Combustion and Flame 152 (2008) 604–615

www.elsevier.com/locate/combustflame

Investigation of NO

x

conversion characteristics

in a porous medium

S. Afsharvahid

∗

, P.J. Ashman, B.B. Dally

The University of Adelaide, Adelaide, South Australia 5005, Australia

Received 30 March 2007; received in revised form 16 June 2007; accepted 19 June 2007

Available online 13 August 2007

Abstract

The conversion of nitric oxide (using CNG/air as fuel/oxidizer) inside a porous medium is investigated in this

study. Unlike freely propagating flames, porous burners provide a solid medium that facilitates heat exchange
with the gaseous phase. The heat exchange allows the stabilization of a variety of fuel mixtures from lean to
rich and with a variety of calorific values. In addition, it allows the control of the reaction zone temperature
and thus the control of pollutant formation while maintaining flame stability. An experimental porous burner
was designed and manufactured for this purpose. The effects of equivalence ratio and flow velocity on the flame
stabilization, NO

x

and TFN (total fixed nitrogen) conversion ratios, and temperature profiles along the burner are

investigated. In addition, numerical calculations using the PLUG flow simulator model and the GRI 3.0 kinetic
mechanism reveals the key reactions which control the conversion efficiency. It was found that under slightly
fuel-rich conditions (ϕ

 1.3) NO

x

mostly converts to N

2

with a maximum conversion ratio of 65%, while for

higher equivalence ratios (ϕ > 1.3) a large proportion of NO

x

converts to NH

3

. Results from experiments and

numerical modeling showed that the temperature profile along the burner has significant effects on the NO

x

and

TFN conversion ratios. It was also found that temperatures between 1000 and 1500 K are most desirable for NO

x

and TFN conversion in the porous burner. Analysis of the chemical paths for the low- and high-equivalence-ratio
cases showed that the formation of nitrogen-containing species under very rich conditions (ϕ > 1.3) is due to the
increased importance of the HCNO path as compared to the HNO path. The latter is the dominant path at low
equivalence ratios (ϕ

 1.3) and leads to the formation of N

2

. The NO concentration in the initial mixture was

found to improve the conversion by up to 20% at low equivalence ratios (ϕ

 1.3) and to have negligible effect at

higher equivalence ratios.

©

2007 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Porous burner; NO

x

reburning; TFN

*

Corresponding author. Fax: [phone].
E-mail address:

[e-mail]

(S. Afsharvahid).

1. Introduction

The emission of nitrogen oxides (NO

x

) from com-

bustion systems has attracted substantial interest for a
long time. Their contribution to acid rain and smog
and strictly imposed regulatory requirements have
motivated researchers worldwide to investigate new

0010-2180/$ – see front matter

© 2007 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

doi:10.1016/j.combustflame.2007.06.017

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

605

and innovative methods of preventing or reducing
NO

x

emission from fossil fuel combustion. New

combustion technologies with low NO

x

emission

have been developed and utilized. The lifetimes of
most stationary combustion systems are measured in
tens of years; hence the treatment of exhaust gases
to abate the emission of NO

x

becomes more viable.

Different approaches to NO

x

control developed in the

past; include burner replacement, fuel staging, fuel re-
burning, steam or water injection, selective catalytic
reduction, and selective noncatalytic reduction

[1]

.

In many instances these have led to a reduction in
combustion efficiency and have required expensive
retrofits

[2]

. Among several reducing techniques

[2]

,

reburning

[3]

is a cost-effective technology

[4]

in

which nitric oxide is consumed using fuel as a re-
ducing agent. This method was developed in the
early 1950s and was termed “reburning” by Wendt
et al.

[3]

. NO

x

reduction of 50–70%

[5]

is achiev-

able using such an approach. The process involves
partial oxidation of the reburning fuel, under fuel-rich
conditions, followed by the reaction of hydrocarbon
radicals (CH

x

) and NO. This results in the forma-

tion of intermediate nitrogen species such as HCN
and subsequent conversion to N

2

. More recently, de-

tailed analysis of the reburning chemistry by Dagaut
et al.

[6]

and an earlier study by Kilpinen et al.

[7]

have revealed the important role of the HCCO rad-
ical, which reacts with NO to produce either HCNO
or HCN, both of which are then subsequently reduced
to N

2

. In general, reburning happens when gaseous,

liquid, or solid hydrocarbon fuels are injected down-
stream of the main combustion zone. Further reduc-
tion of NO

x

species, through advanced reburning and

second-generation advance reburning, is also possi-
ble via the addition of agents such as ammonia

[8]

or urea

[9]

. NO

x

reduction up to 90% is achiev-

able using advanced reburning technologies, which
invariably involve catalytic surfaces. Such successful
results encouraged researchers to conduct bench-scale
and pilot-scale reburning experiments in order to de-
velop the technology to a mature stage where it can
be implemented in existing combustion systems.

One of the earliest industrial-scale systems for

NO

x

reburning was developed in Japan. Their sys-

tem was capable of reburning 50% of the NO

x

on a

full-scale boiler by Mitsubishi in early 1980s. Later
Babcock and Wilcox

[10]

in Japan successfully ap-

plied the technology to a few wall-fired utility burn-
ers. Since then the reburning process has been inves-
tigated in many different combustion systems such as
flow reactors

[11]

and jet-stirred reactors

[12,13]

.

Despite the successful progress achieved, reburn-

ing is still considered a very complex process that
relies on several parameters that influence its per-
formance. It also involves a number of interrelations

among these parameters, which makes it very difficult
to study the influence of each variable individually.
The most important parameters that control the re-
burning efficiency are finite-rate mixing, equivalence
ratio, and reaction zone temperature

[7,14,15]

.

