st
21 International Symposium on Plasma Chemistry (ISPC 21)
Sunday 4 August – Friday 9 August 2013
Cairns Convention Centre, Queensland, Australia
Study of nitric oxide and carbon monoxide production in plasma assisted combustion by Quantum Cascade Laser Absorption Spectroscopy
G.D. Stancu1,2, M. Simeni Simeni1,2, C.O. Laux1,2
CNRS-UPR 288, Laboratoire E.M2.C, Grande Voie des Vignes, 92295 Châtenay-Malabry, France
2
Ecole Centrale Paris, Grande Voie des Vignes, 92295 Châtenay-Malabry, France
1
Abstract: Nitric oxide and carbon monoxide are pollutant and key intermediate species in
air-hydrocarbon combustion systems. Here, nanosecond repetitively pulsed (NRP) discharges
were used to stabilized lean methane-air flames. Mid-IR Quantum Cascade Laser Absorption
Spectroscopy was developed to perform in situ CO and NO density measurements in plasma
assisted combustion environments. NRP discharges were optimized for low NO emissions.
Keywords: nitric oxide, carbon monoxide, QCLAS, plasma assisted combustion, nanosecond
pulsed discharges.
1. Introduction
The interest for using lean combustion is the reduction
of pollutants, in particular the NOx emission. The disadvantage of the lean combustion is the initiation of flame
instabilities, which can lead to flame extinction, hence to
unburned hydrocarbons (loss of energy) and pollutants
release. The use of short-pulsed discharges to ignite and
to enhance flame stability has been of great interest during last decade. Particularly, the use of nanosecond repetitively pulsed (NRP) discharges to stabilize lean combustion types is a very promising approach [1,2]. These discharges produce large amounts of atomic oxygen [3,4], a
key radical for the flame stabilization mechanism for an
average power (∼ 100 W) of a small percentage of the
flame power (∼ 10 kW). However, little is known on NO,
CO and unburned hydrocarbons emissions and on the
discharge optimization possibilities for pollutants reduction in plasma-assisted lean flames [5].
Recent studies [6,7], have shown that the NRP discharges produce large amounts of nitric oxide compared
to flame emissions. Here first results of discharge optimization are presented.
In situ real time monitoring of pollutant species is also
a large concern, being very challenging [8] in environments such combustion power plants. Complex techniques,
e.g.
Laser
Induced
Fluorescence
or
Chemiluminescence analysis can be intrusive, need difficult calibration methods and are strongly dependent on
the knowledge of the quenching of the excited states
[9,10]. The ex situ UV, IR absorption based gas analyzers
can lead to large errors in particular for the short-lived
species. These difficulties were overcome here by using
high-resolution (10-3cm-1) rotation-vibration resolved absorption spectroscopy. Nitric oxide and carbon monoxide
emissions were measured using Mid-IR Quantum Cascade Absorption Spectroscopy at 1900.076 cm-1 and at
2055.40 cm-1, respectively, and at temperatures up to 1100
K. Diagnostics challenges such as spectral overlapping
with large hot absorption features (water, carbon dioxide),
the line strength and the line profile temperature and
pressure dependencies, and the detection at trace density
levels were addressed using spectroscopic simulations and
sensitive multi-pass White cells.
2. Experimental set-up
Figure 1 shows a scheme of the QCLAS (a QMACS
system) and plasma-assisted combustion setups. Lean
CH4-air premixed flames were stabilized by the NRP discharges at total flow rates of 4 - 5.8 m3/h. For efficient
stabilization the discharge was placed in the recirculation
zone, behind a bluff body. The NRP discharges were generated using a pin-to-pin electrode configuration (gap
distance 1-5 mm), by 10-ns high voltage pulses (3-9 kV)
at pulse repetition frequencies of 5-30 kHz.
Fig 1. QCLAS and plasma-assisted combustion experimental
set-up.
Experiments were performed with the flame confined in
a metallic tube of 50 cm length and 8 cm diameter. The
measurements were done at the tube outlet. The temperature of the burned gases at the tube outlet was measured
st
21 International Symposium on Plasma Chemistry (ISPC 21)
Sunday 4 August – Friday 9 August 2013
Cairns Convention Centre, Queensland, Australia
2.1
1.8
50
1.5
1.2
Intensity (mV)
where S(T) is the line strength of a specific absorption line
at the temperature T and N is the concentration of particles in all states.
