You are on page 1of 7

Combustion and Flame 217 (2020) 314–320

Contents lists available at ScienceDirect

Combustion and Flame


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

The temperature dependence of the laminar burning velocity and


superadiabatic flame temperature phenomenon for NH3 /air flames
Xinlu Han a, Zhihua Wang a,∗, Yong He a, Yingzu Liu a, Yanqun Zhu a, Alexander A. Konnov b
a
State Key Laboratory of Clean Energy Utilization, Zhejiang University, Hangzhou 310027, PR China
b
Division of Combustion Physics, Lund University, Lund S-22100, Sweden

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

Article history: Combustion of ammonia (NH3 ) as a carbon-free alternative fuel has been recently widely studied, with
Received 10 March 2020 vast majority of the burning velocity data obtained at room temperature. In the present study, the lami-
Revised 16 April 2020
nar burning velocity SL of NH3 /air mixtures has been measured at unburnt gas temperature Tu from 298 K
Accepted 17 April 2020
to 448 K, covering equivalence ratios from 0.85 to 1.25 and at 1 atm using the heat flux method. Kinetic
Available online 7 May 2020
simulations were made with five literature mechanisms developed for NH3 combustion, i.e., Nakamura et
Keywords: al., Otomo et al., San Diego, Okafor et al., and Mei et al. mechanisms, and the influence of radiation heat
Laminar burning velocity losses was considered. Using the obtained burning velocity data at different temperatures, the temper-
Elevated temperature ature dependence coefficients α in SS0L = ( TTu0 )α were derived, and compared with different models’ pre-
L u
Temperature dependence
dictions. Further analyses of the temperature dependence of SL were carried out through examination of
Ammonia flame
the overall activation energy, temperature and species profiles as well as the reaction paths, and a unique
flame structure at the rich side of adiabatic NH3 /air flames was found, which resembles ‘over-rich’ phe-
nomena in hydrocarbon flames. At equivalence ratio larger than 1.1 ± 0.05, the NH3 /air flames become
so rich that (1) the NH2 radical overwhelms the H and OH radicals in maximum mole fraction; (2) after
the flame front, H2 O converts back to H2 with NO formed at the same time, causing the superadiabatic
flame temperature phenomena, i.e. adiabatic flame temperature being lower than the maximum achieved
in the flame. Moreover, local minimum NO concentration is found right after the over-rich NH3 /air flame
front, which may be helpful in reducing NO emissions from NH3 flames in practical applications.
© 2020 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction the same pressure range Mei et al. [8] provided several oxy-fuel
NH3 burning velocity data. There are also NH3 /O2 flame results re-
Ammonia (NH3 ) has recently gained more and more attention ported by Liu et al. [9] where the effect of pressure from 0.5 to
by the combustion community as a promising alternative fuel. 1.6 atm was investigated. However, it is noted that these literature
Comparing with traditional fossil fuels, NH3 releases no carbon data for NH3 /air flames are all at room temperature, and the data
dioxide (CO2 ) when burning, which is advantageous facing the obtained at elevated temperatures are only found in the early work
threat of global warming. Also, due to its matured usage in, e.g., of Cohen [10] from 323 to 423 K employing a flat flame burner,
fertilizer industry, the fuel NH3 can be incorporated into modern and in the recent work of Lhuillier et al. [11] from 298 to 473 K
life with less investment on its production and transportation. using spherical flame method.
As an important global combustion characteristic, laminar burn- Besides, temperature dependence of NH3 burning velocity in
ing velocity (SL ) of NH3 has been investigated in several experi- the form of Eq. (1) [12,13],
mental studies. For the measurements fuelled with pure NH3 , most  α
experiments were concentrated on NH3 /air combustion at room SL Tu
= (1)
temperature and pressure, e.g., Takizawa et al. [1], Ronney [2], SL0 Tu0
Pfahl et al. [3], Jabbour and Clodic [4], and Han et al. [5]. The
and the corresponding α coefficient are also seldom investigated
NH3 /air burning velocities at pressures up to 5 atm have been
in literature. The α coefficient is important because with SL0 at a
measured by Hayakawa et al. [6] and Ichikawa et al. [7], and for
reference temperature Tu0 (e.g., 298 K) known, specified α value is
helpful for deriving SL at various Tu ,which is also powerful indica-