NO

x

reburning has been extensively analyzed for

a wide range of temperatures. Maintaining a uniform
constant temperature over the experimental domain
has made the experiments a real challenge. Bilbao
et al.

[16]

have conducted a range of experiments in

a 1500-mm-long quartz tube that has a diameter of
23 mm. Observing a longitudinal temperature profile,
they assumed that the temperature is almost constant
in a 600-mm zone in the middle of the tube. They used
this zone to determine the gas residence time. Their
results showed that for moderately low temperatures
(between 600 and 1100

◦

C), NO

x

reduction efficiency

increases as the temperature increases. It was found
that at a given temperature, the fuel effectiveness in
NO

x

reduction follows the sequence acetylene, ethyl-

ene, ethane, natural gas, and methane. It is also found
that natural gas and methane are of greater interest
for high-temperature applications, while acetylene is
more suitable for low-temperature cases. Dagaut et al.

[12]

explained this sequence through the importance

of HCCO intermediate for the production of HCN and
extended the list to include propene. They noted that
more HCCO is produced from acetylene fuels than
from the other reburning fuels, especially at lower
temperatures.

Using similar experimental techniques, Dagaut

and Ali

[17]

extended this study to include a liquefied

petroleum gas (LPG) blend for a range of equivalence
ratios and a temperature range of 950–1450 K. They
found that LPG as a reburning fuel follows the same
general oxidation path already delineated for simple
alkanes.

Williams and Pasternack

[18]

used a McKenna flat

flame burner to investigate the intermediates of pre-
mixed flames for a variety of fuels doped with NO.
They showed that for a fixed temperature of

∼1800 K

the methane, ethane, and ethylene flames all have
similar reburning chemical pathways, while acetylene
flames are quite different. They noted that acetylene
produces about three times as much CN and NCO
compared to other fuels.

It is clear from the above that moderately low tem-

peratures, fuel-rich conditions, and sufficiently long
residence times are required to achieve reduction of
NO

x

using hydrocarbon fuels. It is also apparent that

well-controlled conditions are required to better opti-
mize the conversion efficiency.

Porous burners

[19–26]

were found to be a suit-

able medium to achieve such controlled conditions.
The large inner surface area of the porous bed en-
sures efficient heat transfer

[27]

between the solid and

606

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

gas phases. Although, in most porous burners, overall
flow is calculated in the laminar range, in the vicinity
of the small pores within the porous media there is a
likelihood of turbulent gas flow. This effect locally in-
creases the effective diffusion and heat transfer within
the gas phase. These phenomena result in a different
temperature pattern to the adiabatic flame temperature
for different equivalence ratios. The maximum tem-
peratures in porous media for a range of equivalence
ratios between 1 and 2 are considerably below the adi-
abatic flame temperature. This great advantage results
in better NO

x

conversion and makes porous burners a

suitable postprocessor for NO

x

reburning.

In addition in porous burners, it is easier to control

the bed temperature and flame location by changing
either the flow rate or the equivalence ratio. Porous
burners also have reasonably uniform temperature
profile (axially and radially), which is believed to
be a result of the following enhanced heat transfer
processes:

• The high rate of solid to solid and solid to gas

radiation heat transfer caused by favorable radi-
ation characteristics of the SiC foams.

• The effective gas–solid convective heat transfer

by a constant stream of hot gases passing through
the solid phase downstream.

• The high rate of solid–solid conductive heat

transfer through the interconnected structure of
the silicon carbide (SiC) foams.

These characteristics are indications of a promis-
ing and desirable medium for the NO

x

conversion

process. Furthermore, relatively low operating tem-

perature leads to lower CO concentrations for fuel-
rich conditions

[28]

.

Despite an extensive literature search, there seems

to be little previous research on NO

x

reduction us-

ing porous burners. Besides our experimental study
in 2005

[29]

, which reported on the effects of equiva-

lence ratio, flow velocity, and input NO level on NO

x

reburning efficiency, Binguea et al.

[30]

were the only

group that experimentally studied the effect of equiv-
alence ratio on NO reburning in porous burners.

This study investigates flame stabilization over a

wide range of flow velocities and equivalence ratios
in a laboratory-scale porous burner and reports exten-
sively on parameters that could improve NO

x

con-

version inside a porous burner both experimentally
and numerically: equivalence ratio, temperature, resi-
dence time, and input NO level.

2. Experimental setup

An experimental porous burner was developed and

manufactured in the School of Mechanical Engineer-
ing at the University of Adelaide.

Fig. 1

shows a

sketch of the porous burner that mainly consists of
a mixing chamber followed by a heat exchanger and
an insulated ceramic porous bed.

The mixing chamber, which has a relatively large

volume, helps to achieve a better mixing of the inlet
gases. The inlet gases can include air, compressed nat-
ural gas (CNG), NO, and diluents such as nitrogen or
carbon dioxide, which could vary in ratio for different
cases.

The heat exchanger is made of four rows of lon-

gitudinal and lateral 1/4



copper tubes that use water

Fig. 1. Schematic of the porous burner apparatus.

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

607

as the coolant. The tubes are fitted in a copper box
(

150

× 150 × 100 mm) which is filled with flint clay.

Flint clay has a low thermal conductivity and during
the experiments was always kept below 200

◦

C by the

cooling coils. This flint clay section prevents flame
flashback and hence extends the operating envelop of
the porous burner. The small layer of flint clay be-
tween the cooling coils and porous bed is added to
provide preheating of the fresh reactants.

