Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Combustion and Flame 255 (2023) 112912

Contents lists available at ScienceDirect

Combustion and Flame


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

Effect of CO2 dilution on structures of premixed syngas/air flames in a


gas turbine model combustor
F. Pignatelli a, S. Derafshzan b, D. Sanned b, N. Papafilippou c, R.Z. Szasz a, M.A. Chishty d,
P. Petersson e, X.-S. Bai a, R. Gebart c, A. Ehn b, M. Richter b, D. Lörstad f, A.A. Subash b,∗
a
Division of Fluid Mechanics, Department of Energy Sciences, Lund University, Lund, Sweden
b
Division of Combustion Physics, Department Physics, Lund University, Lund, Sweden
c
Division of Energy Science, Department of Engineering Sciences and Mathematics, Luleå University of Technology, Luleå, Sweden
d
Research Institutes of Sweden (RISE), Piteå 941 38, Sweden
e
Dantec Dynamics A/S, Skovlunde, Denmark
f
Siemens Energy AB, Finspång, Sweden

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

Article history: The impact of CO2 dilution on combustion of syngas (a mixture of H2 , CO, and CH4 ) was investigated in
Received 7 November 2022 a lab-scale gas turbine model combustor at atmospheric pressure conditions. Two mild dilution levels of
Revised 15 June 2023
CO2 , corresponding to 15% and 34% of CO2 mole fraction in the syngas/CO2 mixtures, were experimentally
Accepted 15 June 2023
investigated to evaluate the effects of CO2 dilution on the flame structures and the emissions of CO and
NOx. All experiments were performed at a constant Reynolds number (Re = 10 0 0 0). High-speed flame
Keywords: luminescence, simultaneous planar laser-induced fluorescence (PLIF) measurements of the OH radicals
Swirl-Stabilized flames and particle image velocimetry (PIV) were employed for qualitative and quantitative assessment of the
Gas turbine model combustor resulting flame and flow structures. The main findings are: (a) the operability range of the syngas flames
Syngas combustion
is significantly affected by the CO2 dilution, with both the lean blowoff (LBO) limit and the flashback limit
CO2 Dilution
shifting towards fuel-richer conditions as the CO2 dilution increases; (b) syngas flames exhibit flame-
PIV
OH-PLIF pocket structures with chemical reactions taking place in isolated pockets surrounded by non-reacting
Flame pocket fuel/air mixture; (c) the inner recirculation zone tends to move closer to the burner axis at high CO2
Emissions dilution, and (d) the NOx emission becomes significantly lower with increasing CO2 dilution while the
CO emission exhibits the opposite trend. The flame-pocket structure is more significant with increased
CO2 dilution level. The low NOx emissions and high CO emissions are the results of the flame-pocket
structures.
© 2023 The Author(s). Published by Elsevier Inc. on behalf of The Combustion Institute.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)

1. Introduction Consequently, the different compositions of syngas give rise to


a significant change in flame characteristics, e.g., laminar flame
Synthetic gas, or syngas, represents a viable alternative en- speed [4–7], ignition delay time [8,9], and flame extinction strain
ergy source to natural gas in gas turbine applications thanks to rate [10]. Furthermore, varying syngas compositions affects the
its availability and combustion characteristics [1]. Syngas is a operating conditions of typical engineering applications, such as
combustible gas mixture mainly composed of hydrogen (H2 ) and gas turbines [11,12]. Therefore, there is a need to understand the
carbon monoxide (CO) that can be produced through gasification combustion characteristics and properties of syngas flames to
of various feedstocks, e.g., coal, biomass or solid waste [2,3]. How- ensure the safe operation of the existing burners [11,13].
ever, depending on the production process, the feedstocks, and the Modern gas turbine engines are usually operated with pre-
quality of the raw materials, syngas may also contain impurities mixed flames at fuel-lean conditions to mitigate NOx emissions
and a small amount of light hydrocarbons, e.g., methane (CH4 ). and to achieve high thermal efficiency [14–16]. Using high H2
Thus, the combustion characteristics of syngas are not unique. content fuel mixtures, such as syngas or hydrogen-enriched fossil
fuels, is a feasible way to reduce CO2 emissions. However, using
syngas represents a challenge due to the presence of H2 in the fuel

Corresponding author. mixture. The distinctive characteristics of hydrogen, e.g., high dif-
E-mail address: arman.subash@forbrf.lth.se (A.A. Subash). fusivity, high flame speed, and high flame temperature even at low

https://doi.org/10.1016/j.combustflame.2023.112912
0010-2180/© 2023 The Author(s). Published by Elsevier Inc. on behalf of The Combustion Institute. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/)
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

equivalence ratios, can significantly alter the range of operating weakening the IRZ. Furthermore, the resulting flame shape in the
conditions for gas turbine engines [17]. Previous studies showed combustor is mainly affected by the laminar flame speed, thus, by
that syngas flames are prone to flashback [11,18]. Indeed, several the syngas composition. A numerical study by Li et al. [29] on the
experimental works under gas turbine-like conditions reported effect of CO2 dilution on a syngas mixture flame in a partially pre-
smaller operability windows, shifted toward fuel-lean conditions, mixed gas turbine combustor revealed that the flame temperature
for different hydrogen-blended mixtures [2,12,19–21]. Thus, syngas is significantly reduced with the CO2 dilution and that the pres-
and, in general, high H2 -content fuels are often diluted with ence of CO2 modifies the radical pool reducing the concentrations
additive species, e.g., N2 , CO2 and Ar, to prevent flashback and of H and OH radicals in the reaction zones.
to mitigate high flame temperature and, consequently, further to In countries where there exists a considerable forest reserve
decrease the NOx emissions [12,22]. However, the thermal and and pulp and paper industry, the so-called Black Liquor (BL) is
chemical properties of diluents in the mixture may significantly available, which can be used in large-scale (10 0–50 0 MW thermal)
affect the flame stability, leading to blowout and flame extinction power plants [26,48], or can be gasified to syngas. The composition
[21,23]. Thus, a crucial task is to understand how flame stability of syngas from black liquor is characterized by the major species
limits vary with the fuel composition and additive dilution to compositions, H2 and CO2 , almost equal to around 1/3 of the entire
operate the device safely. composition each, CO concentrations around 28% and other minor
There exists extensive literature investigating experimentally species, e.g., CH4 , at around 1–2%. To increase the heating value of
and numerically the impact of CO2 dilution in syngas mixtures the mixture, methane is usually kept in the mixture while part of
in gas turbine-related applications [2,11,12,24–30]. The CO2 dilu- the carbon dioxide can be removed. This results in an unavoidable
tion is one of the most promising ways of reducing the NOx emis- variation of flame structures between the two extreme composi-
sions compared with other diluents, such as N2 and Ar, since it tions, with or without CO2 dilution in the syngas mixture. To our
can participate in chemical reactions [25,31–33]. The CO2 inhibits knowledge, there is a lack of study on the effect of CO2 dilution
the flame by reducing the flame temperature and the concentra- on the syngas flame structures in gas turbine combustors. In par-
tions of H, O, and CH radicals in the reaction zone [25,34–36]. ticular, optical experiments based on laser diagnostic methods are
Williams et al. [30] reported that the levels of NOx in a lab-scale lacking. Optical experiments that can characterize the flow struc-
swirling flame of CO2 -diluted syngas mixtures were very low (be- tures and the distribution of key radical species are needed to un-
low 2 ppm). The NOx emissions correlate well with the adiabatic derstand how the flame and the flow vary with CO2 dilution to
flame temperature. In fact, with a constant O2 content in the ox- safely operate the existing devices and to improve the fuel flexibil-
idizer, an increase of CO2 dilution decreases the adiabatic flame ity of a burner.
temperature of the flame, thus the NOx emissions. On an indus- Set against this background, we conducted experimental studies
trial gas turbine, Lee et al. [24] reported that the NOx reduction of lean premixed syngas flames in a laboratory-scale gas turbine
per unit of power varies logarithmically with the diluent’s heat ca- model combustor (GTMC) using laser-based diagnostic methods,
pacity. Thus, predicting the NOx emissions of an energy production including simultaneous OH-PLIF and PIV, to further understand the
plant is possible. Despite the benefit of low NOx emission in CO2 flame structure under stable conditions. The emission levels of the
diluted syngas flames, the participation of CO2 in the chemical re- most relevant species are also measured at the combustion cham-
actions results in the LBO limits shifting toward higher equivalence ber outlet to help the understanding of flame characteristics. Be-
ratios as compared to pure syngas mixtures [12]. Furthermore, as fore performing the optical experiments, the operability limits of
shown by Mo et al. [37] in a bluff-body stabilized syngas flame, the the GTMC for the syngas combustion are identified experimentally
CO2 dilution affects the flame stability more than the N2 dilution. for three CO2 dilution levels. Then, the flame structures are stud-
Swirlers are commonly used in gas turbine combustors to en- ied under stable operating conditions. The results are analyzed to
hance the fuel/air mixing process as well as to stabilize the flame understand the impact of syngas composition both on the flame
in the combustor due to the formation of an inner and an outer structures and emissions. The paper is organized as follows: first,
recirculation zone (IRZ and ORZ, respectively) [38–42]. Although the experimental facilities and the employed experimental tech-
the variety of syngas compositions has been the objective of sev- niques are described, then the characterization of the fuel mix-
eral studies aiming at the characterization of the fundamental ture composition as well as the operability limits are discussed in
flame characteristics, e.g., laminar flame speed [5,43], lean blow- Section 3.1 while the flame and flow structures and the experi-
out (LBO) limits, as well as the level of emissions under gas mental emission data are provided in Sections 3.2 and 3.3, respec-
turbine-like conditions [11,12,27,30,44,45], only a few studies have tively. Conclusions are summarized in Section 4.
been focused on characterizing the flame and on investigating the
turbulence-combustion interaction in practical devices. Park et al. 2. Experimental setup and laser diagnostics
[46] showed the impact of the syngas composition on the flame
shape in a down-scaled diffusion-flame type gas turbine combus- 2.1. Burner facility
tor. They reported that at constant load, the higher reactivity of
hydrogen makes the flame shorter and broader. An increase of 3% The burner used during the experimental campaign is the latest
in hydrogen mole fraction decreased the flame length by 3.2% and version of the CECOST burner [20,49–51]. Inspired by the EA/AEV
increased the flame angle by 4.2%. PIV measurements on a modular burner [52], the burner was developed to feature the complex
laboratory scale gas turbine combustor by Dam et al. [47] revealed combustion processes in typical industrial gas turbine engines, i.e.,
that a H2 -CO syngas mixture presents a wider recirculation zone thermo-acoustic instabilities, lean blow-out and flashback [53,54].
compared to methane fuel, which leads to a significant increase An isometric view and a cross-section of the burner are shown
of the pressure gradient inside the combustor. Furthermore, the in Fig. 1. The burner includes a premixed fuel/air swirl genera-
propensity of the flame to flashback increases with the hydrogen tor consisting of four quarter-cones. The cones are approximately
content in the mixture. The numerical analysis of different CO2 - 2 mm thick, and the half-angle of the cones is around 25 de-
diluted syngas compositions carried out by Papafilippou [26] on grees. The swirler is mounted to and held by a lance and as-
the TECFLAM swirl burner showed qualitatively that the magnitude sembled inside a metallic mixer pipe of 220 mm in length and
of the reverse flow in the combustor is significantly higher for the 54 mm in diameter, where the mixing of bulk airflow and fuel
syngas as compared to the methane flames. Syngas mixtures show takes place. A honeycomb is mounted inside the mixer-pipe up-
increased axial momentum and tend to reduce the swirl intensity, stream of the swirler to break up the large eddy structures of the

