Vous êtes sur la page 1sur 25

This article was downloaded by: [Hernando Yepes]

On: 22 April 2013, At: 22:57


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Combustion Science and Technology


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gcst20

Experimental and Numerical Investigation of the Novel


Low NOx CGRI Burner
a a a
B.A. FLECK , A. SOBIESIAK & H.A. BECKER
a
Centre for Advanced Gas Combustion Technology Queen's University, Kingston, ON, K7L 3N6
Version of record first published: 27 Apr 2007.

To cite this article: B.A. FLECK , A. SOBIESIAK & H.A. BECKER (2000): Experimental and Numerical Investigation of the Novel
Low NOx CGRI Burner, Combustion Science and Technology, 161:1, 89-112

To link to this article: http://dx.doi.org/10.1080/00102200008935813

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
Combust. Sci. and Tech., 2000. Vol. 161. pp. 89-112 o 2000 OPA(Overseas Publishers Association) N.V.
Reprints available directly from the publisher Published by license under the
Photocopying permitted by license only Gordon and Breach Science Publishers imprint.
Printed in Malaysia

Experimental and Numerical


Investigation of the Novel Low NOx
eGRI Burner
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

B.A. FLECK', A. SOBIESIAK and HA BECKER

Centre for Advanced Gas Combustion Technology Queen's University, Kingston,


ON,KlL3N6

(Received November 30, 1999; Revised June 29. 2000)

This paper reports on an experimental and numerical investigation of the near field and combustion
zone of a burner that realizes a FOD! (Fuel/Oxidant Direct Injection) strategy for furnace firing.
FOD! is an non-premixed method of reactants delivery and employs a direct discharge of fuel and
oxidant jets into the furnace chamber. The jets entrain significant quantities of furnace gases that have
been cooled by furnace heat transfer. The fuel and oxidant streams arrive at the reaction zone diluted
by furnace gases, lowering the temperature of the reaction and reducing NOx emissions. FOOl
involves three-feed mixing processes linked with non-adiabatic reactions at unusually low tempera-
tures and reactant concentrations. The burner investigated here consists of fourteen fuel and air ports
arranged in a circle around a central pilot flame. The global performance characteristics of the burner
demonstrate the effectiveness of FOOl in NOxreduction and its visually flameless oxidation process.
The mathematical modelling of the burner using a commercial CFD package was capable of ade-
quately predicting jet trajectories and primary flow structures in the furnace. The predicted tempera-
ture and species concentrations depart from measured values thus raising a question of how to
effectively model three-feed processes under severe non-adiabatic conditions.

Keywords: Natural gas burner; flame less; low NO,; Furnace

1 INTRODUCTION

Market demands and legislative regulations on combustion technologies have


motivated the continued development of more energy efficient and less polluting
combustion systems. Burning of natural gas, one of the most favourable fuels
environmentally, can still result in emissions of oxides of nitrogen: NO (nitric
oxide) and N0 2 (nitrogen oxide), usually lumped as "NOx" . In industrial fur-
• Corresponding author: present address Dept. of Mech. Eng. University of Alberta, Edmonton.
Albena T6G2G8. Canada.

89
90 B.A. FLECK er of.

naces, increased energy efficiency is obtained by combustion air preheating, usu-


ally realized in recuperative or regenerative heat exchangers. The requirement of
low NOx in conjunction with high combustion air preheat represents a conflict of
interest between energy savings and emissions reduction goals and a challenge
for combustion researchers and design engineers.
The Canadian Gas Research Institute (CGRI) burner is a multi-jet, nominally
non-premixed, gas-fired burner intended to give 10w-NOx performance at high
air preheats. Design principles of the burner include a direct fuel and oxidizer
injection into the furnace atmosphere and directing all of the fuel on the outside
of the combustion air flow. Fuel and air undergo extensive mixing with recircu-
lating furnace gases (products of combustion) before meeting each other. The
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