Corresponding author. tor of the SL data consistency [14,15]. For NH3 /air flames, the only
E-mail address: wangzh@zju.edu.cn (Z. Wang). experimentally derived α values from Cohen [10] and Lhuillier et

https://doi.org/10.1016/j.combustflame.2020.04.013
0010-2180/© 2020 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
X. Han, Z. Wang and Y. He et al. / Combustion and Flame 217 (2020) 314–320 315

al. [11] are in conflict. Cohen [10] reported the α coefficient to be


approximately equal to unity, while that derived by Lhuillier et al.
[11] is around 2 at stoichiometry. A fitted α correlation equation
from numerical simulations can be found in the work by Gold-
mann and Dinkelacker [16] using kinetic mechanism from Mathieu
and Petersen [17]. Considering the uncertainty in experiments and
kinetic modelling, the limited works are insufficient for accurate α
evaluation of NH3 /air flames.
In light of the background outlined above, the aims of this
investigation are: (1) to measure the laminar burning velocity of
NH3 /air flames at different unburnt temperatures; (2) to validate
kinetic mechanisms using the experimental SL ; (3) to evaluate the
temperature coefficient α and analyse the flame structure changes
at different equivalence ratios from experimental and numerical
data.

2. Experimental details

The experiment setup used herein is similar to that in our re-


cent work on NH3 flames [5]. Atmospheric adiabatic burning veloc-
ities of NH3 /air flames were measured by the heat flux method at
unburnt temperatures 298–448 K with 25 K step. For each unburnt
temperature, the measured equivalence ratios φ were changed
from 0.85 to 1.25 with 0.05 increment. Influence of the buoyancy
instability and NH3 adsorption on metal surfaces of the burner was
considered negligible for the measurement results [5]. The mea-
sured burning velocities at 298 K (Tu0 ) were used as the reference
SL0 when deriving Eq. (2).
 
ln SL /SL0
αTu →Tu0 =   (2)
ln Tu /Tu 0

The uncertainty of the measured SL , SL , was evaluated from


the thermocouple readout scatterings of the burner temperature as
well as from the uncertainties arising from Mass Flow Controllers.
The uncertainty of the α coefficient derived from experimental SL
was evaluated through the uncertainty propagation as
 
1 S L SL0
αTu0 →Tu =   + (3)
ln Tu /Tu0 SL SL0

More details about the experimental uncertainty can be found


in our previous publications [5,18,19].

3. Numerical details

Flame simulations in this work were carried out using ANSYS Fig. 1. Laminar burning velocity of NH3 /air flames at 1 atm, (a) present experimen-
CHEMKIN 17.0 [20], Premixed Flame Speed module, and calcula- tal results at different Tu ; (b) 298 K data comparing with literature experimental
and simulation results considering radiative heat loss; (c) 448 K data comparing
tions of the equilibrium state were achieved through the Equi- with simulation results considering radiative heat loss.
librium module. The literature kinetic mechanisms used here are
Nakamura et al. [21], Otomo et al. [22], San Diego [23], Okafor et
al. [24], and Mei et al. [8], all targeting at the oxidation of NH3 as
main fuel component. Soret effect and multicomponent transport 4. Results and discussion
were considered in all simulation cases.
Nakamura et al. [25] reported that considering radiation heat 4.1. Burning velocity and its temperature dependence
loss strongly influences the simulated burning velocity results of
NH3 /air flames, which mimics more the situation of real exper- Figure 1 shows the present measurements of the laminar burn-
imental flames with inevitable flame radiation. Therefore, in the ing velocity at different Tu . At 298 K, the present experimental re-
present study, simulations with radiative heat loss were made to sults agree well with the literature data from Han et al. [5], Tak-
compare with the experimental SL and the derived α coefficients. izawa et al. [1], Ronney [2], Pfahl et al. [3], Jabbour and Clodic [4],
Calculations of the radiative losses were achieved by adding the Hayakawa et al. [6], Ichikawa et al. [7], Mei et al. [8], and Lhuillier
absorption coefficients of NH3 , NO, N2 O, and H2 O species [25] into et al. [11]. Error bars show the experimental uncertainty, which is
the thermodynamic input file of different mechanisms and then around 10% for majority of the experimental data. This uncertainty
enabling the ‘Include Gas Radiation’ option in CHEMKIN. Separate range is not as fine as that for hydrocarbon flames using the same
simulations without radiative heat loss were also made in order to heat flux method, which is mostly attributed to the low measure-
analyse the adiabatic flame structure of NH3 /air flames. ment sensitivity [18] of NH3 flames resulting in the increased un-
316 X. Han, Z. Wang and Y. He et al. / Combustion and Flame 217 (2020) 314–320

ln (SL /SL0 )
Fig. 2. α = ln (Tu /Tu0 )
derived from Tu0 = 298 K and Tu = 448 K for NH3 /air flames.

