquid fuels, Proc. Combust. Inst. 35 (2015) 3581–3589.
lower NOx production in the secondary combustion zone. [16] P. Kumar, T.R. Meyer, Experimental and modeling study of chemical-kinetics
• The optimum low NOx primary zone equivalence ratio for mechanisms for H2-NH3-air mixtures in laminar premixed jet flames, Fuel 108
CH4 –NH3 –air combustion was found to vary from 1.30 to 1.35 (2013) 166–176.
[17] J. Li, H. Huang, N. Kobayashi, Z. He, Y. Nagai, Study on using hydrogen and
as ammonia heat fraction in the fuel decreased from 0.30 to ammonia as fuels: combustion characteristics and NOx formation, Int. J. Energy
0.10. Res. 38 (2014) 1214–1223.
• Significantly low emissions such as 49 ppmv of NOx, 2 ppmv of [18] A. Ichikawa, A. Hayakawa, Y. Kitagawa, K.D.K.A. Somarathne, T. Kudo,
H. Kobayashi, Laminar burning velocity and Markstein length of ammo-
CO and approximately zero N2 O, HCN and NH3 emissions with nia/hydrogen/air premixed flames at elevated pressures, Int. J. Hydrogen En-
a combustion efficiency of 99.8% were achieved using rich-lean ergy 40 (2015) 9570–9578.
premixed combustion at 0.25 MPa. [19] D. Pugh, P. Bowen, A. Valera-Medina, A. Giles, J. Runyon, R. Marsh, Influence of
steam addition and elevated ambient conditions on NOx reduction in a staged
premixed swirling NH3 /H2 flame, Proc. Combust. Inst. 37 (2019) 5401–5409.
[20] R.C. Rocha, M. Costa, X.-.S. Bai, Chemical kinetic modelling of ammo-
nia/hydrogen/air ignition, premixed flame propagation and NO emission, Fuel
Declaration of Competing Interest 246 (2019) 24–33.
[21] X. He, B. Shu, D. Nascimento, K. Moshammer, M. Costa, R.X. Fernandes, Au-
The authors declare that they have no known competing finan- to-ignition kinetics of ammonia and ammonia/hydrogen mixtures at interme-
diate temperatures and high pressures, Combust. Flame 206 (2019) 189–200.
cial interests or personal relationships that could have appeared to [22] A.J. Reiter, S.C. Kong, Combustion and emissions characteristics of compres-
influence the work reported in this paper. sion-ignition engine using dual ammonia-diesel fuel, Fuel 90 (2011) 87–97.
416 E.C. Okafor, K.D. .A. Somarathne and R. Ratthanan et al. / Combustion and Flame 211 (2020) 406–416

[23] A.J. Reiter, S.C. Kong, Demonstration of compression-ignition engine combus- [38] A. Valera-Medina, R. Marsh, J. Runyon, D. Pugh, P. Beasley, T. Hughes, P. Bowen,
tion using ammonia in reducing greenhouse gas emissions, Energy Fuels 22 Ammonia-methane combustion in tangential swirl burners for gas turbine
(2008) 2963–2971. power generation, Appl. Energy 185 (2017) 1362–1371.
[24] A. Valera-Medina, S. Morris, J. Runyon, D.G. Pugh, R. Marsh, P. Beasley, [39] OpenFOAM 1.7.1 (2010) Available at:http://openfoam.org/version/1- 7- 1/.
T. Hughes, Ammonia, methane and hydrogen for gas turbines, Energy Proce- [40] T. Poinsot, D. Veynante, Theoretical and Numerical Combustion, 2nd ed., R.T.
dia 75 (2015) 118–123. Edwards, Inc., Pennsylvania, USA, 2005, pp. 164–169.
[25] E.C. Okafor, Y. Naito, S. Colson, A. Ichikawa, T. Kudo, A. Hayakawa, H. Kobayashi, [41] F. Nicoud, F. Ducros, Subgrid-Scale stress modelling based on the square of the
Experimental and numerical study of the laminar burning velocity of velocity gradient tensor, Flow Turbul. Combust. 62 (1999) 183–200.
CH4 –NH3 –air premixed flames, Combust. Flame 187 (2018) 185–198. [42] S. Menon, P.-.K. Yeung, W.-.W. Kim, Effect of subgrid models on the computed
[26] E.C. Okafor, Y. Naito, S. Colson, A. Ichikawa, T. Kudo, A. Hayakawa, H. Kobayashi, interscale energy transfer in isotropic turbulence, Comput. Fluids 25 (1996)
Measurement and modelling of the laminar burning velocity of methane-am- 165–180.
monia-air flames at high pressures using a reduced reaction mechanism, Com- [43] S.M. Correa, Turbulence-chemistry interactions in the intermediate regime of
bust. Flame 204 (2019) 162–175. premixed combustion, Combust. Flame 93 (1993) 41–60.