recirculation of products that have lost considerable thermal energy by heat


transfer in the furnace is induced by burner jet entrainment and is shaped by fur-
nace aerodynamics. This mixing strategy results in exceptionally low NOx emis-
sions •. The global performance characteristics of CGRI burners fired into the
Centre for Advanced Gas Combustion Technology (CAGCT) research furnace
are described in a paper by Sobiesiak et aI. (1998). The burner NO x emission lev-
els ranged from 2 to 40 ppm @ 3% 02' with the high levels being reached only at
very high furnace exhaust-gas temperatures. Combustion was taking place with-
out any visible or audible flame, thus producing a combustion zone rather than
flame. Stability was acceptable, given tested turndown (75%), and the absence of
furnace pressure fluctuations.
The strategy featured in the CGRI burner has been applied to a limited degree
in burners designed by the International Flame Research Foundation (IFRF) (Jes-
see et aI., 1997), in the TwinBed II burner of the North American Manufacturing
Company (1995), and in a patent issued to Tokyo Gas (Nakamachi et al., 1990).
The FLOX burner investigated by WUnning and WUnning (1997) employed a
direct fuel and oxidizer injection, however, with an conventional single fuel jet
on the burner axis. This study by WUnning and WUnning identified a set of oper-
ating conditions with NOx emissions ranging from 6 to 100 ppm @ 3% 02 and
for which no visible flame was observed. To mark these conditions, the authors
coined a term "flarneless oxidation" (WUnning and Wunning, 1997). The CGRI
burner definitely operates in this so-called flameless oxidation mode as both
burners' design principles allow for an internal furnace gas recirculation which is
an order of magnitude higher than in normal burners.
The remarkable performance of the CGRI burner has its roots in the special
features of flow and mixing in the burner near field. In this paper, details of the
near field as determined by experimental and mathematical modelling work are
• EPA regulations (1995) for this facility would require NOx emissions around 100 ppm. signifi-
cantly higherthan the observed levels in this study.
INVESTIGATION OF CGRI BURNER 91

examined. The objective of this research is to characterize the CGRl burner near
field structure in terms of gas composition, velocity and temperature fields. The
experiments involved studies of the 320 kW CGRl burner mounted on the
CAGCT research furnace using gas sampling probes, microthermocouple probes
and laser Doppler velocimetry. Mathematical modelling with CFD was
employed using the commercial code TASCflow.

2 EXPERIMENTAL APPARATUS AND PROCEDURE


Downloaded by [Hernando Yepes] at 22:57 22 April 2013

The burner is comprised of seven air and seven fuel ports arranged in a circular
pattern around a central premixed pilot flame which continuously bums during
operation (Fig. I). The burner ports are angled at 15° and 10° from the burner
axis with diameters of 6.35 mm and 19.1 mm, for fuel and air respectively. The
15° fuel port angle in this investigation is much smaller than the fuel port angles
examined in the previous work (Sobiesiak et al., 1998) where they ranged from
30° to 65°. The relative angle between the fuel and air ports of 5° compared to
the value of 20° to 55° for the previous work.

Front Section

Fuel Air Nozzle


Nozzle
Angle

Fuel Nozzle
Angle

Furnace Wall

eGRI Burner Schematic


FIGURE 1 Schematic of CGRI burner (00110 scale)

The burner is fired into the CAGCT research furnace laboratory which is of
semi-industrial scale designed for research on gas-fired burners with recupera-
tive preheat (Fig. 2). For the purposes of this investigation, the furnace was con-
92 B.A. FLECK et al.

CAGCT Research' Furnace


O. 0®
f---.----- -----..
® Wall burner (ahaded one used)
·············~·--;(JIJ
qQ]
@] Roof burner (nol used) . , ,
c[]] [Q] Recuperelor
" ' '

ImI iii wfndows iqJJ


" roof porte used (Ol""rs not shown)
ic@J
....
I
~--._. --- --- .. ..
:idJ
---_ -_.-. _.- --_. ,. __.'
1\ ,

.Q .