certainty caused by thermocouple readout scattering. The compar- Fig. 3. Ea


and x fitted using adiabatic simulation results.
2R
ison of the present experimental data at 423 K with the literature
data from Cohen [10] and Lhuillier et al. [11] can be found in the
Supplementary material. α with other datasets show clearly that the present SL results are
Figure 1 also includes simulation results considering radiative consistent at different Tu .
heat losses. For each mechanism, same tendency is found at 298 K
and 448 K, while due to the exponential increase of SL at ele- 4.2. Adiabatic flame structure analyses
vated Tu (see Eq. (1)), larger absolute discrepancies among different
mechanisms are found at 448 K. Simulations using Nakamura et In the following, analyses were carried out using adiabatic sim-
al. mechanism [21] are always higher than the present experimen- ulation results, thus focusing on the flame structure changes aris-
tal data, regardless of equivalence ratio φ . Mechanisms from San ing from chemistry. Considering the oxidation process using a one-
Diego [23] and Mei et al. [8] under-estimate by around 20% the step global reaction, the overall activation energy Ea and reaction
fuel-lean and over-estimate over 40% the fuel-rich side SL results order n should be different when there are different flame struc-
of the present experiments. Overall, Otomo et al. [22] and Okafor tures, which then can be identified. Ea and a synthetic parameter
et al. [24] mechanisms have the best prediction performance, of related to reaction order, x, can be derived from the α coefficient
which the relative difference compared with the present experi- through Eq. (4) [18],
mental data is within ± 10%.
Figure 2 shows the temperature coefficient α derived from the Ea
αTu0 →Tu = ·X +x (4)
present experimental SL data and from the simulated SL with ra- 2R
diative heat loss. Also shown are the experimentally derived val- where X is related to the adiabatic flame temperature. Analyses
ues from Lhuillier et al. [11] and correlation reported by Gold- of Ea using this equation have helped identify the flame struc-
mann and Dinkelacker [16]. For the experimental data, minimum ture change between the moderate-rich and over-rich regions of
α 298 K → 448 K = 1.86 is found at φ = 1.1 with also lowest un- methane, ethane, and propane+air flames in our previous publi-
certainty of αT 0 →Tu = 0.54. The simulation results by San Diego cations [18,19], where details of the derivation procedure can be
u
[23] and Mei et al. [8] mechanisms have best agreement with found.
the experimental data, in contrast with their inferior performance Figure 3 shows the relationship of 2ERa and x vs. φ . The selected
on predicting SL compared to Otomo et al. [22] and Okafor et al. four mechanisms reproduce well either the experimental SL or α
[24] mechanisms in Fig. 1. The present α values agree well with in Figs. 1 and 2. In Fig. 3, rich-side turning points at φ = 1.15 are
that from Lhuillier et al. [11] at φ = 1.0 and 1.2, while the φ = 0.8 found from the simulation results using Otomo et al. [22], Okafor
point was not measured in the present experiments. As has been et al. [24], and Mei et al. [8] mechanisms, where the overall activa-
mentioned in Introduction, the α value of Cohen [10] is equal to tion energy Ea and x reaching valley and peak values, respectively.
unity, which is significantly lower than the present and Lhuillier Results by San Diego mechanism [23] show similar tendency, while
et al. [11] data, thus not shown in the figure. The α derived from the turning point being leaner at φ = 1.1. The presence of the turn-
Goldmann and Dinkelacker [16] correlation exceeds by 10% that ing points in Ea and x profiles indicates that there may be flame
derived from the present experiment and simulations at around structure change in the fuel-rich side of NH3 /air flames.
φ = 1.1, which can be attributed to uncertainties of the selected Figure 4 shows adiabatic flame simulation results at 298 K of
mechanism [17] and of the correlation fitting processes. Effect of the spatial profiles of temperature, net heat release rate, and mole
radiative heat loss consideration on the simulated α is shown in fractions of H2 O, H2 , OH, H, N2 , NO at φ = 1.1 and φ = 1.15.
the Supplementary material to be notable at especially the off- The net heat release rate was adopted directly from the ANSYS
stoichiometry sides, indicating the need to consider it in validating CHEMKIN output. Results by Otomo et al. [22], Okafor et al. [24],
not only burning velocity of ammonia, but also temperature coef- and Mei et al. [8] mechanisms were chosen because they have
ficients α . Since α is a powerful indicator of the SL data consis- turning points at the same equivalence ratio in Fig. 3. In Fig. 4, they
tency [14,15], the agreement of the present experimentally derived show the same tendencies in the spatial profiles investigated. Con-
X. Han, Z. Wang and Y. He et al. / Combustion and Flame 217 (2020) 314–320 317