[27] A. Ichikawa, Y. Naito, A. Hayakawa, T. Kudo, H. Kobayashi, Burning velocity and [44] G. Bulat, E. Fedina, C. Fureby, W. Meier, U. Stopper, Reacting flow in an indus-
flame structure of CH4 /NH3 /air turbulent premixed flames at high pressure, trial gas turbine combustor: les and experimental analysis, Proc. Combust. Inst.
Int. J. Hydrogen Energy 44 (2019) 6991–6999. 35 (2015) 3175–3183.
[28] S. Li, S. Zhang, H. Zhou, Z. Ren, Analysis of air-staged combustion of NH3 /CH4 [45] T. Lucchini, F. Contino, G.D. Errico, Coupling of in situ adaptive tabulation and
mixture with low NOx emission at gas turbine conditions in model combus- dynamic adaptive chemistry: an effective method for solving combustion in
tors, Fuel 237 (2019) 50–59. engine simulations, Proc. Comb. Inst. 33 (2011) 3057–3064.
[29] C.F. Ramos, R.C. Rocha, P.M.R. Oliveira, M. Costa, X.-.S. Bai, Experimental and ki- [46] J.A. Miller, M.D. Smooke, R.M. Green, R.J. Kee, Kinetic modelling of the oxida-
netic modelling investigation on NO, CO and NH3 emissions from NH3 /CH4 /air tion of ammonia in flames, Combust. Sci. Technol. 34 (1983) 149–176.
premixed flames, Fuel 254 (2019) 115693. [47] X. Han, Z. Wang, M. Costa, Z. Sun, Y. He, K. Cen, Experimental and kinetic mod-
[30] H. Xinlu, W. Zhihua, M. Costa, S. Zhiwei, H. Yong, C. Kefa, Experimental and elling study of laminar burning velocities of NH3 /air, NH3 /H2 /air, NH3 /CO/air
kinetic modeling study of laminar burning velocities of NH3 /air, NH3 /H2 /air, and NH3 /CH4 /air premixed flames, Combust. Flame 206 (2019) 214–226.
NH3 /CO/air and NH3 /CH4 /air premixed flames, Combust. Flame 206 (2019) [48] CHEMKIN-PRO 17.2, ANSYS, Inc., San Diego (2016)
214–226. [49] M. Stöhr, I. Boxx, C.D. Carter, W. Meier, Experimental study of vortex-flame in-
[31] P. Glarborg, J.A. Miller, B. Ruscic, S.J. Klippenstein, Modelling nitrogen chem- teraction in a gas turbine model combustor, Combust. Flame 159 (2012) 2649.
istry in combustion, Prog. Energy Combust. Sci. 67 (2018) 31–68. [50] U. Stopper, W. Meier, R. Sadanandan, M. Stöhr, M. Aigner G. Bulat, Experimen-
[32] V.J. Wargadalam, G. Loffler, F. Winter, H. Hofbauer, Homogenous formation of tal study of industrial gas turbine flames including quantification of pressure
NO and N2 O from the oxidation of HCN and NH3 at 60 0-10 0 0 °C, Combust. influence on flow field, fuel/air premixing and flame shape, Combust. Flame
Flame 120 (20 0 0) 465–475. 160 (2013) 2103–2118.
[33] T. Takagi, T. Tatsumi, M. Ogasawara, Nitric oxide formation from fuel nitrogen [51] J.W. Daily, Laser induced fluorescence spectroscopy in flames, Prog. Energy
in staged combustion: roles of HCN and NHi, Combust. Flame 35 (1979) 17–25. Combust. Sci. 23 (1997) 133–199.
[34] J.B. Mereb, J.O.L. Wendt, Reburning mechanisms in a pulverized coal combus- [52] Y. Fan, W. Lin, S. Wan, Y. Suzuki, Investigation of wall chemical effect using
tor, Proc. Combust. Inst. 23 (1990) 1273–1279. PLIF measurement of OH radical generated by pulsed electric discharge, Com-
[35] P.G. Kristensen, P. Glarborg, K. Dam-Johansen, Nitrogen chemistry during bust. Flame 196 (2018) 255–264.
burnout in fuel-staged combustion, Combust. Flame 107 (1996) 211–222. [53] R. Sadanandan, M. Stöhr, W. Meier, Simultaneous OH-PLIF and PIV measure-
[36] P. Dagaut, P. Glarborg, M.U. Alzueta, The oxidation of hydrogen cyanide and ments in a gas turbine model combustor, Appl. Phys. B 90 (2008) 609–618.
related chemistry, Prog. Energy Combust. Sci. 34 (2008) 1–46. [54] S.J. Klippenstein, L.B. Harding, P. Glarborg, J.A. Miller, The role of NNH in NO
[37] J. Jójka, R. Ślefarski, Dimensionally reduced modeling of nitric oxide forma- formation and control, Combust. Flame 158 (2011) 774–789.
tion for premixed methane-air flames with ammonia content, Fuel 217 (2018)
98–105.

You might also like