Absorption techniques provide line-of-sight integrated
signals. Assuming cylindrical symmetry for the plasma-flame species an Abel inversion can be used to recover the local absorption coefficients. From lateral absorbance measurements (absorbance, i.e. τ(ν) = ln(I0/I)(ν)) we
obtained the absorbance function, τ(ν,x). The local absorption coefficients as a function of the radius, r, can be
obtained from:
-3
where I0(v) is the incident intensity, k(v) is the absorption
coefficient, and labs is the absorption length. The integrated absorption coefficient is related to species density by,
(2)
k d
ST N
16
3. Principle of diagnostics
Under conditions of weak absorption, the radiation intensity I(v) transmitted through a homogeneous sample is
given by the Beer–Lambert law,
(1)
I( ) I0 ( )exp( k labs)
Carbon monoxide is one major combustion intermediate species. Emitted in large amounts it is also a toxic
pollutant and its presence in the exhaust gases is an indication of combustion incompleteness. More details about
the results on CO measurements can be found in references [7,12], here a summary is given.
The density of CO was corrected for the linestrength
temperature dependency, at 1100 K the line strength of
the P21 transition increases by a factor 5.5 compared to
300 K. In figure 2 the absolute CO density at the tube
outlet in the exhaust gases is represented. From
Abel-inverted one-pass lateral scan absorption measurements the radial density profile was obtained based on
equation (3). No CO density was measured if the laser
beam was not passing above the tube exit.
CO density (10 cm )
by thermocouples and was found to be 1100 ( 70) K with
small variations function of experimental conditions. A
multi-pass White cell (up to 28 passes) was used to improve the sensitivity of the absorption system.
0.9
0.6
k(r, )
1
r
x
r2
2055.2
(x, )
x2
dx
(3)
The presence of large amounts of water and carbon dioxide in the flame and the exhaust gases (main combustion products) makes the CO and NO detection by Mid IR
absorption techniques challenging. The high temperature
of the flame and exhaust gases (600 - 2000 K) results in
exciting a larger number of transitions from higher rotational and vibrational quantum numbers. The effects of
the high temperature and of the high pressure broadening
(atmospheric pressure) lead to large spectral overlapping
of the plasma-flame species. The CO transition in the
electronic ground state, the fundamental band v(1,0) (P 21),
permit measurements at 1100 K with a negligible spectral
overlapping. The selection of the NO transition was more
difficult due to the very strong spectral overlapping of the
hot water and the carbon dioxide, which were present in
the exhaust gases (19 % H2O and 9.5 % CO2). The electronic ground state X1/2, fundamental band v(1,0) (R6.5)
transition allow measurements of NO if the background
absorption is known. Based on the Hitran data base
(HITEMP-2010) accurate line strengths were obtained to
correct for the Boltzmann population redistribution over
the rotational vibrational level system at high temperatures [11].
4. Results
x
25
0
0.3
R
lateral position
2055.4
-1
2055.6 (cm )
0.0
0
10
20
30
40
radial distance (mm)
Fig 2. CO densities function of radial distance at tube outlet. Lateral absorption spectra are show in the inset.
The density variation across the tube diameter (1.8-2.1
x1016 cm-3) was due to gas temperature, which was showing a reverse radial dependency. The molar fraction was
therefore constant at tube outlet. For equivalence ratios
(ER) ≥ 1 (ER=1 represents the stoichiometric composition) the measurements were done with one laser path
while for ER < 1 a White multi-pass cell was used. Using
NRP discharges in lean condition produces no significant
change of the CO density. For rich flames, about 20-30 %
more CO was measured. The typical CO density for lean
flames (equivalent ratio ER= 0.7 - 0.95) was on the order
of 5x1013 cm-3 and 6 x 1016 cm-3 for the rich flame
(ER=1.05). Because the burned gases flow in the confinement tube at temperatures in the range of 1100 – 2000
K and the combustion of CO is very efficient
(CO+1/2O2CO2) at high temperature, the CO density
strongly decreases (to 8 ppm). In rich flames, large
amounts of CO (<1%) are measured because the methane
combustion in the confinement tube was not complete.