Fig. 4. Spatial profiles of temperature and net heat release rate (a) and (b), and mole fractions of H2 O and H2 (c) and (d); OH and H (e) and (f); N2 and NO (g) and (h) at
298 K. Panels in the left and right columns represent φ = 1.1 and φ = 1.15 flames, respectively. Position of the maximum net heat release rate set at distance = 1 cm.
318 X. Han, Z. Wang and Y. He et al. / Combustion and Flame 217 (2020) 314–320

Fig. 5. Reaction pathways in NH3 /air flames calculated using Okafor et al. mecha- Fig. 6. Rate of production of OH at the distance = 3.0 cm in NH3 /air flames calcu-
nism at 298 K. Left: φ = 1.1; right: φ = 1.15. lated using Okafor et al. mechanism at 298 K.

trary to φ = 1.1 flame, simulation results at φ = 1.15 show supera-


diabatic flame temperature (SAFT), i.e., the maximum flame tem-
perature exceeding the equilibrium (adiabatic) one, which is in line
with the negative net heat release rate after the flame front staying
for a much longer distance at φ = 1.15 than at φ = 1.1. It is also
noticed that after the flame front, the [H2 O] and [H2 ] for φ = 1.15
flame gradually decreases and increases, respectively, while this
tendency can’t be found in φ = 1.1 flame. Based on the above com-
parison, the H2 O is apparently converting to H2 in φ = 1.15 NH3
flame tail (the region downstream the large temperature gradient
flame front), which was also found for the methane, ethane, and
propane+air flames in their over-rich regions [19]. For the [OH]
and [H] profiles in Fig. 4(e) and (f), decreases of their concentra-
tions right after the flame front are found at both equivalence ra-
tios, however after this decrease, [OH] and [H] take a much longer
distance to reach the equilibrium states for the φ = 1.15 flame than Fig. 7. Maximum mole fractions of dominant intermediate species in NH3 /air
for φ = 1.1 flame. Difference can hardly be found in the [N2 ] pro- flames, simulated at 298 K using Okafor et al. mechanism.
files, while in the same panels (g, h), the [NO] continually goes
down at φ = 1.1, and shows non-monotonic behaviour at φ = 1.15.
The differences discussed above confirm essential changes in the consumed partly by R2 and R3 at φ = 1.15, and the increase of [H]
flame structure around the turning points seen in Fig. 3. in Fig. 4 can be attributed to the production of H through these
Figure 5 shows the reaction pathways in φ = 1.1 and φ = 1.15 two reactions together with that through R5. R4 also consumes
NH3 /air flames. Data in this figure were taken from the 298 K H2 O, together with the H2 O consumption through R1, responsible
simulation results using Okafor et al. [24] mechanism, at a dis- for the decrease of [H2 O] in Fig. 4. At the same time, NO is pro-
tance = 3 cm (2 cm downstream the position of the maximum duced by R2; and from the analyses of HNO ROP (not shown here),
net heat release rate). Obvious difference can be found in Fig. 5 for the HNO formed by R3 is quickly consumed by R5 to form NO, ex-
the direction of arrows connecting H2 and H2 O species, validating plaining the increase of [NO] in Fig. 4.
the assumption discussed above that there is H2 O–H2 conversion
in φ = 1.15 NH3 +air flame tail. For φ = 1.1 and φ = 1.15 reac- OH + H2 = H + H2 O (R1)
tion pathways, direction differences are also noted for the arrows
attaching to OH. N + OH = NO + H (R2)
Figure 6 shows the rate of production (ROP) of OH, which fol-
lows the finding of the pathways arrow direction changes in Fig. 5. NH + OH = HNO + H (R3)
The mechanism and distance chosen in Fig. 6 are the same as
those in Fig. 5. Though the total OH ROP is nearly the same for H + OH + M = H2O +M (R4)
φ = 1.1 and φ = 1.15 flames, the absolute ROP contribution of
several reactions is found dramatically larger for φ = 1.15 than H + NO + M = HNO + M (R5)
for φ = 1.1 flames, therefore the ROP at φ = 1.1 has been mul-
tiplied by 10 in Fig. 6 to facilitate comparison. There are 3 re- In contrast to the φ = 1.15 case discussed above, at φ = 1.1,
actions, R1-R3, showing opposite sign of ROP at different equiva- directions of R1-R3 and R5 are all different and the absolute ROP
lence ratios. At φ = 1.15, R1 has the largest positive OH ROP, and of R1-R3 is indeed small, resulting in that (1) [H2 O], [H2 ], [OH], [H]
the increase of [H2 ] in Fig. 4 at the distance = 3.0 cm can be ex- staying unchanged with the net heat release rate approaching zero
plained since R1 produces OH and H2 at the same time. The SAFT and no SAFT found; (2) [NO] decreases at the distance = 3.0 cm.
and negative net heat release rate can also be understood since the Figure 7 shows the maximum mole fractions of dominant in-
H2 O–H2 conversion through R1 is endothermic. The OH formed is termediate species in NH3 /air flames, in order to inspect the flame
X. Han, Z. Wang and Y. He et al. / Combustion and Flame 217 (2020) 314–320 319