The porous bed is securely placed and sealed on

top of the heat exchanger and consists of four sili-
con carbide disks. Each disk is 50 mm thick and has
a diameter of 150 mm, a porosity of

∼90%, and 10

pores per inch (ppi). The porous bed is insulated us-
ing Kaowool paper (2600 grade) followed by Kaofil
Pumpable. R-type thermocouples are inserted circum-
ferentially inside the porous bed through premanufac-
tured grooves, at 25-mm intervals (staggered at 90

◦

).

Thermocouple bodies are insulated using 710 ceramic
(aluminum oxide), which has an emissivity very close
to the silicon carbide. Similar emissivity ensures that
the thermocouples are measuring the gas phase tem-
perature and are not affected by radiation from the
porous medium. The temperatures along the porous
medium were collected along the centerline of the
burner at 5-s intervals.

Gas samples were withdrawn from both the mix-

ing chamber and the exhaust gases using stainless
steel tubes. Gas samples are analyzed using a Chemi-
luminescent NO

x

analyzer (ML 9841B) for NO and

NO

x

(with NO

2

determined by difference). The NO

x

analyzer has a range of 0 to 20 ppm with a resolu-
tion of 0.001 ppm. For NO and NO

x

concentrations

in excess of 20 ppm, the sample gases are diluted with
N

2

using a dual channel sample diluter (ECOTECH

1412PD). The sample diluter was equipped with two
individual dilution channels (ratios of 5:1 and 8.9:1),
enabling the analysis of NO and NO

x

concentrations

up to 890 ppm.

The air and fuel are fed through the mixing cham-

ber, wherein NO and other diluents such as nitrogen
and carbon dioxide could be added to the mixture.
The mixture was ignited on the top, and the flame was
propagated upstream along the porous bed. For most
cases, flames were stabilized at a location very close
to the top of the heat exchanger. The effects of dif-
ferent equivalence ratios, flow velocities, and initial
NO

x

levels in the inlet gas mixture on flame stabi-

lization and NO

x

and TFN conversion efficiency are

investigated.

Here, the NO

x

conversion is defined as the ratio

of the outlet NO

x

concentration to the inlet NO

x

con-

centration. Similarly, the TFN (total fixed nitrogen)
conversion ratio is calculated as

TFN conversion ratio

= 1 − [TFN]/[NO

in

]

× 100,

where

[TFN]

= [NO] + [NO

2

]

+ [HCN] + [NH

3

]

+ 2 × [N

2

O].

Note that [TFN], corresponding to the overall concen-
tration of nitrogen compounds except N

2

, is the best

indicator of the efficiency of the reburning process.

3. Numerical modeling

The objective of the numerical calculations is to

provide insight into the chemical kinetics and NO

x

conversion chemistry for the different experimental
conditions considered in this study.

The PLUG simulator from the CHEMKIN 3.6

[31]

package was used in this study. The PLUG simu-

lator is designed to model a nondispersive and one-
dimensional chemically reacting ideal gas mixture
flowing in a channel of arbitrary geometry.

The burner was designed to have a uniform radial

temperature profile. In addition, the tubular burner
shape, the very high porosity of the foams, and the rel-
atively large burner diameter result in the flow being
similar to that of a plug flow reactor. In the model-
ing calculations, the reactor temperatures were set to
values equal to those measured experimentally. Other
assumptions, including fully mixed inlet gases, uni-
form gas concentration across the burner, and small
pressure drop along the burner, were also achieved
experimentally. The GRI 3.0 chemical kinetic mech-
anism

[32]

was used for all kinetic modeling calcula-

tions. Some other assumptions were also made, which
are discussed later.

The pressure drop (N/m

2

) across the porous me-

dia

[33–35]

is calculated as

(1)

P

= LV

bed



μ

f

K

+

ρF

√

K

V

bed



.

In the above equation V

bed

(m/s), μ

f

(kg s/m), K

(m

2

), ρ (kg/m

3

), F (m

−1

), and L (m) are the cross-

sectional mean velocity (also called Darcian or seep-
age velocity

[34]

), dynamic viscosity, permeability,

gas density, inertia coefficient, and bed length, respec-
tively.

For the conditions used in our experiments, it was

found that the maximum calculated pressure drop
along the porous foam is 243 Pa. With such a small
pressure drop, the assumption of isobaric conditions
is valid. Hence, atmospheric pressure was assumed
for all the calculations.

The fuel used in the experiments was CNG, which

comprised methane, ethane, propane, butane, nitro-
gen, and carbon dioxide, with volumetric fractions as
given in

Table 1

.

608

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

Table 1
CNG composition used in numerical calculations

CNG component CH

4

C

2

H

6

C

3

H

8

C

4

H

10

N

2

CO

2

Volume fraction 90.9% 5.0% 1.1% 0.3%

0.3% 2.4%

Since GRI 3.0 does not include any reactions of

C

4

H

10

, the 0.3% of C

4

H

10

in the CNG was replaced

with the same amount of C

3

H

8

for the calculations.

4. Results and discussion

In the following, the burner characteristics under

different operating conditions are discussed. In addi-
tion, experimental results are presented that validate
and clarify the assumptions made earlier in the nu-
merical modeling section.

Later on, the results from the experiments are pre-

sented, along with the numerical calculations. Both
experiments and modeling were performed for air
flow rates ranging from 50 to 200 slpm, equivalence
ratios of 1.0 to 2.0, and different inlet NO concentra-
tions up to 1800 ppm. The effect of critical parameters
on NO

x

and TFN conversion efficiency, such as flow

rate (resident time), equivalence ratio, and initial NO

x

concentration, is also presented and discussed.