2
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 1. Isometric view and vertical cross-section of the CECOST burner, along with details of the swirler (a.1), the swirler blades (a.2), and the fuel injector (b).

flow. The fuel and air mixing is further enhanced in a premix- 2.2. Experimental techniques
ing tube (length of 100 mm and diameter of 50 mm) mounted
on the top of the swirler. The combustion chamber has a 140 × Fig. 2 illustrates the experimental apparatus employed in this
140 mm cross-section and 400 mm length and is mounted on a investigation. The setup consists of a system for PLIF imaging of
metal base plate on the top of the premixing tube. A contrac- OH radicals, a PIV system for velocity measurements, an emission
tion plate with an 80 mm diameter hole at the center is mounted measurement system, and a high-speed camera system for imaging
on the top of the combustor. The combustor and premixing tube of the flame luminescence in both the combustion chamber and
are made of quartz glass, providing optical access for laser-based the premixing tube to investigate eventual flashback events. The
experiments. optical and probing diagnostic techniques are described below in
Based on previous studies on the same burner [19,55], the swirl detail.
number, S, has been estimated to be approximately 0.6. S is defined
as follows [56]: 2.2.1. Optical diagnostics
A sketch of the optical setup employed in the experiments is
R shown in Fig. 2. The flame structure was visualized using PLIF
0 ŪyŪt r 2 dr
S= R 2 (1) imaging of OH radicals, indicating the reaction and post-flame
R 0 Ūy rdr zones. The OH-PLIF was performed using an Nd:YAG laser at
532 nm pumping a tunable dye-laser using Rhodamine 590 dye.
where Ūy , Ūt and R denote the mean axial velocity, the mean tan- After frequency doubling the dye-laser output, the wavelength was
gential velocity, and the radius of the premixing tube (25 mm). tuned to 283.321 nm, corresponding to the Q1(8) transition in
The bulk airflow was provided at room temperature by a blower OH with 16 mJ/pulse. The PIV system consists of a high-speed
(Rietschle SAP 300, maximum airflow mass rate of 55 g/s) through diode-dumped, dual cavity Nd:YLF laser (Litron LDY304-PIV) oper-
the bottom inlet of the mixer pipe as indicated in Fig. 1. The fuel ated at a wavelength of 527 nm. The two laser beams were spa-
was injected radially outwards through a spiral-pipe injection sys- tially overlapped using a beam splitter and expanded by sheet-
tem at the bottom of the mixer pipe, placed slightly above the bulk forming optics consisting of a negative cylindrical lens ( f = −75
airflow inlet. The two ends of the spiral pipe are the fuel inlets. The mm) and a positive spherical one ( f = +500 mm). The resulting
fuel entered the mixer pipe through 16 holes evenly distributed Gaussian sheet was guided toward the flame centerline (approxi-
along the surface of the spiral pipe (see Fig. 1b). The diameter of mately 120 mm high).
each hole is 1 mm, and the direction of injection is perpendicu- Imaging of the OH-PLIF was performed using an ICMOS camera
lar to the airflow stream. The fuel flow rates were regulated sep- (Andor iStar SCMOS) with a native resolution of 2560 × 2160 pix-
arately using two different mass flow controllers (Alicat Scientific els with an individual pixel size of 20 μm and 16-bit depth. The
MC-100SLPM, with accuracy ±0.8% reading and ±0.2% at full load). resolution was reduced using on-chip binning to 1280 × 1080 pix-
To quantify the fuel/air mixing at the entrance plane of the els allowing for higher signal-to-noise ratio (SNR). The PLIF images
combustion chamber, PLIF measurements of a tracer species, ace- were collected at a frequency of 20 Hz. A UV-Nikkor lens was used
tone, were carried out in a horizontal plane, 5 mm downstream with a focal length of 105 mm and an f-number set to 4.5. A 320 ±
of the exit plane of the premixing tube [49]. The PLIF signal in- 20 nm bandpass filter was fixed to eliminate interference signals. A
dicated that the spiral pipe injection system resulted in a nearly high-speed camera (Photron SA-Z) and a dual frame camera (Phan-
homogeneous fuel/air mixture at the entrance plane of the com- tom v311) were used for flame luminescence and PIV imaging. The
bustion chamber. two cameras were aligned orthogonal to each other. The PIV and

3
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 2. 3D schematic illustration of the experimental setup.