temperature thus lower equilibrium NO concentrations than the


adiabatic CH4 flames, while NO is abundant in the NH3 flame front
(see Figs. 4(g) and (h) and 7), and its concentration can only de-
crease in very long adiabatic post flame zone, which is hard to
achieve in real applications, making NOx emission a main concern
in the NH3 combustion. The present recognition of the over-rich
NH3 flame structure is therefore important because there is al-
ways a local minimum of [NO] within 1 cm after the over-rich
flame front (see Fig. 4(h)), i.e., a low [NO] level can be reached
with much shorter adiabatic zone needed, which may be helpful
for reducing the NOx emission in practical applications.

5. Conclusions

New measurements of the NH3 /air burning velocities at differ-


ent unburnt temperatures and 1 atm have been performed using
Fig. 8. Differences of the local maximum flame temperature and corresponding the heat flux method. Five recently published ammonia combus-
equilibrium flame temperature for NH3 /air flames at 298 K. tion mechanisms were compared with the experimental data. The
temperature coefficients α of NH3 /air flames were derived from
the burning velocity results, and obvious differences were found
structure change and identify important radicals at different φ . At in the simulation results with and without radiation heat loss.
the rich side of flame compositions, with φ increasing, the max- By analyses of the overall activation energy, temperature and
imum concentrations of all radicals go down except NH2 , whose species profiles as well as the reaction pathways, characteristics of
maximum concentration increases gradually and overwhelms that the over-rich NH3 /air flames were revealed. At equivalence ratio
of OH, H, and NO at φ > 1.15. From the integrated ROP analyses larger than 1.1 ± 0.05, the NH3 +air adiabatic flame becomes so
by Okafor et al. [24] and in our previous study [5], typical fuel- rich that (1) the NH2 radical overwhelms the H and OH radicals
rich species N2 H2 , N2 H3 , N2 H4 , and H2 NN are generated mostly by in maximum mole fraction; (2) after the flame front, H2 O con-
NH2 , and thus the relative abundance of NH2 compared to other verts back to H2 with NO formed at the same time, causing the
species points out that the flame is ‘very rich’. Based on this char- superadiabatic flame structure. In the over-rich NH3 /air flames, a
acteristic, the names ‘moderate-rich’ and ‘over-rich’ can be given point with local minimum NO concentration is found right after
to the regions before and after the aforementioned flame structure the flame front, which may be helpful in reducing the NO emis-
turning point, respectively. sions from NH3 flames in real applications.
Analyses presented from Figs. 3–7 have been repeated using
different mechanisms as well as at different unburnt temperatures,
Declaration of Competing Interest
similar results were found with the turning equivalence ratios to
over-rich flames. They are slightly different using different mech-
The authors declare that they have no known competing finan-
anisms, nevertheless all indicating that the shown over-rich flame
cial interests or personal relationships that could have appeared to
characteristics are common features and not a peculiarity arising