4.1. Characteristics of the burner

For conventional open flames the maximum flame

temperature occurs at an equivalence ratio slightly
larger than unity.

Fig. 2

shows the maximum mea-

sured temperature in this porous burner plotted against
the equivalence ratio for a CNG flame with an air
flow rate of 150 slpm. It is clear that the tempera-
ture increases with equivalence ratio until it becomes
constant at

∼1600 K prior to the flame becoming un-

stable and blows off the burner. The equivalence ratio
where the flame blows out was found to be different
for different flow rates. For example, flow rates of 50,
100, and 150 slpm are stable up to equivalence ratios
of 2.1, 2.0, and 1.9, respectively, which indicates that
lower flow rates have a wider range of flame stability.

Fig. 2

also shows the adiabatic flame temperature

and the ratio of heat extraction. It is clear that the mea-
sured temperature is much lower than the adiabatic
flame temperature at lower equivalence ratios, and
this difference decreases close to the blowout con-
ditions. This trend is consistent with that observed
for the heat extraction ratio. It is found that flames
with higher equivalence ratios stabilize further down-
stream from the heat exchanger and hence the ratio is
lower.

These trends are found to be consistent with

other trends reported in the literature, e.g., Binguea

Fig. 2. Flame temperature in the porous burner compared
with adiabatic flame temperature for different equivalence
ratios and air flow rate of 150 slpm (top). Solid line shows
the adiabatic flame temperature; (solid line, F) represents
maximum temperature; and (dashed line, F) shows the ratio
of heat extracted by the heat exchanger to the firing rate.

Fig. 3. Temperature versus equivalence ratio for a flow ve-
locity of 25 cm/s (top) from another porous burner

[36]

.

Propagation wave velocity plotted versus different equiva-
lence ratios (bottom)

[36]

.

et al.

[36]

.

Fig. 3

(top) shows the measured flame

temperature as well as the calculated adiabatic flame
temperature plotted against the equivalence ratio for
a similar burner

[36]

. It is clear that the difference

between the measured flame temperature and the adi-
abatic flame temperature decreases with the increase
in equivalence ratio. Also plotted in

Fig. 3

(bottom)

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

609

Fig. 4. 3D temperature profile presenting centerline axial temperature for different equivalence ratios and inlet air flow rate of
100 slpm.

Fig. 5. Measured radial temperature profile for an air flow
rate of 50 slpm and an equivalence ratio of 1.5. (Solid line,
F) and (solid line, 2) show radial temperatures at 85 and
135 mm above heat exchanger, respectively.

is the wave propagation rate for the same equivalence
ratio.

Fig. 4

shows the measured temperature profiles

along the centerline of the burner for different equiva-
lence ratio cases. The profiles exhibit a sharp increase
from ambient temperature very close to the heat ex-
changer (x

= 0), which points to the ignition of the

mixture and the axial location of the flame front. It is
clear from this figure that with an increase in equiva-
lence ratio the location of the peak shifts downstream
and away from the heat exchanger. Further down-
stream the temperature drops slightly because of the
heat loss through the burner walls until a large drop
close to the burner exit. This drop is believed to be
caused by radiation to the surroundings.

Fig. 5

shows measured radial temperature profiles

at two axial locations along the burner. These tem-
perature profiles were measured using thermocouples

located at 85 and 135 mm above the heat exchanger.
These temperatures were measured in flames with an
air flow rate of 50 slpm, an equivalence ratio of 1.5,
and at 90-s intervals.

It is clear that in the region around the center with

a diameter of at least 60 mm, the temperature varies
by less than 5 K, while at the burner edge the drop
is on the order of 20 K. This shows that temperature
measurements in the middle are not affected by heat
losses to the surrounding through side walls.

As a result, reasonably uniform axial tempera-

ture as shown in

Fig. 4

(between x

= 50 mm and

x

= 150 mm), in addition to a uniform radial tem-

perature as shown in

Fig. 5

(between r

= 0 mm and

r

= 30 mm) makes this porous burner a desirable

medium for the NO

x

and TFN conversion process.

These measurements also vindicate our assumption
for the numerical calculations, especially the one-
dimensional flow and the use of the centerline profile
as a representative of the temperature profile along the
burner.

4.2. Effect of equivalence ratio

Fig. 6

shows the NO

x

and TFN conversion effi-

ciency plotted versus equivalence ratios for an air flow
rate of 100 slpm. For all cases 100 ml of NO was
added to the mixture.

It is clear that NO

x

conversion increases when

equivalence ratio, and with it fuel concentration, in-
creases. NO

x

conversion starts at ϕ

= 0.93, increases

sharply up to ϕ

= 1.2, and plateaus at ϕ > 1.5. It was

found that the maximum NO

x

conversion is reached

at an equivalence ratio of 1.5 for 50 slpm, 1.7 for
100 slpm, and 1.9 for 150 slpm (not shown). By in-
creasing the fuel/air ratio beyond these values, the

610

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

Fig. 6. Measured NO

x

conversion ratio and calculated NO

x

and TFN conversion ratios plotted versus equivalence ra-
tio for an air flow rate of 100 slpm and initial NO level of
100 ml. Black solid line, gray solid line, and (dashed line,
2) represent the calculated TFN conversion, calculated NO

x

conversion, and experimental NO

x

conversion, respectively.

The dotted line shows the calculated residence time for dif-
ferent equivalence ratios.

NO

x

conversion efficiency remains the same for a

small range and eventually starts to drop. At the same
time, increasing the equivalence ratio far beyond these
values makes the flame unstable and blows it out of
the burner.