In figure 3 the NO density is shown as function of ER.
The largest nitric oxide was found to be in the range of
st
21 International Symposium on Plasma Chemistry (ISPC 21)
Sunday 4 August – Friday 9 August 2013
Cairns Convention Centre, Queensland, Australia
In figure 4 changes of NO function of the peak voltage
is show for constant pulse repetition frequency of 30 kHz
and for 4-mm electrode gap distance. The nitric oxide
density decreases more than 2 times when the peak voltage changes from 8.5 to 5.5 kV.
240
NO density (ppm)
100 - 130 ppm (2.4 - 3.2 x 1015 cm-3) and was obtained
when NRP was running in pure air. The air was injected at
the same flow rates as those corresponding to air in combustion ER experiments. The measurements were performed at room temperature. No significant temperature
changes at the tube exit were observed. The measurements were performed at peak voltage 5.3 kV, electrodes
gap distance 4-mm and at repetition frequency of 30 kHz.
The applied voltage was the same for air and air-fuel. The
average discharge power is in the order of one percent of
flame power. For flame with plasma we see that the NO
density drastically decreases to the range of 20-30 ppm.
The total density reduction due to temperature increase at
the tube exit (from 300K to 1100K) was considered in
calculation of the molar fraction. The additional decrease
of a factor 4 represents the NO reduction due to combustion presence.
210
180
150
120
90
60
30
0
5.5
6.0
6.5
7.5
8.0
8.5
9.0
voltage (kV)
160
NO density (ppm)
7.0
140
Fig 4. NO molar fraction function of applied voltage (4-mm
120
gap distance, 30 kHz pulse repetition frequency).
100
plasma
plasma-flame
flame
80
60
In Figure 5 the NO density was measured for discharge gap distance varying from 1 to 5 mm. It can be
seen again a very important decrease of the NO molar
fraction by decreasing the gap distance.
40
150
0
0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05
ER
Fig 3. NO molar fraction as function of the equivalence
ratio for plasma, plasma+flame, flame (4-mm gap distance, 30 kHz pulse repetition frequency, 5.3 kV peak
voltage).
In the flame case the NO densities are found in the
range of 1.2-25 ppm for equivalent ratio range of 0.75 1.05. This corresponds to NO reduction from rich to
lean by a factor 20, a typical reduction for methane-air
flames. For 0.7 ER the NO density was below the detection limit.
The very large NO emissions by the discharge need to
be reduced. Optimization of discharge parameters in order
to significantly reduce NO density while still stabilizing
the lean flame is necessary.
A couple of experiments were carryout using only the
discharge in air, NO being measured at the tube outlet.
The air was injected at the same flow rates as those corresponding to air in combustion ER experiments.
NO density (ppm)
20
120
90
60
30
0
1
2
3
4
5
gap distance (mm)
Fig 5. NO molar fraction function of electrode gap distance (30
kHz pulse repetition frequency, 5.3 kV peak voltage).
Reducing the discharge repetition frequency (results
not presented here) reduces also the nitric oxide density.
The reduction of the NO density by decreasing the voltage and the pulse repetition frequency is explained by the
decrease of the discharge average power. The residence
time of the gas between the electrodes can explain the
results from figure 5 (decay of NO when reducing the gap
distance). The number of discharge pulses seen by the gas
st
21 International Symposium on Plasma Chemistry (ISPC 21)
Sunday 4 August – Friday 9 August 2013
Cairns Convention Centre, Queensland, Australia
between the electrodes is larger for 5-mm gap distance
than for 1-mm inter-electrode gap.
So, NO emissions from the NRP discharges can be
reduced if the discharge repetition frequency, discharge
applied voltage and inter-electrode gap distance is reduced. However stabilisation of methane-air lean flames
by NRP discharges occurs only in the spark discharge
regime and for certain domain of discharge parameters.