flame luminosity signals were separated with a dichroic mirror quisition time of 120 seconds) were collected for each point of in-
centered at 530 nm (±10) placed 270 mm from the combustor that terest to ensure relevant statistical data.
reflected the scattered light from PIV seeding particles at 527 nm
onto the PIV camera while passing the flame luminescence to the
high-speed camera, allowing for simultaneous imaging of the same 3. Results and discussion
measurement region. The flame luminescence camera had a native
resolution of 1280 × 800 pixels with an individual pixel size of 3.1. Mixture properties and burner stability limits
20 μm and 12-bit depth. A Nikon objective with a focal length of
105mm with an f-number set to 2.8 was used. Flame luminescence Table 1 lists the composition based on the volume fraction of
images were collected at a frame rate of 20 kHz. For the PIV cam- the CO2 -diluted syngas mixtures investigated in this work. The
era, the native resolution was 1280 × 1080 pixels with an individ- baseline mixture (SYN0 in Table 1) consists of 0% CO2 , 3% CH4 , 44%
ual pixel size of 20 μm and 12-bit depth. A 70 mm Nikon objective CO and 53% H2 and it is referred to as the “pure syngas”. In the
was mounted on the PIV camera, combined with a 12 mm exten- mixtures denoted SYN15 and SYN34, the pure syngas (SYN0) was
sion ring. An f-number of 5.2 was used, and the frame rate was diluted with 15% and 34% CO2 , respectively. The flames were op-
1 kHz. erated at a constant Reynolds number (Re) of 10 0 0 0. The Re was
The PIV and the OH-PLIF systems were controlled by an ex- determined based on the diameter of the premixing tube, the bulk
ternal trigger, making the acquisition of simultaneous flow/flame axial flow speed of the mixture at the exit of the premixing tube,
structure information possible. Each OH-PLIF pulse was centered in and the kinematic viscosity of air at room temperature. Due to the
between two consecutive PIV laser pulses every 50 couples of PIV different composition, the fuel/air mole ratio at stoichiometric con-
laser beams. For each investigated flame condition, 30 0 0 single- ditions (FAmole,s ) increases with CO2 dilution.
shot PIV images were collected. Thus, 60 simultaneous PIV/OH- The effect of CO2 dilution on the laminar flame speed, radi-
PLIF pictures were acquired. Furthermore, to ensure an appropriate cal concentrations in the reaction zone, and adiabatic flame tem-
amount of data for statistical analysis of the flame structure, an ex- perature is analyzed first. Numerical simulations of freely propa-
tra set of 500 single-shot OH-PLIF images were collected for each gating planar laminar premixed flames under adiabatic conditions
selected case. were carried out using the Cantera code [57] and the widely used
GRI-3.0 [58] chemical kinetic mechanism. Fig. 3 shows the lami-
2.2.2. Emission analyzer nar flame speed (SL ), the adiabatic flame temperature (Tad ), and
The exhaust gases have been analyzed using a Fuji Electric 3 the maximum mole fractions of H, O, and OH radicals across the
component ZKJ system for CO concentrations while NOx concen- flames for different equivalence ratios . The laminar flame speed
trations have been measured with an Alfakomp Eco Physics CLD increases with the equivalence ratio for the investigated lean and
822 M hr system. Gases discharged from the combustor have been slightly rich flames ( < 1.1), while it decreases monotonically
sampled at a sampling rate of 1 Hz with a gas sample probe sys- with increasing CO2 dilution. A qualitatively similar trend is shown
tem (M&C PSP 40 0 0) consisting of a probing tube characterized in the adiabatic flame temperature plot. However, the adiabatic
by an inner diameter of 4 mm placed 30 mm above the circu- flame temperatures of the CO2 -diluted flames peak at  ∼ 1.05,
lar contraction section of the combustor (see Figs. 1 and 2). The
dry species of interest are then directed to an emission analyzer
Table 1
rack, where their concentrations are measured. The system’s accu- Compositions and stoichiometric fuel/air mole ratios (FAmole,s ) of the investigated
racy was 1 ppm on uncorrected NOx and CO values in ppm. All the syngas mixtures.
values extracted from the data logger of the analyzer have been
Mixtures CH4 [%] CO [%] H 2 [%] CO2 [%] FAmole,s
later corrected to the reference condition of 15% O2 . The emission
trends have been monitored during the experiments till their sta- SYN0 3 44 53 0 0.3855
SYN15 2.55 37.4 45.05 15 0.4233
bility to minimize the measurement errors. Once the species mea- SYN34 2 29 35 34 0.5836
surements were stable, 120 data samples (corresponding to an ac-

4
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 3. Numerically predicted (a) laminar flame speed (SL ), (b) adiabatic flame temperature (Tad ), (c) maximum mole fraction of H (XH ) and O (XO , in logarithmic scale)
and (d) maximum mole fraction of OH (XOH ) as functions of the equivalence ratio , at atmospheric pressure and room temperature (1 bar and 298K) for the three syngas
mixtures listed in Table 1.

and they decrease slightly with  from 1.05 to 1, due to the in-
complete combustion of fuel in the fuel-rich mixtures.
The dilution with CO2 suppresses the chemical reactivity of the
mixture due to the reduced heat of combustion of the syngas mix-
ture, which leads to a reduced adiabatic flame temperature and
lower concentrations of radicals. Subsequently, the laminar flame
speed is also decreased. Since the syngas mixtures contain high
amounts of H2 and CO, the laminar flame speed peaks at fuel-rich
mixtures, and it increases with the H2 mole fraction of the mixture
[6,7].
The impact of CO2 dilution of the flames may be classified as
thermal and chemical effects. The thermal effect of CO2 dilution
is the reduction of flame temperature, whereas the chemical ef-
fect is the reduction of the concentration of radical species (H, O,
and OH). These two effects may be distinguished by comparing the
mole fraction of OH radicals at a constant adiabatic flame temper- Fig. 4. Operability limits for the three different CO2 dilutions.
ature of the three syngas flames, e.g., Tad = 1800 K, Fig. 3. At the
constant adiabatic flame temperature of 1800 K, the thermal effect
is compensated by the increased equivalence ratio at high CO2 di- the F B was defined as the one at which the flame was anchored
lution conditions. The peak OH mole fractions for the three syngas to the swirler.
flames are 3.6×10−3 , 3.2×10−3 , and 2.7×10−3 , for the SYN0, SYN15, Generally, the stability range shifts toward richer mixtures
SYN34 flames, respectively. The corresponding laminar flame speed when increasing the CO2 dilution level, cf. also Table 2. Similar
decreases, along with the decrease of peak OH mole fractions, trends have been reported in the literature [2,12]. Furthermore, the
from 35 cm/s to 31 cm/s as the CO2 dilution increases from 0 to LBO and flashback limits vary non-linearly with the level of CO2
34%. dilution, which is attributed to both the thermal and the chemical
The effect of CO2 dilution on the burner operability limits is effects of CO2 dilution on the overall flame behaviour. The thermal
shown in Fig. 4. The shadow region indicates the stable flame effect can be identified from the adiabatic flame temperature at
regime, which is bounded by the flashback and LBO limits. The ex- LBO and flashback, cf. Table 2. From Fig. 3(b), it is apparent that
periments for each syngas mixture were repeated three times to the adiabatic flame temperature and maximum mole fraction of
ensure the repeatability of the results. The LBO limit is the lowest OH radicals at LBO for the SYN0 and SYN15 mixtures are quite
equivalence ratio below which the flame is blown off the combus- close, suggesting that the thermal effect controls the LBO process
tor. The flashback limit (F B ) corresponds to the condition when for the low CO2 dilution mixtures. However, when the CO2 dilution
the flame propagates upstream in the premixing tube. Specifically, is further increased to 34%, the adiabatic flame temperature and

5
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Table 2
Experimentally measured equivalence ratios at lean blow-off and flashback limits, and numerically predicted laminar flame speed
[cm/s], adiabatic flame temperature [K], and maximum mole fraction of OH at lean blow-off limit and flashback limit for different
syngas mixtures.

Mixtures LBO SL,LBO Tad,LBO XOH,LBO F B SL,F B Tad,F B XOH,F B

SYN0 0.38 8.08 1402 0.60∗ 10−3 0.54 30.51 1751 3.01∗ 10−3
SYN15 0.40 8.22 1410 0.56∗ 10−3 0.56 26.84 1717 2.61∗ 10−3
SYN34 0.48 11.25 1484 0.85∗ 10−3 0.70 30.87 1804 2.72∗ 10−3