It is clear from this figure that the NO

x

and TFN

conversion profiles exhibit different behavior at dif-
ferent equivalence ratios. While NO

x

conversion in-

creased when increasing the equivalence ratio, the
predicted TFN concentration at the output revealed
different results. The analysis showed that NO does
not necessarily convert to N

2

for all conditions. For

equivalence ratios close to unity, most of NO is pre-
dicted to be converted to N

2

with low concentrations

of HCN, NH

3

, and N

2

O. As the fuel concentration in

the mixture increases, TFN conversion follows a trend
similar to NO

x

conversion for moderately fuel-rich

conditions (ϕ

 1.2) and opposite trends for higher

equivalence ratios. For ϕ > 1.2 most of the NO is
converted to N-containing species such as N

2

O, NH

3

,

and HCN and not to N

2

.

In comparing the experimental and numerical re-

sults in

Fig. 6

, one can notice that the measured

and predicted NO

x

conversion agree better at higher

equivalence ratios (ϕ > 1.5) and there is an incon-
sistency in the profile for moderately fuel-rich flames
(

1 < ϕ < 1.5). Nonetheless the trend of increased

NO

x

conversion with the increase in equivalence ra-

tio was captured well. However, the inconsistency is
believed to be caused by the physical location of the
flame front. Having equivalence ratios close to unity
results in higher propagation waves upstream, as pre-
sented in

Fig. 3

(bottom).

This causes the flame to move upstream close to

the heat exchanger, as shown in

Fig. 7

, and makes

the temperature measurement rather difficult. It is
worth noting that only the temperature at 25 mm
above the heat exchanger was recorded in the exper-
iments. These values were then fed into the PLUG
flow reactor. In the lower-equivalence-ratio cases the
flame front is likely to have been closer to the heat
exchanger, with a higher temperature than the one
recorded at 25 mm. This will lead to lower measured
NO

x

conversion, as seen in

Fig. 6

.

To understand the contribution of the various N-

containing species to the predicted NO

x

and TFN

conversion efficiency, an analysis of the chemical
pathway was conducted for two equivalence ratios.

Fig. 8

presents species mole fractions along the reac-

tor for a case with an equivalence ratio of 1.1, an air
flow rate of 100 slpm, and 100 ml of NO in the mix-
ture. Under these conditions (i.e., ϕ

= 1.1), the inlet

NO (896 ppm) is reduced predominately to molecu-
lar N

2

at the burner exit. The extent of NO conversion

to N

2

is calculated as 67%, based on the conserva-

tion of atomic nitrogen. Some NO is also converted
to NH

3

(42 ppm), N

2

O (30 ppm), and HCN (5 ppm),

Fig. 7. Axial temperature profile for different equivalence ratios and inlet air flow rate of 100 slpm (left). The same profiles are
repeated detailing the flame front (right). (Dashed line, "), (solid line, 2), and (dotted line, Q) represent equivalence ratios of
1.1, 1.5, and 1.9, respectively.

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

611

Fig. 8. Calculated concentrations of major N species for an
air flow rate of 100 slpm, an equivalence ratio of 1.1, and
100 ml of input NO. Note the different scales in the upper
and lower graphs.

but in much smaller amounts (see

Fig. 8

). The outlet

concentration of TFN is 291 ppm.

The mechanism of NO reduction under these con-

ditions, which has been elucidated by analyzing of the
rates of progress for each species, is shown schemati-
cally in

Fig. 9

. While reactions R212 and R214 form a

cycle, with NO reduced to HNO and back again, there
remains a net flux of NO to HNO. The reverse of re-
actions R197 and R280 accounts for the reduction of
HNO to NH. The larger flux of NH is reduced to N

2

O

(R199) and subsequently to N

2

(R183, R185), while a

smaller flux is reduced directly to N

2

(R198). A small

amount of NH is also reduced via NH

2

(

−R202) to

NH

3

(

−R277 and −R278):

H

+ NO + M → HNO + M,

(R212)

HNO

+ H → H

2

+ NO,

(R214)

HNO

+ H

2

→ NH + H

2

O,

(

−R197)

HNO

+ CO → NH + CO

2

,

(

−R280)

NH

+ NO → N

2

O

+ H,

(R199)

Fig. 10. Calculated concentrations of major N species for an
air flow rate of 100 slpm, an equivalence ratio of 1.7, and
100 ml of input NO. Note the different scales in upper and
lower graphs.

N

2

O

+ H → N

2

+ OH,

(R183)

N

2

O (

+ M) → N

2

+ O (+ M),

(R185)

NH

+ NO → N

2

+ OH,

(R198)

NH

+ H

2

→ NH

2

+ H,

(

−R202)

NH

2

+ H

2

→ NH

3

+ H,

(

−R277)

NH

2

+ H

2

O

→ NH

3

+ OH.

(

−R278)

Under very fuel-rich conditions (i.e., ϕ

= 1.7), the

mechanism is different and the inlet NO (848 ppm)
is reduced predominantly to NH

3

at the burner exit,

with smaller extents of NO reduction to N

2

and HCN,

as shown in

Fig. 10

. The extent of NO conversion

to NH

3

is 43% with an NH

3

outlet concentration of

487 ppm. Smaller amounts of NO are converted to
N

2

(92 ppm) and HCN (80 ppm). Under these con-

ditions, the outlet concentration of N

2

O is negligible.

The outlet concentration of TFN is 658 ppm and thus
only 22% of the inlet NO is converted to N

2

.

Here the NO reduction is initiated by reaction with

HCCO (R274) to form HCNO. HCNO is then re-

Fig. 9. Mechanism of NO reduction for an air flow rate of 100 slpm, an equivalence ratio of 1.1, and 100 ml of input NO. The
thickness of arrows is indicative only and is not accurately scaled.