--
......of,
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

4.5m

~ .-An
-
~
... --- _.. - _. --- "" .••.• ,<
r
1m III III

.. J 1______ '
..:.:.:.:.- --- _. ----- .:.:...
e:-:-:-:-
3m - - - :

FIGURE 2 Orthographic view of the CAGCf furnace

figured as shown in Fig. 3 which gives the general layout with the water-cooled
movable wall (used for near field temperature, sampling and velocity measure-
ments) on the near field. Its dimensions are nominally 3 x 4.5 x 1 metres in the x,
y and z directions respectively. This furnace configuration is slightly different
than that of the previous work (Sobiesiak et al., 1998) where three burners were
fired or a single burner was fired into a "tunnel" furnace formed by blocking off
part of the the CAGCT furnace with refractory bricks. Because of this, previous
works eliminated, for the most part, any cross flows such as those generated in
the current work due to the open furnace geometry. Such cross flows are of inter-
est since they frequently affect burner behaviour in practical industrial applica-
tions. The floor of the furnace is a water-cooled heat sink and the heat flux to it
was measured. Refractory insulation can cover all or some portion of the heat
sink to limit the heat flux to the sink and thus controls the furnace temperature.
For this work, the heat sink was partially insulated, exposing I m 2 of the bare
water-cooled floor. This resulted in a furnace exhaust gas temperature
=
Te 975°C and a recuperative air preheat temperature Ta= 290°C. Because the
furnace safety systems rely on UV emissions from a burner flame to indicate nor-
mal operation with combustion, the pilot flame (the only source of UV in the fur-
nace with this burner) was continuously run to prevent furnace shutdown.
INVESTIGATION OF CGRI BURNER 93

k: y

Ea exhaust wall
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

m floor (sink)
movable water
cooled waIl I!il refractory
FIGURE 3 Schematic of CAGer furnace and movable wall for probing the burner near field

The burner operated with natural gas at fuel mass flow rate (6.75 gls), giving
320 kW combustion heat release. Primary air delivery was adjusted to give
3 ± 0.2 mole % oxygen (dry basis) in the exhaust. Other parameters of the burner
operation are given in Table I. Measurements were taken after at least four hours
of continuous full fire. After this warm-up, the refractory temperatures continued
to gradually increase by as much as 20°C for the remainder of a day's measure-
ments; this transient accounts for the majority of the uncertainty in test condi-
tions.

TABLE I Characteristics and operatingconditions for the CGRI burner. The uncertainties are for a
given measurement (instrument uncertainty) with the maximum daily fluctuations given in
parentheses

Fuel (natural gas) Air

Portdiameter, mm 6.35 ± 0.1 19.5 ±O.I


Port angle 15° ± 3° 10° ± 1°
Burner inlet temperature, "C 18± I (±8) 290± I (±20)
Mass flow rate, g/s 6.75 ± 0.05(±0.25) 127.8 ± 0.9(±4.7)
Burner port velocity, mls 43.2 ± 0.5(±3) 102 ± 1(±9)
Jet Reynolds number 1.8 x \(f 4.2 x \(f

(;z (7 jets ),N 0.283 ± 0.006(±0.033) 12.8 or 13.6 ± 0.3(±1.7)


94 B.A. FLECK el al.

The three primary measurement techniques that were employed were gas sam-
pling probes, micro-thermocouples and laser Doppler velocimetry (LDV). For
each of these, detailed probing of the burner near field was done using a travers-
ing rig which allowed free movement in all three directions. This information
was complimented by burner refractory temperature measurements with fixed
thermocouples and heat flux measurements through the water-cooled floor pan-
els.

Gas sampling

The sample gas entered through a small-bore (0.794 mm inside diameter)


Downloaded by [Hernando Yepes] at 22:57 22 April 2013

ceramic tube. The attributes of the gas sampling system used are well docu-
mented (Turnbull, 1995). The sample line was heated (- 60 D C) to prevent water
condensation before reaching the sample preconditioner. The sample was ana-
°
lysed on a dry basis for 2, CO 2 , CO, NOx and CH 4 content.
Turbulent concentration fluctuations were for the most part filtered out by the
30 second time constant of the gas sampling system. Additional averaging was
imposed by taking the arithmetic mean of the analyser output over 10 seconds.