the maximum mole fraction of OH radicals at LBO are significantly cooling of the combustor wall. Previous studies have shown that
higher than in the low CO2 dilution mixtures. This indicates that methane/air flames and hydrogen-enriched methane/air flames ex-
the chemical effect is also essential in the high CO2 -diluted flames hibited M-shaped flames in a similar burner [19,51]. However, the
since under the same adiabatic flame temperature or OH radical present syngas flames and CO2 -diluted syngas flames typically ex-
concentration, the high CO2 -diluted flames are more prone to ex- hibit a V-shape. As will be shown later, the flow structure, i.e., the
tinction. Similar conclusions can be drawn for the flashback limits. shape of the inner recirculation zone (IRZ), is essentially similar in
It appears that the thermal effect of CO2 dilution on the flashback all syngas flames, which is the reason for the similar flame shape
is dominant in the low CO2 dilution flames, whereas the chemical of all the cases shown in Fig. 5.
effect is becoming important in the high CO2 dilution flames. For Despite the similar flame shape, the luminosity of the flames
all flame cases, a flashback occurs when the laminar flame speed are significantly different in the flames shown in Fig. 5. The SYN0
is about 30 cm/s, Table 2. flame at  = 0.53 is close to flashback and the luminosity of this
The chemical effect of CO2 dilution may be understood from flame is the strongest. The SYN34 flame at  = 0.47 is close to
the direct participation of CO2 in the chemical reactions. Several LBO, and has the lowest luminosity. It is clear that at similar equiv-
authors [59,60] have shown that CO2 affects the water-gas shift alence ratios, the increase in the CO2 dilution leads to a decrease
reaction, CO + OH = CO2 + H, affecting the radical pool. Accord- in flame luminosity, which is consistent with the decreasing reac-
ing to Park et al. [60], increasing CO2 levels decrease the concen- tivity and concentration of H, O and OH radicals, as discussed in
trations of OH, O, and H radicals, leading to an overall reduction the previous section.
of mixture reactivity. Furthermore, high concentrations of CO2 en- Fig. 6 shows instantaneous flame luminescence images for the
hance the recombination reaction of H + O2 + M → HO2 + M, thus SYN0 and SYN34 flames for an equivalence ratio of  = 0.53. Data
reducing the concentration of H radical levels in the flame and, were acquired with a frequency of 1 kHz and an exposure time of
consequently, affecting the branching reaction H + O2 ↔ O + OH 50 μs. The white dashed line indicates the exit plane of the pre-
[1,11,12,61]. This is consistent with the numerical results shown in mixing tube. Consistently with the mean flame luminosity fields
Fig. 3(c and d). To compensate for the loss of radicals in the high discussed earlier, the intensity of flame luminosity in the SYN0
CO2 dilution flames, the values of LBO and F B must therefore be flames is significantly higher than in the SYN34 flames. The in-
significantly increased. stantaneous flame luminescence images also show fine structures,
A third effect of CO2 dilution on the flashback and LBO limits particularly in the SYN34 flames. The SYN0 flames are close to the
can be pointed out. The combustion characteristics of syngas mix- flashback, and the leading front of the flame is already entering
tures are highly dependent on the H2 concentration in the mix- the premixing tube. The SYN34 flames are stabilized downstream
tures. Thus, it is expected that the LBO process is sensitive to of the exit plane of the premixing tube, and the leading front po-
the concentration of H2 , since hydrogen, due to its high molecu- sition appears to oscillate in the axial direction. The oscillation of
lar diffusivity and high flame speed [62], is the last species able the leading flame front in the SYN34 flames could be due to the
to burn during the extinction process as compared to the rest of precessing vortex core (PVC) in the IRZ [64] since the flame is sta-
the species of the mixtures. As shown below, at high CO2 dilution bilized in the IRZ. The oscillation may also be due to the local
levels, the flame structures significantly differ from that of low CO2 flame extinction and re-ignition. The onset of local extinction and
dilution flames. The high CO2 dilution flames exhibit significant lo- re-ignition could be flame/turbulence interaction or the poor mix-
cal extinction in the reaction zones, which certainly affect the LBO ing of the fuel and air in the premixing tube [65]. In the present
and flashback limits. The effect of CO2 dilution on local extinction flames, the high-level CO2 dilution suppresses the reactivity of the
will be discussed in more detail in the following. fuel/air mixture, making the flame vulnerable to local extinction
under intensive turbulence straining.
3.2. Effect of CO2 dilution on the flow and flame structures Fig. 7 shows the temporal evolution of the flame areas com-
puted based on the flame luminescence for the three different
The syngas flame structures were visualized using high-speed mixtures at the equivalence ratio of 0.53. The area has been com-
flame luminescence imaging at different equivalence ratios at sta- puted by binarizing every 500 single-shot images, i.e., setting the
ble conditions. After background luminosity subtraction, the mean value of 1 for regions where the signal was higher than a thresh-
luminosity was computed from 500 instantaneous images. Fig. 5, old (set to 10% of the peak signal intensity of the SYN0 flame) and
showing mean flame images, indicates a V-shape luminous re- 0 where it was lower. Next, the resulting time-dependent signals
gion for all three syngas mixtures. In a V-shaped flame, the lead- have been normalized by the highest value of the flame area of
ing front of the luminous region is on the centerline of the the SYN0 flame. It is visible that the luminous flame area fluc-
burner around the exit of the premixing tube, whereas the flame tuates continuously, indicating flame expansion and recession. For
fronts near the combustor liner point toward the combustor out- the SYN0 flame, the flame area fluctuates around 75% of its max-
let [19,63]. In contrast, an M-shaped flame is referred to the flame imum area, whereas the SYN15 flame area fluctuates around 50%
whose fronts near the combustor liner point toward the outer re- of the maximum flame area of the SYN0 flame. The SYN34 flame
circulation zone (ORZ) at the corner of the combustor [19]. shows the smallest flame area, which fluctuates around 20% of the
In M-shaped flames, the hot gas and radicals in the ORZ ig- maximum flame area of the SYN0 flame. From Fig. 7, one can see
nite the reactants and anchor the flame in the ORZ, whereas in V- that the flame area of the SYN34 flame exhibits a low-frequency
shaped flames, the reactions in the ORZ are quenched due to the peak at approximately 10 Hz (occurring once every 100 ms). The

6
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 5. Averaged flame luminosity for the three investigated mixtures. The luminosity of the SYN34 flames has been multiplied by a factor of 2. D is the diameter of the
premixing tube. r and y indicate radial and axial directions, respectively.

Fig. 6. Instantaneous flame luminescence sequence for SYN0 (top) and SYN34 (bottom) mixtures under the condition of  = 0.53. The luminosity of the SYN34 flames has
been amplified by a factor of 2.

7
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 7. Flame luminescence signal for the three different mixtures at  = 0.53 (left) and SYN34 flame luminescence single-shot at t = 352 ms (right). Intensities are
normalized by the highest signal value.

Fig. 8. Adiabatic flame temperature (Tad ) as a function of laminar flame speed SL based on the numerical results shown in Fig. 3. The arrows indicate the PIV and OH PLIF
experimental points.

luminous region of the SYN34 flame is downstream of the pre- Table 3


Properties of the investigated mixtures at the selected equivalence ratios, . SL ,
mixing tube and in the IRZ, cf. Figs. 6 and 7. Thus, the flame area
m˙ Syn , m˙ CO2 and Q refer to the laminar flame speed [cm/s], mass flow rates of syngas
is affected by the PVC. Such low-frequency periodic flame fluctua- and CO2 [g/s], and the thermal load [kW].
tions were not observed in the SYN0 and SYN15 flames. The lead-
ing fronts of SYN0 and SYN15 flames are in the premixing tube Mixtures  SL Tad m˙ Syn m˙ CO2 Q

upstream of the IRZ, cf. Figs. 5 and 6. Subsequently, the luminous SYN0 0.47 19.2 1598 1.2 0 31.5
areas of these flames are unaffected by the PVC and do not show SYN15 0.50 19.2 1608 0.84 0.47 22.06
SYN34 0.56 18.1 1609 0.57 0.92 14.96
a low-frequency flame area oscillation.
PIV and OH PLIF experiments were conducted for the three
syngas mixtures under stable flame conditions with similar adia-
batic flame temperatures (Tad ) and laminar flame speeds (SL ). Fig. 8 investigated cases including the corresponding mass flow rates of
shows the variation of Tad as a function of SL , which is a re-plot syngas (m˙ Syn ) and CO2 (m˙ CO2 ), as well as the values of the corre-
of the numerical results presented in Fig. 3a and b. At the se- sponding thermal input (Q) is provided in Table 3.
lected flame conditions, the adiabatic flame temperature is about Fig. 9 shows snapshots of OH-PLIF images at four different
1600 K, with a maximum difference of 9.61 K between the SYN0 instants of time for the three syngas flames. The t = 0 indicates an
and SYN34 mixtures. The laminar flame speed for the selected ex- arbitrary time during the stationary operation of the combustor.
perimental conditions varies between 18.1 cm/s and 19.2 cm/s. The The V-shaped flame structures are evident in the OH-PLIF images,
equivalence ratios are 0.47, 0.50, and 0.56 for the SYN0, SYN15, and similar to the chemiluminescence images discussed earlier. The
SYN34 mixtures, respectively. A summary of the properties of the flame without CO2 dilution (SYN0) exhibits fairly continuous OH

8
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 9. Instantaneous OH-PLIF snapshots for the SYN0 flames at  = 0.47 (top), the SYN15 flames at  = 0.50 (middle) and the SYN34 flames at  = 0.56 (bottom).