612

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

Fig. 11. Mechanism of NO reduction for an air flow rate
of 100 slpm, ϕ

= 1.7, and 100 ml of input NO. Note that

arrows’ thicknesses are indicative and are not accurately
scaled.

duced via three separate channels (

Fig. 11

): the major

channel is via reaction R270 to form HNCO. The mi-
nor channels are via R271 forming HCN as a stable
product and via R272 forming NH

2

as an intermedi-

ate. HNCO is reduced directly to NH

2

(R265), which

then reacts with either H

2

(

−R277) or H

2

O (

−R278),

forming NH

3

as the major stable product. A minor re-

action sequence exists in which NO is reduced first to
HNO via reaction R212. HNO reacts with either CO
(

−R280) or H

2

(

−R197) to form NH, which then re-

acts with NO to form either N

2

(R198) as a stable

product or N

2

O (R199) as an intermediate; N

2

O re-

acts directly with H (R183) to form stable N

2

:

HCCO

+ NO → HCNO + CO,

(R274)

HCNO

+ H → H + HNCO,

(R270)

HCNO

+ H → OH + HCN,

(R271)

HCNO

+ H → NH

2

+ CO,

(R272)

HNCO

+ H → NH

2

+ CO,

(R265)

NH

2

+ H

2

→ NH

3

+ H,

(

−R277)

NH

2

+ H

2

O

→ NH

3

+ OH,

(

−R278)

H

+ NO + M → HNO + M,

(R212)

HNO

+ CO → NH + CO

2

,

(

−R280)

HNO

+ H

2

→ NH + H

2

O,

(

−R197)

NH

+ NO → N

2

+ OH,

(R198)

NH

+ NO → N

2

O

+ H,

(R199)

N

2

O

+ H → N

2

+ OH.

(R183)

It is clear from the above that different chemical paths
dominate at different equivalence ratios that lead to
the production of other intermediates such as NH

3

.

Fig. 12. Measured axial temperature profiles for differ-
ent air flow rates and for an equivalence ratio of 1.1.
(Dotted line, 2), (dashed line, F), (solid line, "), and
(dashed–dotted line, Q) represent air flow rates of 50, 100,
150, and 200 slpm, respectively.

The predictions of

Fig. 11

are broadly consistent

with the modeling results of Dagaut et al.

[6]

for sim-

ilar fuels and at similar equivalence ratios. Some dif-
ferences are apparent, which are to be expected given
the different kinetic mechanisms employed. Notably,
the GRI 3.0 mechanism does not include the impor-
tant channel HCCO

+ NO → HCN + CO

2

, and so

this is missing from

Fig. 11

. Thus the current cal-

culations are likely to underpredict the formation of
HCN

[6]

. This, of course, could not be verified in the

present experiments since [HCN] was not measured.

4.3. Effects of flow velocity (residence time)

Flame front location in porous burners is a func-

tion of the flow velocity (flow rate), the equivalence
ratio, and the rate of heat extraction from the system.
The residence time is related not only to the flow ve-
locity but also to the other parameters; i.e., for similar
equivalence ratios, increasing the flow velocity results
in higher flame temperatures, as seen in

Fig. 12

, and

a decrease in the residence time.

It is now well established that the increase in the

mixture flow rate affects the residence time in three
ways:

1. Increasing the mixture flow rate increases super-

ficial velocity and actual velocity inside the bed,
which leads to a shorter residence time, as pre-
sented in

Fig. 13

.

2. Increasing the mixture flow rate moves the flame

location further downstream. This decreases the
burner effective length, L

− f , in the following

equation and with it the effective residence time.
Effective residence time (τ

eff

)

in porous burners

is defined as the time at which the gas mixture
is exposed to temperatures where chemical reac-
tions can happen. The equation relates the effec-

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

613

Fig. 13. Measured and calculated NO

x

conversion ratios

plotted versus air flow rates for an input NO level of 100 ml
and ϕ

= 1.1. Black solid line, gray solid line, and (dashed

line, 2) represent calculated TFN conversion, calculated
NO

x

conversion, and experimental NO

x

conversion, respec-

tively. Dotted line represents the residence time for different
cases.

tive residence time to the operating conditions of
the burner:

τ

eff

=

T

in

(L

− f )A

up

ε

T

(x)

˙

Q

up

(2)

+

x



x

0

A

up

ε

dx

˙

Q

up

+ A

up

ln



T (x)

T

in

 .

The first term on the right-hand side represents
the gas residence time after the flame front and
the second term represents the gas residence time
between the inlet and the flame front (preheating
stage).
Applying boundary conditions x

= x

0

= 0,

T

(x)

= T

max

, and no preheating results in

(3)

τ

eff

=

T

in

(L

− f )A

up

ε

T

max

˙

Q

up

.

In the above equations, T

in

, T

max

, L, f , A

up

,

and ε are the gas mixture inlet temperature, flame
temperature, porous bed length, flame location,
upstream cross-sectional area of the burner, and
porosity, respectively. ˙

Q

up

is the inlet volumetric

flow rate.

3. Increasing the mixture flow rate increases the fir-

ing rate and the heat exchange between the solid
and the gas phases and consequently the flame
temperature, T

max

, in the above equation and

hence decreases the density and residence time,
as indicated in

Fig. 12

.

In summary, by increasing the flow velocity ˙

Q

up

,

the flame temperature T

max

and the flame location

(distance between flame front and the heat exchanger)
f

increase as well. This will decrease residence time

and as a consequence conversion efficiency, as can be
seen in

Fig. 13

. It should also be noted that in the

temperature range of 1000 to 1500 K, increasing the
flame temperature for the same residence time could
result in better NO

x

to N

2

conversion efficiency.