Thermocouples

The thermocouples were of Type "S"(Pt/Pt-lO%Rh) of 25.4~m (0.00 1") or


50.8~m (0.002") wire, prewelded, purchased from Omega Engineering Inc.. The
manufacturer states that the weld size is about three times the wire diameter.
The thermocouple is assumed to follow the first order response function,
dTjunction
T gas = T junction +T dt (1)

where a sample set of time constants r is found by monitoring the thermocouple


response to heating from an electrical pulse and a number of different furnace
locations. The set of time constants were fitted to the parameters of mean veloc-
ity and temperature. This was a compromise given the available information,
since ideally the time constant for each thermocouple is a function of instantane-
ous velocity, temperature and gas composition.
A correction for the radiative heat transfer between the thermocouple junction
and surroundings was also made using measured gas and refractory temperatures
and an estimated linear fit for the bead emissivity. The magnitude of the correc-
tion was between -5 and 20 D e.
A data acquisition rate of 400 Hz was found to be fast enough to retrieve any
signal higher than the noise level and to prevent aliasing problems. A sample size
INVESTIGATION OF CGRI BURNER 95

N = 8192 was used (to allow the use of the FFr for postprocessing) giving a sam-
ple duration of 20.48 seconds.
The derivative of the junction temperature in equation (I) was found in the
spectral domain and applied only to the Fourier coefficients above the noise floor
level, thus effectively filtering out the effect of noise on transient compensation.
The uncertainty in the mean temperature measurements was between 5 and 10
degrees outside the pilot ignoring catalytic effects. In relative terms, the uncer-
tainty in the RMS fluctuations was up to 16%.

Laser Doppler Velocimetry


Downloaded by [Hernando Yepes] at 22:57 22 April 2013

A 3W Argon ion laser tuned for maximum even light distribution between the
blue and green colours was aimed through a collimator and into a TSI COLOR-
BURST beam separator and into a TSI fibre optic probe. The system operated in
backscatter mode and was oriented to measure the components of velocity in the
direction of the burner axis (x) and in the vertical (z) direction. The receiving
fibre was connected to a TSI COLORLINK which was linked back to the
COLORBURST to drive the Bragg cell. The resultant two channel signal was
fed into a TSI I990-C counter system with its own adjustable bandpass active fil-
ters which was set to use 32, 16, or 8 zero crossings to establish burst frequen-
cies, depending on the flow velocity. The comparison criterion was I % for all
measurements, set so that erroneous signals generated by noise (Heitor et aI.,
1993) were rejected. The frequency data were sent through a data bus to a micro-
computer which handled the data processing, using software developed at the
CAGCT.
A water-cooled probe shroud was employed (purchased from the IFRF (Dugue
and Weber, 1994)) which kept the TSI probe and fibre optic cable at a safe tem-
perature. The primary air flow was seeded using an inverse cyclone seeder with
titanium dioxide powder.
Correction for velocity bias (McLaughlin and Tiederman, 1973) was made
employing the residence time calculated using the reciprocal of the particle speed
as best estimated by the two components of velocity measurements. The effects
of variable seeding density was estimated using the research of Birch and Dod-
son (1980). For the air jets, this indicated an added uncertainty in the mean and
RMS velocities were below 2.5% and 5% respectively.
Because combustion in a flow, generated by the CGRI burner takes place
throughout the furnace, the gradients in density are very much lower than those
generated by conventional burners. Thus systematic errors due to density gradi-
ents (Ancimer and Fraser, 1994) were expected to be negligible.
96 B.A. FLECK et al.

Because of the combination of strict data comparison criterion (I %) used, the


rate of validated data was low (0.1 to 50 Hz). The number of validated data
points for a given location was between 100 and 500, giving sample durations
which ranged from 10 s and 17 minutes. The low data rate ensured the statistical
independence of each measurement. Since particle arrival times follow a Poisson
distribution (Larsen and Buchhave, 1986), the shortest temporal spacing of data
for 50 Hz data rate should be on the order of 2 ms, which is similar to the largest
integral time scale expected in this flow. The relative uncertainty for each meas-
urement is 3.5%.
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

Numerical Modelling

Numerical modelling of this flow was performed using the commercial software
package TASCflow. This was done on a Cartesian grid of 1.5 x 105 nodes with
grid embedding used to make efficient use of available computer memory. Tur-
bulence modelling was handled using the k - E model.
Combustion was modelled with a simple two-step eddy diffusivity combustion
model with two reactions, based on a scheme suggested by Westbrook and Dryer
(1981),
(2)