distribution, while, with increasing the CO2 in the mixture, this homogeneity can be due to insufficient fuel/air mixing in the
OH distribution becomes more discrete and is found in isolated mixer pipe and the premixing tube. Acetone PLIF measurements
pocket regions. These pocket structures show discontinuation of of the present burner showed a homogeneous fuel/air mixture at
reaction zones, indicating local quenching of the reactions in the the exit of the premixing tube [49]. Thus, the impact of fuel/air
flames with high CO2 dilution. mixture in-homogeneity on the in-homogeneous OH field in the
In the SYN0 flames, although the local extinction is not signif- three flames shown in Fig. 9 can be excluded. Furthermore, it
icant, the OH signal does show high-intensity islands connected is expected that pressure oscillation due to combustion instabil-
to low-intensity regions. This phenomenon is attributed to the ity, e.g., thermoacoustic instability, can lead to airflow rate oscilla-
diffusional-thermal flame instability often observed in lean hy- tion. The fuel flow rate may also oscillate when injected through
drogen premixed flames [66,67]. Since the diffusion coefficient of the 16 small holes in the spiral-pipe fuel injection system (see
hydrogen is significantly higher than other major species in the Section 2.1). The fuel nozzle velocity for the investigated cases is
flame, under lean premixed conditions, hydrogen molecules tend 84 m/s, 70 m/s, and 60 m/s, respectively, for the SYN0, SYN15, and
to diffuse to form rich islands where the reactivity is enhanced SYN34 cases. The flow is far from choking; thus, the fuel flow rate
substantially. Due to the differential diffusion, a high-intensity OH oscillation may be significant if there is a strong pressure oscil-
region is formed in the burning flame pockets, leaving weakly lation in the mixing pipe. This can give rise to an oscillation of
burning regions in between the high-intensity OH region [68]. the fuel/air ratio and possibly lead to an in-homogeneous OH field.
With increasing levels of CO2 dilution, the reactivity of the initially No significant pressure oscillation could be observed in the cur-
weak OH region is further suppressed, and local extinction occurs rent stable flame operation, including the three flames shown in
in a broad region, leaving the burning pockets in small isolated re- Fig. 9. Thus, it can be concluded that the onset of in-homogeneous
gions. OH field and isolated OH pockets is not due to fuel/air mix-
Although the numerically simulated laminar flame speed and ture in-homogeneity resulting from insufficient mixing or pressure
adiabatic flame temperature are similar in the SYN0, SYN15, and oscillation.
SYN34 flames listed in Table 3, the actual flame temperature is ex- An estimation of the Karlovitz (Ka) number is performed based
pected to be lower due to the heat loss to the combustor walls. on the definition of Peters [69], and the PIV data collected for
The SYN34 flame has a lower thermal input than the SYN0 and this work indicates Ka is 0.489, 1.06, and 1.16 for the SYN0 flame,
SYN15 flames, which may lead to a lower flame temperature in the SYN15 flame, and SYN34 flame, respectively. Thus, the flames here
SYN34 flame when considering the heat loss to the walls. This can investigated belong to the flamelet regime [69]. However, the flame
also contribute to the reduced reactivity and isolated OH pockets pocket structure shown in the OH distribution in Fig. 9 is signif-
in the SYN34 flame. icantly different from the structure of flamelet combustion, e.g.,
In-homogeneity of the fuel/air mixture could contribute to the premixed methane/air flames in a Bunsen burner [70] or over a
in-homogeneous OH field in these flames. Reactant mixture in- swirl burner [71]. In methane/air flames, the Lewis number is close

9
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 10. Instantaneous flow field (colored with the velocity magnitude) superimposed to corresponding instantaneous OH-PLIF snapshot (top row), and OH-PLIF maximum
gradient field overlapped to the corresponding strain field (bottom row) for the SYN0 flame at  = 0.47, SYN15 flame at  = 0.50 and SYN34 flame at  = 0.56, respectively.

to unity. The diffusional-thermal instability is insignificant; thus, of hydrogen in the mixture. Besides, the PIV and OH-PLIF fields
the flame pocket structure is seldom seen. do not show a clear correlation between the OH pockets and lo-
Fig. 10 shows snapshots of simultaneous PIV and OH-PLIF fields cal flow eddies. Thus, it can be concluded that the onset of flame
for the three syngas flames (top row) and the gradients of the OH- pockets is due to diffusional-thermal flame instability [66,67] am-
PLIF signal distributions superimposed to the instantaneous strain plified by the suppression of reactivity of CO2 dilution.
rate distribution (bottom row). The most substantial gradients of Fig. 11 shows the OH-PLIF PMs (top row) and mean velocity
the OH-PLIF signal can be used as markers to approximate the po- fields (bottom row) for all the investigated cases. Similar to the
sition of the flame front [72,73]. edge extraction methodology used in Fig. 10, the PMs have been
The strain rate has been determined from the 2-D PIV data us- obtained by binarizing the single-shot OH-PLIF images using the
ing the following: same threshold. Then, the binarized fields were ensemble averaged
 2  2  2 to obtain the OH-PLIF probability maps (PMs) [74,75], which gives
∂u 1 ∂ u ∂v ∂v a maximum probability (℘) of 1 (corresponding to a 100% react-
S= + + + (2) ing region) and minimum probability of 0 (corresponding to non-
∂r 2 ∂y ∂r ∂y
reacting regions).
where u and v denote the velocity components along the radial (r) The probability maps are helpful in understanding the mean
and axial (y) direction, respectively. flame structures since they do not suffer from the pulse-to-pulse
The PIV field clearly shows the IRZ around the combustor’s energy fluctuations and the spatial decay of the laser beams. The
centerline and the outer recirculation zones (ORZ) in the lower- PMs indicate the chemical reactions in the flame to reduce with an
left and lower-right corners of the combustor. The fuel/air mix- increase in CO2 dilution, as also observed in the snap-shot images
tures flow into the combustor between the two recirculation zones. in Figs. 9 and 10. The maximum probability is 0.658 in the SYN0
The reaction zone represented by the OH PLIF signals is shown flame, 0.40 in the SYN15 flame, and 0.178 in the SYN34 flame. The
in the shear layer between the ORZ and the IRZ. It is shown SYN34 flame, having the lowest probability, corresponds to the iso-
from Fig. 10 that the strain rate field correlates neither with the lated OH pockets in these flames (with local extinction of the reac-
quenched region nor with the reaction region of the flame. Local tions). This feature is very different from flamelet combustion (i.e.,
extinction is, therefore, not due to the high strain rate; this be- with a spatially continuous reaction zone) of methane/air flames at
haviour has also been reported in [1] where local extinction of fu- similar low Karlovitz numbers and in flamelet models that do not
els with a high H2 content is attributed to differential diffusion consider differential diffusion [76].

10
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 11. Probability of having OH-PLIF signal in a spatial location in the combustion chamber (top) and mean flow field (bottom). Contours of ℘ = 0.2 (in magenta) and
0.5 (in black) are superimposed to the PMs of the OH-PLIF field. The inner recirculation zone (in white) is overlapped on the mean flow field. The mean velocity U  was
normalized with the bulk velocity Ubulk .

Fig. 12. The width of the inner recirculation zone (IRZ /D) along the centerline of the combustor (y/D) for the different syngas flames and an isothermal cold flow.

The mean flow field was obtained by averaging 30 0 0 instanta- the premixing tube. Cold flow results from our previous investiga-
neous velocity fields (corresponding to 3 ms) for each flame. To ob- tion [50] were added as references. It is visible that the thermal
tain the instantaneous velocity field from a PIV image, the single- expansion of reactive gases affects the different velocity compo-
shot PIV image has been post-processed by applying a multi-pass nents expanding the inner recirculation zone.
cross-correlation algorithm, selecting an initial interrogation win- The mean flow field in the SYN15 flame is somewhat similar to
dow of 64 × 64 pixels and progressively reducing to 16 × 16 pix- that of the SYN0 flame, whereas the mean flow structure of the
els, with an overlap of 50% for both. The impact of CO2 dilution on SYN34 flame is significantly different. The mean IRZ of the SYN34
the mean flow field is visible in Figs. 11 and 12. Fig. 12 shows the flame is narrower than those in the lower CO2 dilution flames, and
width of the IRZ as a function of the distance to the exit plane of the reverse flow axial velocity peaks at y/D ∼ 1.25, which is much

11
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

Fig. 13. Emissions of NOx with errorbars (left) and CO (right) on dry basis for the investigated syngas compositions, from measurements (a,b) and 1D planar laminar flame
simulations (c,d).