It is worth noting that in reality, increasing the to-

tal flow rate results in a reduced residence time and
higher temperatures along the burner. If the temper-
ature is within the desirable range of 1000–1500 K,
there will be a competition between the two effects
of production and destruction of NO. For tempera-
tures above 1500 K both temperature increase and
residence time decrease degrade NO

x

and TFN con-

version efficiency.

Fig. 13

shows a comparison between numerical

and experimental conversion results for different air
flow rates and for ϕ

= 1.1. Also plotted in

Fig. 13

is

the change of residence time with the air flow rate.
It is clear that the calculated NO

x

and TFN con-

centrations show trends fairly similar to that mea-
sured experimentally (NO

x

conversion only), albeit

with a difference in the absolute values. This differ-
ence is larger at lower flow rates and lower equiva-
lence ratios. This discrepancy is again believed to be
a product of the experimental setup, where the max-
imum flame temperature happens closer to the heat
exchangers and is not captured by the existing ther-
mocouples. It is worth noting that the above mixture
strength range can vary slightly from one system to
another. This range gets narrower when the flow rate
is increased as the flame moves further downstream.
As a result, discrepancy between the experimental
and numerical results was smaller for higher air flow
rates (e.g., 200 slpm in

Fig. 13

) or higher equiva-

lence ratios, where the flame stabilizes further down-
stream. In previous work

[29]

, the same phenomena

were observed experimentally and good agreement
was achieved in ultrarich combustion regimes and at
lower flow rates.

It is clear from

Fig. 13

that increasing the flow rate

from 50 to 200 slpm increases the rate of consumption
by a factor of 3.7, while the residence time decreases
by a factor of 4.75. This eventually results in a de-
crease of NO consumption and hence NO

x

and TFN

conversion efficiency.

4.4. Effect of NO

x

levels in the mixture

Fig. 14

shows the NO

x

and TFN conversion effi-

ciency plotted versus the inlet NO concentration for
an air flow rate of 100 slpm and ϕ

= 1.1. It was

found that the NO concentration in the inlet mixture
can change the conversion efficiency by up to 20%.
In general, the higher the NO level in the reactant
streams, the better the conversion efficiency. How-
ever, this trend is only sustained up to a certain level.

614

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

Fig. 14. Measured and calculated NO

x

conversion ratio plotted versus input NO level for air flow rate of 100 slpm, ϕ

= 1.1

(left), and ϕ

= 1.7 (right). Black solid line, gray solid line, and (dashed line, 2) represent calculated TFN conversion, calculated

NO

x

conversion, and experimental NO

x

conversion, respectively.

The desirable NO concentration was found to be a
function of the flow rate and almost independent of
the equivalence ratio. In other words, for any flow rate
there is an optimum inlet NO

x

/air ratio (or more accu-

rately a narrow range) that gives the best conversion
efficiency. This ratio seemed to be almost the same
for different equivalence ratios and decreased as flow
rate increased.

From this figure it is clear that for the lower equiv-

alence ratios there is a clear trend of increase in con-
version with NO initial level. For example, increasing
the initial NO level from 20 to 100 ml for an equiva-
lence ratio of 1.1 increases the conversion efficiency
by 20%, while for richer fuel conditions (ϕ

= 1.7)

the conversion efficiency increases by just 4% for the
same range of NO input levels.

5. Summary and conclusions

A laboratory-scale porous burner was designed

and constructed to examine the parameters con-
trolling NO

x

and TFN conversion inside a porous

medium. The experimental porous burner showed
consistent, reliable and stable behavior for a wide
range of conditions. Results from experiments were
compared to those calculated using the PLUG simu-
lator of the CHEMKIN package and with the GRI 3.0
kinetic mechanism.

It was found that porous burners are a suitable

medium for NO

x

conversion because of the ease with

which temperature, equivalence ratios, and residence
time can be controlled. This feature allows maximum
optimization to improve the conversion efficiency un-
der varied inlet conditions.

It was found that the best conversion was achieved

for a maximum bed temperature below 1500 K and
that NO

x

conversion of 90% is possible. It was

also found that for the slightly rich equivalence ra-

tio ϕ < 1.3, most of the NO converts into N

2

, while

for higher-equivalence-ratio cases, more intermediate
species are formed, in particular NH

3

. The best TFN

conversion efficiency (65%) was found at ϕ

= 1.1,

T

= 1370 K, and an air total flow rate of 100 slpm.

For similar equivalence ratios, it was also found

that lower flow rates or flow velocities result in bet-
ter conversion by increasing the residence time and
decreasing the temperature to the desirable temper-
ature range. While residence time was the most im-
portant parameter in the conversion process, lower
temperatures (for lower flow rates) were also found
to improve both NO

x

and TFN conversion. In addi-

tion, lower mixture flow rates were found to have a
wider range of flame stability.

Analysis of the chemical pathways for a low- and

high-equivalence-ratio case showed that the forma-
tion of nitrogen-containing species under very rich
conditions is due to the increased importance of the
HCNO path as compared to the HNO path. The lat-
ter is the dominant path at low equivalence ratios and
leads to the formation of N

2

.

The NO concentration in the initial mixture can

change the conversion by up to 20% at low equiva-
lence ratios (ϕ

 1.3) and have a negligible effect at

higher equivalence ratios.

Acknowledgment

The authors thank Mr. Graham Kelly, the combus-

tion laboratory manager, for his assistance in the con-
struction and maintenance of the experimental porous
burner.