(3)

incorporating the five active species CH 4 , 0z, CO, HzO and COz. The reaction
rate was determined by comparing the mixing rate given by the 8DM (Magnus-
sen and Hjertager, 1976) and the rate determined based on chemical kinetics. The
slower of the two was assumed to be rate determining, and thus was used as the
reaction rate. The mixing rate is given by

(4)

where r is the mass ratio of fuel to oxidant for the given reaction (0.333 and 1.75
for the reactions above). The reaction rates determined by the chemical kinetics
(Westbrook and Dryer, 1981) are
s..»: = 6.7 x 1012Te-2535/T[CH4jo.2[02]1.3 (5)

for the first reaction and


R r2,kin = 1014.6Te-2013/TrCOI[H20jO.5[02jo.25 (6)

for the second (temperatures are in Kelvin).


INVESTIGATION OF CGRI BURNER 97

A uniform wall temperature of 800°C was chosen based on refractory tempera-


ture measurements from the experiments. Thermal radiation was modelled
assuming optically dense gas. The coupled multi-grid solver was generally able
to bring residuals down four orders of magnitude to a mean normalized level of -
10-5 .

3 RESULTS AND DISCUSSION

During the initial warm-up phase, pressures inside the furnace tended to fluctuate
significantly. This was due to the unstable nature of the combustion at low tem-
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

peratures. Unlike conventional burners, the combustion process for burners with
direct fuel injection, such as the CGRI burner, is highly dependent on the ambi-
ent furnace temperature. This is because the temperature of entrained furnace
gases largely dictate the temperature of combustion.
However, when the furnace refractory temperatures exceeded about 700°C, the
pressure fluctuations abated and the near field around the burner where combus-
tion takes place showed no indication of a luminous flame, thus operating in the
f1ameless oxidation mode. For these conditions, the only visual evidence that the
burner continued to operate was the evenly distributed glow from the refractory
walls. This stable operation could be maintained for up to 75% turndown.
Figure 4 shows the downstream evolutions of the measured fields of oxygen
and methane content. The figure shows how both the air and fuel jets close to the
burner (x = 125 mm) maintained their separate distinct structure while slightly
further downstream (x = 350 mm), the fuel jets were induced into the core of the
ring of air jets. After this point, the peak methane content remained close to the
burner axis. The slight deviation (in the -y direction) of the combined methane
and air away from the burner axis was due to the fact that the majority of the
entrained furnace gases came from the rear (+Y side of the axis) of the furnace.
The methane and oxygen content profiles (Fig. 5) show the peaks due to the
inlet jets at x = 215. Note that the methane jets are significantly more diluted than
the air jets at this point since their inlet content is 95% methane compared to the
-21 % oxygen for the air. Thus at this point where the fuel and air jets were
essentially segregated, the fuel stream was, to a first approximation, diluted with
7-8 parts furnace gas to every one part of fuel. By the same rough estimate, the
inlet air was diluted to 2 parts primary air to I part furnace gas.
An estimate of the jet dilution prior to their significant interaction may also be
made using the strong-jetlweak-jet model (Grandmaison et al., 1998). Based on
this work, the point of confluence of the air and fuel jets occurred in the vicinity of
the downstream location x = 350, which is in accord with the above results. From
98 B.A. FLECK et al.
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

....
,

....
,
FIGURE 4 Contours of oxygen (a) and methane (b) levels. The contour interval is 1 mole % and 2
mole % (dry basis) for oxygen and methane respectively, for all plots. The x axis is not to scale com-
pared yand z. Distancesin millimetres
INVESTIGATION OF CGRJ BURNER 99

14.---------------------,

x=215
··--0·--. x=350
x=700
.......&....... x=900
o
: .\
.,. ,.
, ,
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

......... _._.~.-.-

qoo 600 900 1200


Y,mm

14

x=215
---- .. ---- x=350
x=700
_._._.-&-._._.- x=900
.a
'0
"#.