closer to the combustor inlet plane than the lower CO2 dilution operating conditions are close to zero and well below the experi-
flames that peak at y/D ∼ 2.5. Since the three flames have theoret- mental uncertainty of 1 ppm (see Section 2.2). Thus, the accuracy
ically similar laminar flame speeds, adiabatic flame temperatures, and precision of the measurements may not be significantly re-
and Reynolds numbers, the mean flow field is also expected to be liable. However, statistical analysis on 120 data samples for each
similar. The different mean flow structure of the 34% CO2 dilution case indicates that the NOx emission for all investigated syngas
flame is clearly due to the flame-pocket structure that leads to a mixtures tends to increase with the increasing equivalence ratio.
much lower mean flame probability, thus, a lower mean heat re- A similar low level of NOx emission for syngas flames has been
lease rate in the reaction zone. This lower mean heat release rate observed in the experiments of Samiran et al. [11], which is at-
leads to the flow field closer to the isothermal (cold) flow, with a tributed to the low flame temperature in lean syngas flames. The
narrower IRZ in the radial direction. increase of NOx with  is due to the increasing adiabatic flame
The mean flow results show that in the studied flame condi- temperatures. In contrast, the CO emission tends to decrease with
tions, the upstream stagnation point is in the premixing tube, thus, the increase of the equivalence ratio. The measured NOx emissions
is not visible in Fig. 11. As a result, the leading flame front of the can be compared with the numerical predictions shown in Fig. 13c,
SYN0 flame is inside the premixing tube; see also the flame lumi- obtained using the GRI 3.0 mechanism discussed earlier. The trend
nosity data shown in Figs. 5 and 6. The leading flame front is ex- of NOx in the experiments and the simulations is similar; however,
pected to be slightly upstream of the stagnation point of the IRZ, the measured NOx emissions are lower than the numerical predic-
where the flow velocity is relatively low. The SYN34 flame is fre- tions. This discrepancy may be due to the flame-pocket structure
quently pushed downstream to the main combustion chamber, as of these syngas flames, giving rise to a much lower mean flame
shown in Figs. 5 and 6, which is likely because the leading front temperature in the combustor than in the simulated planar adia-
of the flame is subject to a high flame stretch rate, and as such batic laminar flame.
the reactions at the leading fronts are frequently quenched due to Another reason for the lower measured NOx is the heat loss
high flame stretch rate and the reduced reactivity in the high CO2 to the walls in the experiment, which gives rise to a lower flame
dilution condition. temperature than the adiabatic flame temperature predicted in the
numerical simulations. As discussed earlier, the thermal input de-
3.3. Effect of CO2 dilution on post-flame emissions creases with the increase in CO2 dilution. Since the cold wall area
is the same in these flames, a flame with lower thermal input will
Fig. 13 a and b show the effect of CO2 dilution on the emis- generally have a lower temperature. This trend is consistent with
sions of NOx and CO in the exhaust gases. The results were en- the measured NOx in the three flames.
semble averaged using 120 data samples and the error bars in the The level of measured CO emissions for all three syngas flames
figures show the standard deviation of the data samples. The de- is significantly higher than their corresponding laminar flame nu-
tected dry NOx mole fractions were normalized to the 15% O2 con- merical results, and the trend of decreasing CO emission with in-
dition. The data were collected for flames in the whole stable op- creasing equivalence ratio is also opposite to that in the numer-
eration regime. As shown Fig. 13a, the NOx emissions for all the ical simulations, cf. Fig. 13(b and d). The measured CO emission

12
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

trend with increasing  is also opposite to the trend observed for in the same burner, and it is attributed to the flame-pocket
the premixed methane/air flames in the same combustor, cf. our structure in syngas flames.
previous work [20]. The high CO emission in the measurements
is likely attributed to the flame-pocket structure of the present
Declaration of Competing Interest
syngas flames. The fuel is consumed in the flame pockets, but
the unburned fuel/air mixture outside the flame pockets is leaked
The authors declare that they have no known competing finan-
into the exhaust gas. The incomplete combustion of these mix-
cial interests or personal relationships that could have appeared to
tures in the downstream region of the combustor is likely the rea-
influence the work reported in this paper.
son for high CO emissions. With an increasing equivalence ratio,
the flame pockets increase in size, and less leakage of the un-
burned mixture to the exhaust gas is expected. This explains the Acknowledgements
decreasing trend of CO emission with an increasing equivalence
ratio. If the flame were of flamelet type, i.e., without local extinc- The authors gratefully acknowledge the financial support from
tion and flame pockets, there would not be fuel leakage to the ex- the Swedish Research Council, Swedish Energy Agency, and
haust gas since the fuel would be oxidized under these lean con- Siemens Energy AB through Project No. 44120-1 and within the
ditions. The flame temperature would increase with increasing , Centre for Combustion Science and Technology (CECOST, 22538),
and the water-gas shift reaction CO + OH = CO2 + H would in- the Knut & Alice Wallenberg foundation (KAW COCALD project).
crease CO emission with increasing  (thus temperature, accord-
ing to the Le Chatelier principle), as predicted in the numerical References
simulations.
[1] T. Lieuwen, V. Yang, R. Yetter, Synthesis Gas Combustion: Fundamentals and
Applications, CRC press, 2009.
4. Conclusions [2] N.A. Samiran, J.-H. Ng, M.N. Mohd Jaafar, A. Valera-Medina, C.T. Chong, H2-rich
syngas strategy to reduce nox and co emissions and improve stability limits
The effect of CO2 dilution on the combustion of syngas mix- under premixed swirl combustion mode, Int. J. Hydrogen Energy 41 (42) (2016)
19243–19255.
ture (53% H2 , 44% CO, 3% CH4 on a volume basis) in a gas tur- [3] N.A. Samiran, M.N.M. Jaafar, J.-H. Ng, S.S. Lam, C.T. Chong, Progress in biomass
bine model combustor was investigated under atmospheric pres- gasification technique – with focus on malaysian palm biomass for syngas pro-
sure and a Reynolds number of 10 0 0 0. The Karlovitz number of duction, Renew. Sustain. Energy Rev. 62 (2016) 1047–1062.
[4] C. Dong, Q. Zhou, Q. Zhao, Y. Zhang, T. Xu, S. Hui, Experimental study on the
the investigated flames is about 1. The main goal was to under- laminar flame speed of hydrogen/carbon monoxide/air mixtures, Fuel 88 (10)
stand the impact of CO2 dilution on the operability ranges of the (2009) 1858–1863.
combustor, on the flame and flow structures, and the CO and NOx [5] J. Natarajan, T. Lieuwen, J. Seitzman, Laminar flame speeds of h2/co mixtures:
effect of co2 dilution, preheat temperature, and pressure, Combust. Flame 151
emissions in the exhaust gases. Optical measurement techniques
(1) (2007) 104–119.
(flame luminescence, OH-PLIF, and PIV) were used to analyze the [6] C. Liu, B. Yan, G. Chen, X.-S. Bai, Structures and burning velocity of biomass
flames in the combustion chamber, while exhaust gas analyzers derived gas flames, Int. J. Hydrogen Energy 35 (2) (2010) 542–555.
were used for the CO and NOx emissions. The main findings are: [7] B. Yan, Y. Wu, C. Liu, J. Yu, B. Li, Z. Li, G. Chen, X.-S. Bai, M. Alden, A. Kon-
nov, Experimental and modeling study of laminar burning velocity of biomass
derived gases/air mixtures, Int. J. Hydrogen Energy 36 (5) (2011) 3769–3777.
• The stable flame regime of the combustor shifts towards fuel- [8] L.D. Thi, Y. Zhang, Z. Huang, Shock tube study on ignition delay of multi-com-
richer conditions with the increasing CO2 dilution in the mix- ponent syngas mixtures – effect of equivalence ratio, Int. J. Hydrogen Energy
ture. The shift of stable flame regime is more significant at 39 (11) (2014) 6034–6043.
[9] H. Lee, L. Jiang, A. Mohamad, A review on the laminar flame speed and ig-
higher CO2 dilution levels. The change in stable flame regime nition delay time of syngas mixtures, Int. J. Hydrogen Energy 39 (2) (2014)
to fuel-richer mixtures with CO2 dilution is attributed to both 1105–1121.
chemical and thermal effects of CO2 , with the chemical impact [10] J.S. Kim, J. Park, O.B. Kwon, D.S. Bae, J.H. Yun, S.I. Keel, A study on flame struc-
ture and extinction in downstream interaction between lean premixed ch4–air
being more significant at high CO2 dilution conditions. and (50% h2+50% co) syngas–air flames, Int. J. Hydrogen Energy 36 (9) (2011)
• At a constant equivalence ratio, the flame luminescence de- 5717–5728.
creases with higher CO2 dilution in the mixture, indicating [11] N.A. Samiran, J.-H. Ng, M.N. Mohd Jaafar, A. Valera-Medina, C.T. Chong, Swirl
stability and emission characteristics of co-enriched syngas/air flame in a pre-
that the chemical reactions are suppressed by the CO2 dilu-
mixed swirl burner, Process Saf. Environ. Protect. 112 (2017) 315–326.
tion. The syngas flame without CO2 dilution exhibits a continu- [12] S. Li, X. Zhang, D. Zhong, F. Weng, S. Li, M. Zhu, Effects of inert dilution on the
ous but non-homogeneous OH PLIF signal distribution. The syn- lean blowout characteristics of syngas flames, Int. J. Hydrogen Energy 41 (21)
gas flames with high CO2 dilution exhibit significantly different (2016) 9075–9086.
[13] D. Pashchenko, Hydrogen-rich fuel combustion in a swirling flame: cfd-mod-
flame structures, with combustion occurring in isolated pockets eling with experimental verification, Int. J. Hydrogen Energy 45 (38) (2020)
surrounded by unburned fuel/air mixtures. This flame structure 19996–20 0 03.
significantly differs from flamelet combustion typically found [14] U. Azimov, E. Tomita, N. Kawahara, Y. Harada, Effect of syngas composition on
combustion and exhaust emission characteristics in a pilot-ignited dual-fuel
in premixed methane/air flames under similar Karlovitz num- engine operated in premier combustion mode, Int. J. Hydrogen Energy 36 (18)
bers. It is attributed to the diffusional-thermal instability of (2011) 11985–11996.
lean premixed flames with high hydrogen concentration in the [15] C.T. Chong, J.-H. Ng, M.S. Aris, G.R. Mong, N. Shahril, S.T. Ting, M.F. Zulkifli, Im-
pact of gas composition variations on flame blowout and spectroscopic charac-
fuel mixture. teristics of lean premixed swirl flames, Process Saf. Environ. Protect. 128 (2019)
• Flow field results at constant adiabatic flame temperature show 1–13.
that the mean flow field for CO2 diluted syngas flames devi- [16] N.A. Samiran, C.T. Chong, J.-H. Ng, M.-V. Tran, H.C. Ong, A. Valera-Medina,
W.W.F. Chong, M.N. Mohd Jaafar, Experimental and numerical studies on the
ates from the pure syngas flame. The inner recirculation zone
premixed syngas swirl flames in a model combustor, Int. J. Hydrogen Energy
(IRZ) moved closer to the combustor center axis. The IRZ be- 44 (44) (2019) 24126–24139.
came narrower under high CO2 dilution conditions because of [17] A. Lipatnikov, J. Chomiak, Molecular transport effects on turbulent flame prop-
agation and structure, Prog. Energy Combust. Sci. 31 (1) (2005) 1–73.
the lower mean heat release due to the local extinction of the
[18] T. Lieuwen, Flashback characteristics of syngas-type fuels under steady and
reaction zones. pulsating conditions, Technical Report, Georgia Institute of Technology, Atlanta,
• The CO2 dilution affects both the NOx and CO emission lev- GA (United States), 2007.
els. In particular, the NOx emissions were lowered with an in- [19] X. Liu, M. Bertsch, A.A. Subash, S. Yu, R.-Z. Szasz, Z. Li, P. Petersson, X.-S. Bai,
M. Aldén, D. Lörstad, Investigation of turbulent premixed methane/air and hy-
crease in dilution. The opposite trend characterizes CO emis- drogen-enriched methane/air flames in a laboratory-scale gas turbine model
sion. This trend is different from premixed methane/air flames combustor, Int. J. Hydrogen Energy 46 (24) (2021) 13377–13388.