References

[1] J. Makansi, W. Bartok, B.A. Folsom, Power (May

1993) 11–28.

S. Afsharvahid et al. / Combustion and Flame 152 (2008) 604–615

615

[2] R.D. Boardman, L.D. Smoot, in: L.D. Smoot (Ed.),

Fundamentals of Coal Combustion, Elsevier, The
Netherlands, 1993.

[3] J.O.L. Wendt, C.V. Sternling, M.A. Matovich, Proc.

Combust. Inst. 14 (1973) 897–904.

[4] C.A. Bertran, C.S.T. Marques, Braz. Chem. Soc. 15 (4)

(2004) 548–555.

[5] C.A. Bertran, C.S.T. Marques, R.V. Filho, Fuel 83

(2004) 109–121.

[6] P. Dagaut, J. Luche, M. Cathonnet, Proc. Combust.

Inst. 28 (2000) 2459–2465.

[7] P. Kilpinen, P. Glarborg, M. Hupa, Ind. Eng. Chem.

Res. 31 (1992) 1477.

[8] C. Shi, A.B. Walters, M.A. Vannice, Appl. Catal. 14

(1997) 175–188.

[9] V. Zamansky, P.M. Maly, V. Lissianski, Second Gen-

eration Advanced Reburning for High Efficiency NO

x

Control, Energy and Environmental Research Corpora-
tion, Irvine, 1999.

[10] Babcock and Wilcox, Demonstration of Coal Reburn-

ing for Cyclone Boiler NO

x

Control. Comprehensive

Report to Congress Clean Coal Technology Program,
1990.

[11] P. Glarborg, M.U. Alzueta, K. Dam-Johansen, Com-

bust. Flame 115 (1998) 1–27.

[12] P. Dagaut, J. Luche, M. Cathonnet, Combust. Flame

121 (2000) 651–661.

[13] P. Dagaut, J. Luche, M. Cathonnet, Fuel 80 (2001) 979–

986.

[14] T. Kolb, P. Jansohn, W. Leuckel, Proc. Combust.

Inst. 22 (1988) 1193–1203.

[15] C.M. Cha, J.C. Kramlich, Combust. Flame 122 (2000)

151–164.

[16] R. Bilbao, A. Milleria, M.U. Alzueta, L. Prada, Fuel

76 (14/15) (1997) 1401–1407.

[17] P. Dagaut, K. Hadj Ali, Fuel 82 (2003) 475–480.
[18] A.B. Williams, L. Pasternack, Combust. Flame 111

(1997) 87–110.

[19] J.R. Howell, M.J. Hall, J.L. Ellzey, Prog. Energy Com-

bust. Sci. 22 (1996) 121–145.

[20] S. Afsharvahid, B. Dally, F. Christo, in: Asia–Pacific

Conference, Nanjing, China, 2003.

[21] P.H. Bouma, L.P.H. Goey, in: First European Confer-

ence on Small Burner Technology and Heating Equip-
ment, Zurich, 1996.

[22] G. Brenner, K. Pickenacker, O. Pickenacker, D. Trimis,

K. Wawrzinek, T. Weber, Combust. Flame 123 (2000)
201–213.

[23] K. Hanamura, R. Echigo, S.A. Zhdanok, Int. J. Heat

Mass Transfer 36 (13) (1993) 3201–3209.

[24] G.A. Lyamin, A.V. Pinaev, Combust. Explos. Shock

Waves USSR 22 (5) (1986) 553–558.

[25] D. Trimis, in: Fluids 2000 Conference and Exhibit,

American Institute of Aeronautics and Astronautics,
Denver, CO, 2000.

[26] L.A. Kennedy, J.P. Binguea, A.V. Saveliev, A.A. Frid-

man, Proc. Combust. Inst. 28 (2000) 1431–1438.

[27] A.J. Barra, J.L. Ellzey, Combust. Flame 137 (1–2)

(2004) 230–241.

[28] J.R. Comparato, in: Western Coal Council, Burning

PRB Coal Seminar, Birmingham, AL, 2001.

[29] S. Afsharvahid, B. Dally, in: 5th Asia–Pacific Confer-

ence on Combustion, Adelaide, Australia, 18–20 July
2005.

[30] J.P. Binguea, A.A. Savaliev, L.A. Kennedy, Proc. Com-

bust. Inst. 31 (2007) 3417–3424.

[31] R.J. Kee, F.M. Rupley, J.A. Miller, M.E. Coltrin, J.F.

Grcar, E. Meeks, H.K. Moffat, A.E. Lutz, G. Dixon-
Lewis, M.D. Smooke, J. Warantz, CHEMKIN Collec-
tion, Release 3.6, Reaction Design, Inc., San Diego,
CA, 2001.

[32] P.S. Gregory, D.M. Golden, M. Frenklach, N.W. Mo-

riarty, B. Eiteneer, M. Goldenberg, C.T. Bowman,
R.K. Hanson, S. Song, W.C. Gardiner, V.V. Lissianski,
Z. Qin, GRI-MECH 3.0,

http://www.me.berkeley.edu/

grimech/

.

[33] K.C. Leong, L.W. Jin, Int. J. Heat Fluid Flow 27 (2005)

144–153.

[34] N. Dukhan, Exp. Fluids 41 (2006) 665–672.
[35] A.P. Philips, H.L. Schram, Am. Ceram. Soc. Bull. 74

(1991) 728–732.

[36] J.P. Binguea, A.A. Savaliev, A.A. Fridman, L.A.

Kennedy, Int. J. Hydrogen Energy 27 (2002) 643–
649.