influence the work reported in this paper.
from a certain mechanism.
Figure 8 shows differences between the local maximum flame
Acknowledgments
temperature (or maximum temperature at the end point of the
computational domain if there is no SAFT behavior) and corre-
This work was supported by the National Natural Science Foun-
sponding equilibrium flame temperature for NH3 /air flames at
dation of China (51876192, 51621005) and National Scholarship for
298 K, which can be used to identify the beginning of the over-
Building High Level Universities (China Scholarship Council).
rich regions predicted by different mechanisms through the inter-
ception points of T = Tmax − Teq and T = 0 lines. Different T
values are seen for different mechanisms in Fig. 8, especially at Supplementary material
the fuel-rich side; and the predicted over-rich turning points vary
within φ = 1.1 ± 0.05. It is also noted that at the leaner side, the Supplementary material associated with this article can be
local maximum temperature, which is around the flame front, is found, in the online version, at doi:10.1016/j.combustflame.2020.
lower than the equilibrium temperature by about 10 K, indicating 04.013.
that there are exothermic reactions slowly releasing heat at the far
away tails of NH3 /air flames. References
Compared with the over-rich flames of hydrocarbons [19], the
[1] K. Takizawa, A. Takahashi, K. Tokuhashi, S. Kondo, A. Sekiya, Burning velocity
over-rich NH3 /air flames share the same characteristics of the SAFT measurements of nitrogen-containing compounds, J. Hazard. Mater. 155 (2008)
and H2 O–H2 conversion after the flame front, while several dif- 144–152.
ferences can also be found. Firstly, there are no obvious over-rich [2] P.D. Ronney, Effect of chemistry and transport properties on near-limit flames
turning points shown in the NH3 /air temperature coefficient α pro- at microgravity, Combust. Sci. Technol. 59 (1988) 123–141.
[3] U. Pfahl, M. Ross, J. Shepherd, K. Pasamehmetoglu, C. Unal, Flammability lim-
files in Fig. 2. Secondly, in the absence of C element in NH3 , the its, ignition energy, and flame speeds in H2 –CH4 –NH3 –N2 O–O2 –N2 mixtures,
radical that overwhelms the importance of H and OH radicals in Combust. Flame 123 (20 0 0) 140–158.
[4] T. Jabbour, D.F. Clodic, J. Terry, S. Kondo, Burning velocity and refrigerant
the flame front is NH2 in NH3 /air flames rather than CH3 in hy-
flammability classification, ASHRAE Trans. 110 (2004) 522–533.
drocarbon flames. Also, the O element (in the form of OH radi- [5] X. Han, Z. Wang, M. Costa, Z. Sun, Y. He, K. Cen, Experimental and kinetic mod-
cal) released by the H2 O–H2 conversion in the over-rich NH3 /air eling study of laminar burning velocities of NH3/air, NH3/H2/air, NH3/CO/air
flames forms NO rather than CO, resulting in the local increase of and NH3/CH4/air premixed flames, Combust. Flame 206 (2019) 214–226.
[6] A. Hayakawa, T. Goto, R. Mimoto, Y. Arakawa, T. Kudo, H. Kobayashi, Laminar
NO concentration after the flame front. As has been discussed by burning velocity and Markstein length of ammonia/air premixed flames at var-
Kobayashi et al. [26], the adiabatic NH3 flames have lower flame ious pressures, Fuel 159 (2015) 98–106.
320 X. Han, Z. Wang and Y. He et al. / Combustion and Flame 217 (2020) 314–320