13
F. Pignatelli, S. Derafshzan, D. Sanned et al. Combustion and Flame 255 (2023) 112912

[20] F. Pignatelli, H. Kim, A. Subash, X. Liu, R. Szasz, X.-S. Bai, C. Brackmann, [48] Y. Jafri, E. Wetterlund, M. Anheden, I. Kulander, Åsa Håkansson, E. Furusjö,
M. Aldén, D. Lörstad, Pilot impact on turbulent premixed methane/air and hy- Multi-aspect evaluation of integrated forest-based biofuel production path-
drogen-enriched methane/air flames in a laboratory-scale gas turbine model ways: part 2. economics, ghg emissions, technology maturity and production
combustor, Int. J. Hydrogen Energy (2022). potentials, Energy 172 (2019) 1312–1328.
[21] O. Tuncer, S. Acharya, J. Uhm, Dynamics, nox and flashback characteristics of [49] X. Liu, Experimental Studies of Turbulent Flames at Gas Turbine Relevant
confined premixed hydrogen-enriched methane flames, Int. J. Hydrogen Energy Burners and Operating Conditions, Ph.D. thesis, Combustion Physics, https:
34 (1) (2009) 496–506. //lup.lub.lu.se/search/files/96073734/Xin_Liu_WEBB.pdf, 2021.
[22] P. Strakey, N. Weiland, G. Richards, Combustion strategies for syngas and high- [50] F. Pignatelli, A.A. Subash, R.Z. Szasz, X.-S. Bai, M. Alden, D. Lörstad, P. Petersson,
-hydrogen fuel, The Gas Turbine Handbook (2006). Pilot impact on swirling combustion characteristics of premixed methane/air
[23] E. Giacomazzi, D. Cecere, F. Donato, F. Picchia, N. Arcidiacono, S. Daniele, P. Jan- flame structure, Accepted for Publication at the 12th Mediterranean Sympo-
sohn, Les analysis of a syngas turbulent premixed dump-combustor at 5 bar, sium, Luxor, Egypt, January 23–26, 2023.
Int. Conf. on Processes and Technologies for a Sustainable Energy, Ischia (2010). [51] H. Feuk, F. Pignatelli, A. Subash, R. Bi, R.-Z. Szász, X.-S. Bai, D. Lörstad,
[24] M.C. Lee, S.B. Seo, J. Yoon, M. Kim, Y. Yoon, Experimental study on the effect M. Richter, Impact of methane and hydrogen-enriched methane pilot injection
of n2, co2, and steam dilution on the combustion performance of h2 and co on the surface temperature of a scaled-down burner nozzle measured using
synthetic gas in an industrial gas turbine, Fuel 102 (2012) 431–438. Special phosphor thermometry, Int. J. Turbomach. Propuls. Power 7 (4) (2022).
Section: ACS Clean Coal [52] T. Sattelmayer, M.P. Felchlin, J. Haumann, J. Hellat, D. Styner, Second-gen-
[25] D.E. Giles, S. Som, S.K. Aggarwal, Nox emission characteristics of counter- eration low-emission combustors for abb gas turbines: burner development
flow syngas diffusion flames with airstream dilution, Fuel 85 (12) (2006) and tests at atmospheric pressure, J. Eng. Gas Turbines Power 36 (1) (1992)
1729–1742. 118–125.
[26] N. Papafilippou, M.A. Chishty, R. Gebart, On the flame shape in a premixed [53] E. Hodzic, Analysis of Flow Dynamics and Flame Stabilization in Gas Turbine
swirl stabilised burner and its dependence on the laminar flame speed, Flow, Related Combustors, Lund University, 2016 Ph.D. thesis.
Turbulence and Combustion 108 (2022) 461–487. [54] E. Hodzic, S. Yu, A.A. Subash, X. Liu, X. Liu, R.-Z. Szasz, X.-S. Bai, Z. Li, M. Alden,
[27] Q. Zhang, D.R. Noble, T. Lieuwen, Characterization of fuel composition effects Numerical and experimental investigation of the cecost swirl burner, ASME
in H2/CO/CH4 mixtures upon lean blowout, J. Eng. Gas Turbine Power 129 (3) Turbo Expo 2018: Turbomachinery Technical Conference and Exposition (2018).
(2006) 688–694. [55] A.A. Subash, S. Yu, X. Liu, M. Bertsch, R.-Z. Szasz, Z. Li, X.-S. Bai, M. Aldén,
[28] D. Littlejohn, R.K. Cheng, D.R. Noble, T. Lieuwen, Laboratory investigations of D. Lörstad, Flame investigations of a laboratory-scale cecost swirl burner at
low-Swirl injectors operating with syngases, J. Eng. Gas Turbine Power 132 (1) atmospheric pressure conditions, Fuel 279 (2020) 118421.
(2009). [56] H.J. Sheen, W. Chen, S. Jeng, T. Huang, Correlation of swirl number for a radi-
[29] S. Li, S. Li, D. Mira, M. Zhu, X. Jiang, Investigation of dilution effects on partially al-type swirl generator, Exp. Therm Fluid Sci. 12 (4) (1996) 444–451.
premixed swirling syngas flames using a les-lem approach, J. Energy Inst. 91 [57] D.G. Goodwin, H.K. Moffat, I. Schoegl, R.L. Speth, B.W. Weber, Cantera: An
(6) (2018) 902–915. Object-Oriented Software Toolkit for Chemical Kinetics, Thermodynamics, and
[30] T.C. Williams, C.R. Shaddix∗ , R.W. Schefer, Effect of syngas composition and Transport Processes, 2022.
co2-diluted oxygen on performance of a premixed swirl-stabilized combustor, [58] G.P. Smith, D.M. Golden, M. Frenklach, N.W. Moriarty, B. Eiteneer, M. Golden-
Combust. Sci. Technol. 180 (1) (2007) 64–88. berg, C.T. Bowman, R.K. Hanson, S. Song, W.C. Gardiner, J. Vitali, V. Lissianski,
[31] S.-G. Kim, J. Park, S.-I. Keel, Thermal and chemical contributions of added h2o Z. Qin, http://www.me.berkeley.edu/gri_mech/, 1999.
and co2 to major flame structures and no emission characteristics in h2/n2 [59] S. Walton, X. He, B. Zigler, M. Wooldridge, An experimental investigation of
laminar diffusion flame, Int. J. Energy Res. 26 (12) (2002) 1073–1086. the ignition properties of hydrogen and carbon monoxide mixtures for syngas
[32] C. Prathap, A. Ray, M. Ravi, Investigation of nitrogen dilution effects on the turbine applications, Proc. Combust. Inst. 31 (2) (2007) 3147–3154.
laminar burning velocity and flame stability of syngas fuel at atmospheric con- [60] J. Park, J.S. Kim, J.O. Chung, J.H. Yun, S.I. Keel, Chemical effects of added co2