In figure 6 the NO density is plotted function of the
applied voltage for 1-mm gap distance and at 30 kHz
pulse repetition frequency. Confined methane-air lean
flames at ER=0.7 were stabilized when the applied voltage was over 4.2 kV. For this condition the measured NO
density produced by the discharge is about 30-40 ppm.
This is a factor 3-4 less than the NO density measured in
figure 3. If the blue curve data (plasma + flame) is downscaled by a factor 3-4 the obtained NO density in
plasma+flame is in the range of NO-flame at ER 0.85.
This suggests that optimising the NRP discharge parameters for low NO density (below the NO emitted by
stoichiometric flames) while providing stabilization of
lean air-methane flame is feasible.
NO density (ppm)
80
60
40
20
0
3.9
4.2
4.5
4.8
5.1
5.4
voltage (kV)
Fig 6. NO molar fraction function of discharge voltage
(1-mm gap distance, 30 kHz).
5. Conclusions
Mid-IR QCLAS has proven to be a suitable technique
highly selective and sensitive, capable of in situ, absolute
NO and CO density measurements in harsh environments
such as plasma-assisted combustion and at high exhaust
temperatures of up to 1100 K. Few orders of magnitude
changes of CO density were measured when changing
from lean to rich flames. For lean flames, the NRP discharges makes no significant CO density changes while
for rich flames 20-30 % increase was observed.
NRP discharge was shown to produce large NO densities which were strongly quenched by the flame presence.
The measured NO densities in the exhaust gases in case of
NRP air discharge and in air/methane flame+NRP was
found to be larger than in case of unassisted flames. Thus,
the main NO formation mechanism given by the
Zel’dovich reactions (thermal NO for flame) needs to be
completed by plasma-induced reactions such as shown in
reference [10].
Optimization of the NRP discharge for reduction of the
NO density while stabilizing lean air-methane flames is
possible by decreasing the gap distance, the applied voltage and the pulse repetition frequency. The strong reduction of NO obtained here after the optimisation of the
NRP discharge (e.g. factor 3-4), suggests that NO emissions in plasma-assisted combustion can be less than NO
density produced by stoichiometric flames. Further experimental investigation need to be done. For understanding the NO formation mechanisms advanced plasma-combustion kinetic models need to be performed.
Acknowledgments: This work was supported by the
ANR PlasmaFlame project (ANR 11BS0902501).
References
[1] G. Pilla, D. Galley, D.A. Lacoste, F. Lacas, D. Veynante, C.O. Laux, IEEE Transactions on Plasma
Science, 34, (6), 2471 (2006).
[2] Starikovskaia, S., J. Phys. D: Appl. Phys., 39 265
(2006).
[3] G.D. Stancu, F. Kaddouri, D.A. Lacoste, C.O. Laux,
40th Plasmadynamics and Laser Conferences San
Antonio TX, USA 3593 (2009).
[4] G.D. Stancu, F. Kaddouri, D.A. Lacoste, C.O. Laux, J
Phys. D: Appl. Phys., 43, 124002 (2010).
[5] D.A. Lacoste, J.P. Moeck, C.O. Paschereit, C.O. Laux,
J. Prop. Power. (2013) in press.
[6] M. Simeni Simeni, G.D. Stancu, C.O. Laux, 10-th
FLTPD, Rolduc, Kerkrade, Netherlands P2-14
(2013).
[7] G.D. Stancu, M. Simeni Simeni, C.O. Laux, 31-st
ICPIG, Granada, Spain, (2013).
[8] X. Chao, J.B. Jeffries, R.K. Hanson, Proceedings of
the Combustion Institute, 33 725 (2011).
[9] M. Uddi, N. Jiang, I.V. Adamovich, W.R. Lempert, J.
Phys. D: Appl. Phys. 42 075205 (2009).
[10] W. Kim, H. Do, M.G. Mungal, M.A. Cappelli, Proc.
Comb. Inst. 31 3319 (2007).
[11] L.S. Rothman, et al. Journal of Quantitative Spe
troscopy and Radiative Transfer, 111, 2139 (2010).
[12] G.D. Stancu, D.A. Lacoste, C.O. Laux, 21-th
ESCAMPIG, Viana do Castelo, Portugal, 272 (2012).