[7] A. Ichikawa, A. Hayakawa, Y. Kitagawa, K.K.A. Somarathne, T. Kudo, [17] O. Mathieu, E.L. Petersen, Experimental and modeling study on the high-tem-
H. Kobayashi, Laminar burning velocity and Markstein length of ammo- perature oxidation of ammonia and related NOx chemistry, Combust. Flame
nia/hydrogen/air premixed flames at elevated pressures, Int. J. Hydrogen En- 162 (2015) 554–570.
ergy 40 (2015) 9570–9578. [18] X. Han, Z. Wang, S. Wang, R. Whiddon, Y. He, Y. Lv, A.A. Konnov, Parametriza-
[8] B. Mei, X. Zhang, S. Ma, M. Cui, H. Guo, Z. Cao, Y. Li, Experimental and ki- tion of the temperature dependence of laminar burning velocity for methane
netic modeling investigation on the laminar flame propagation of ammonia and ethane flames, Fuel 239 (2019) 1028–1037.
under oxygen enrichment and elevated pressure conditions, Combust. Flame [19] X. Han, Z. Wang, Y. He, S. Wang, Y. Zhu, A.A. Konnov, Over-rich combustion
210 (2019) 236–246. of CH4, C2H6, and C3H8 +air premixed flames investigated by the heat flux
[9] Q. Liu, X. Chen, J. Huang, Y. Shen, Y. Zhang, Z. Liu, The characteristics of method and kinetic modeling, Combust. Flame 210 (2019) 339–349.
flame propagation in ammonia/oxygen mixtures, J. Hazard. Mater. 363 (2019) [20] CHEMKIN-PRO 17.0 (15151), ANSYS Reaction Design: San Diego (2016).
187–196. [21] H. Nakamura, S. Hasegawa, T. Tezuka, Kinetic modeling of ammonia/air weak
[10] L. Cohen, Burning velocities of ammonia in air and in oxygen, Fuel 34 (1955) flames in a micro flow reactor with a controlled temperature profile, Combust.
S123–S127. Flame 185 (2017) 16–27.
[11] C. Lhuillier, P. Brequigny, N. Lamoureux, F. Contino, C. Mounaïm-Rous- [22] J. Otomo, M. Koshi, T. Mitsumori, H. Iwasaki, K. Yamada, Chemical kinetic
selle, Experimental investigation on laminar burning velocities of ammo- modeling of ammonia oxidation with improved reaction mechanism for am-
nia/hydrogen/air mixtures at elevated temperatures, Fuel 263 (2020) 116653. monia/air and ammonia/hydrogen/air combustion, Int. J. Hydrogen Energy 43
[12] G.L. Dugger, D.D. Graab, Flame velocities of hydrocarbon-oxygen-nitrogen mix- (2018) 3004–3014.
tures, Symp. (Int.) Combust. 4 (1953) 302–310. [23] U. Mechanism, Chemical-Kinetic Mechanisms for Combustion applications, Me-
[13] G.L. Dugger, D.M. Simon, Prediction of flame velocities of hydrocarbon flames, chanical and Aerospace Engineering (Combustion Research), University of Cal-
Symp. (Int.) Combust. 4 (1953) 336–345. ifornia at San Diego, 2014 http://web.eng.ucsd.edu/mae/groups/combustion/
[14] V.A. Alekseev, A.A. Konnov, Data consistency of the burning velocity mea- mechanism.html.
surements using the heat flux method: hydrogen flames, Combust. Flame 194 [24] E.C. Okafor, Y. Naito, S. Colson, A. Ichikawa, T. Kudo, A. Hayakawa, H. Kobayashi,
(2018) 28–36. Experimental and numerical study of the laminar burning velocity of
[15] A.A. Konnov, A. Mohammad, V.R. Kishore, N.I. Kim, C. Prathap, S. Kumar, A CH4–NH3–air premixed flames, Combust Flame 187 (2018) 185–198.
comprehensive review of measurements and data analysis of laminar burning [25] H. Nakamura, M. Shindo, Effects of radiation heat loss on laminar premixed
velocities for various fuel+air mixtures, Progr. Energy Combust. Sci. 68 (2018) ammonia/air flames, Proc. Combust. Inst. 37 (2019) 1741–1748.
197–267. [26] H. Kobayashi, A. Hayakawa, K. Somarathne, E. Okafor, Science and technol-
[16] A. Goldmann, F. Dinkelacker, Approximation of laminar flame characteristics ogy of ammonia combustion, Proceedings of the Combustion Institute (2018),
on premixed ammonia/hydrogen/nitrogen/air mixtures at elevated tempera- doi:10.1016/j.proci.2018.09.029.
tures and pressures, Fuel 224 (2018) 366–378.

You might also like