dition, Combust. Flame 155 (1) (2008) 145–160. on the extinction characteristics of h2/co/co2 syngas diffusion flames, Int. J.
[33] N. Ding, R. Arora, M. Norconk, S.-Y. Lee, Numerical investigation of dilu- Hydrogen Energy 34 (20) (2009) 8756–8762.
ent influence on flame extinction limits and emission characteristic of [61] S.G. Davis, A.V. Joshi, H. Wang, F. Egolfopoulos, An optimized kinetic model of
lean-premixed h2–co (syngas) flames, Int. J. Hydrogen Energy 36 (4) (2011) h2/co combustion, Proc. Combust. Inst. 30 (1) (2005) 1283–1292.
3222–3231. [62] R. Ono, T. Oda, Spark ignition of hydrogen-air mixture, J. Phys. Conf. Ser. 142
[34] M. Arai, Flue gas recirculation for low nox combustion system, International (1) (2008) 012003.
Joint Power Generation Conference, 20 0 0-7 (20 0 0). [63] T. Guiberti, D. Durox, L. Zimmer, T. Schuller, Analysis of topology transitions of
[35] P. Røkke, J. Hustad, Exhaust gas recirculation in gas turbines for reduction of swirl flames interacting with the combustor side wall, Combust. Theor. Model.
co2 emissions; combustion testing with focus on stability and emissions, Int. 162 (2015) 4342–4357.
J. Thermodyn. 8 (4) (2005) 167–173. [64] M. Stöhr, C. Arndt, W. Meier, Transient effects of fuel–air mixing in a par-
[36] F. Liu, H. Guo, G.J. Smallwood, The chemical effect of co2 replacement of n2 tially-premixed turbulent swirl flame, Proc. Combust. Inst. 35 (3) (2015)
in air on the burning velocity of ch4 and h2 premixed flames, Combust. Flame 3327–3335.
133 (4) (2003) 495–497. [65] J.C. Massey, Z.X. Chen, M. Stöhr, W. Meier, N. Swaminathan, On the blow-off
[37] K. Mo, Y. Zhang, Z. Zhang, Y. Wang, Y. Xiao, X. Hui, Effect of fuel dilution on correlation for swirl-stabilised flames with a precessing vortex core, Combust.
the stability characteristics of syngas diffusion flames, Turbo Expo: Power for Flame 239 (2022) 111741.
Land, Sea, and Air 43130 (2008) 221–228. [66] G. Barenblatt, Y. Zeldovich, A. Istratov, On diffusional thermal stability of a
[38] A. Hayakawa, Y. Arakawa, R. Mimoto, K.K.A. Somarathne, T. Kudo, H. Kobayashi, laminar flame, J. Eng. Phys. Thermophys. 4 (1962) 21.
Experimental investigation of stabilization and emission characteristics of am- [67] G. Sivashinsky, Diffusional-thermal theory of cellular flames, Combust. Sci.
monia/air premixed flames in a swirl combustor, Int. J. Hydrogen Energy 42 Technol. 15 (1977) 137.
(19) (2017) 14010–14018. Special Issue on The 21st World Hydrogen Energy [68] J. Yu, R. Yu, X.-S. Bai, Onset of cellular instability in adiabatic h2/o2/n2 pre-
Conference (WHEC 2016), 13-16 June 2016, Zaragoza, Spain mixed flames anchored to a flat-flame heat-flux burner, Int. J. Hydrogen En-
[39] M. Escudier, J.J. Keller, Recirculation in swirling flow - a manifestation of vor- ergy 38 (34) (2013) 14866–14878.
tex breakdown, AIAA J. 23 (1985) 111–116. [69] N. Peters, Turbulent Combustion, Cambridge Monographs on Mechanics, Cam-
[40] A.K. Gupta, D.G. Lilley, N. Syred, Swirl Flows, Abacus Press, Tunbridge Wells, bridge University Press, 20 0 0.
England, 1984. [70] A. Buschmann, F. Dinkelacker, T. Schäfer, M. Schäfer, J. Wolfrum, Measurement
[41] O. Lucca-Negro, T. O’Doherty, Vortex breakdown: a review, Prog. Energy Com- of the instantaneous detailed flame structure in turbulent premixed combus-
bust. Sci. 27 (4) (2001) 431–481. tion, Symp. (Int.) Combust. 26 (1) (1996) 437–445.
[42] S. Candel, D. Durox, T. Schuller, J.-F. Bourgouin, J.P. Moeck, Dynamics of [71] K.-J. Nogenmyr, C. Fureby, X.-S. Bai, P. Petersson, R. Collin, M. Linne, Large
swirling flames, Annu. Rev. Fluid Mech. 46 (1) (2014) 147–173. eddy simulation and laser diagnostic studies on a low swirl stratified premixed
[43] R.L. Speth, A.F. Ghoniem, Using a strained flame model to collapse dynamic flame, Combust. Flame 156 (1) (2009) 25–36.
mode data in a swirl-stabilized syngas combustor, Proc. Combust. Inst. 32 (2) [72] J.M. Donbar, J.F. Driscoll, C.D. Carter, Reaction zone structure in turbulent non-
(20 09) 2993–30 0 0. premixed jet flames—from ch-oh plif images, Combust. Flame 122 (1) (20 0 0)
[44] T. García-Armingol, J. Ballester, Operational issues in premixed combustion of 1–19.
hydrogen-enriched and syngas fuels, Int. J. Hydrogen Energy 40 (2) (2015) [73] R. Sadanandan, M. Stöhr, W. Meier, Simultaneous oh-plif and piv measure-
1229–1243. ments in a gas turbine model combustor, Appl. Phys. B 90 (3) (2008).
[45] M.C. Lee, J. Yoon, S. Joo, Y. Yoon, Gas turbine combustion characteristics of [74] P. Perona, J. Malik, Scale-space and edge detection using anisotropic diffusion,
h2/co synthetic gas for coal integrated gasification combined cycle applica- IEEE Trans. Pattern Anal. Mach. Intell. 12 (7) (1990) 629–639.
tions, Int. J. Hydrogen Energy 40 (34) (2015) 11032–11045. [75] J. Weickert, B. Romeny, M. Viergever, Efficient and reliable schemes for nonlin-
[46] S. Park, U. Kim, M. Lee, S. Kim, D. Cha, The effects and characteristics of hydro- ear diffusion filtering, IEEE Trans. Image Process. 7 (3) (1998) 398–410.
gen in sng on gas turbine combustion using a diffusion type combustor, Int. J. [76] N. Mukundakumar, D. Efimov, N. Beishuizen, J. van Oijen, A new preferen-
Hydrogen Energy 38 (29) (2013) 12847–12855. tial diffusion model applied to fgm simulations of hydrogen flames, Combust.
[47] B. Dam, G. Corona, M. Hayder, A. Choudhuri, Effects of syngas composition on Theor. Model. 25 (7) (2021) 1245–1267.
combustion induced vortex breakdown (civb) flashback in a swirl stabilized
combustor, Fuel 90 (11) (2011) 3274–3284.

14

You might also like