., ... .
J! ,
0
E

)(

:z:
u
~
i
.
~

4 ,
~
.
qoo 1200
Y,mm

FIGURE 5 Oxygen (a) and methane (b) content on traverses along the furnace mid plane (z = 475
mm)for sevenxlocations. The burner axis is at y = 785
100 B.A. FLECK et al.

this model, the estimated ratio of the entrained to burner feed gases is on the order
of three. Again, this is in good agreement with the peak oxygen content of 8 mole
%', given that the combustion air accounts for the bulk of the burner mass flow.
The furnace gas temperature and the degree of feed air and fuel dilution prior
to combustion are the parameters used by Wiinning and Wiinning to characterize
the regime of flameless combustion. The estimated dilution ratio of 3 for the cur-
rent investigation was most likely an underestimate since the fuel was clearly
more diluted than the air and because the combined flow continued to entrain
large amounts of furnace gas after the point of jet confluence.
The profiles showing methane content (Fig. 5b) at x = 215 mm and x = 350 mm
indicate a higher methane content on the side where the furnace gas entrainment
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

is highest. This is surprising, but may be explained by the fact that the jet trajec-
tory on the open side of the furnace brings the peak more closely centered on the
vertical position of the probe traverse.
A series of temperature profiles is given in Fig. 6 which gives further insight
into the nature of the burner near field. The figure indicates that a high tempera-
ture zone is present around the pilot flame, and is maintained by the combustion
of fuel and excess oxygen in the entrained furnace gas which is induced into the
centre of the air jet ring. Surrounding this hot zone are the cooler temperature air
jets. The interface between the two zones was characterized by peaks in tempera-
ture fluctuations; since this was not accompanied by peaks in mean temperature,
these fluctuations were due to turbulent mixing, not intermittent combustion.
The air jets on the rear side of the furnace, which were exposed to a crossflow
of entrained furnace gases (y > 785 rnm, right side of figure), were diluted at a
faster rate than those on the front side. This is consistent with the concentration
measurements which indicate the strong entrainment from the furnace rear.
Figure 6 also indicates the entrained gas temperature t ~ 975°C and the low
intensity of temperature fluctuations t ~ 6°C. The mean temperatures of the
gases in the rear of the furnace (y » 785 mm near the rear wall where furnace
gases may escape to the exhaust) were measured directly with a thermocouple
inserted through the furnace roof and were found to be similar to this value. This
value of t ~ 975°C is taken as the exhaust gas temperature. This temperature
is lower than the values estimated by Sobiesiak et al. (1998), I I69°C and
1459°c, for the same combustion heat release and heat sink area. The tempera-
ture values in the previous work, calculated from the furnace energy balances
and the total heat losses through the furnace walls, appeared to be excessive,
even if the differences in measured air preheat temperatures are taken into
• The furnace gas has 2.5 mole % O2 and 20 mole % H20 oa a wet basis. For a dilution ratio of
3:1, one would expect oxygen content on the order of 21 x 0.25 + 2.5 x 0.75, or 7.125 mole %,
- 8.4 mole % on a dry basis.
INVESTIGATION OF CGRI BURNER 101

1000
.---- ..
~ •...•...-.• -'-x:.\OO--- ._.-
960 '., ,. . " 30

. ..... ......
'
.P 800 20
.P•
..: <I-
960 10
~-n----1

800 o
1000 40
~ ...,.-•.-. Ds-':iciii'"' _.-
30

.
.P
..:
20 .P•
<I-
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

960 10

800 o
1100 80
x.460
1000 .--_ ..•... -.......... ..or•• _ _-•.. -_.•. -_. 80

.
.P
..:
40

20
.P
<I-

700 o
1200 100
xa360

..........-•...----_
1100 80

1000 ..... - 80 .p
.P
,..: 900
,; 40<':

800 20

700 o
1<400 150

>=216
100
P P
,..: 50 <..:
500 ~-b':=~~>==,..,.-Jo
zab db ,ioo 1100
...."
'" RMB 01fU:tuIIIoNI

FIGURE 6 Temperature measurements on traverses along the furnace mid plane (z =475 mm), The
burner axis is at y = 785. The squares are for mean temperatures (left scale) and the triangles are for
the RMS of the temperature fluctuations (right scale)

account. In this work, the air preheat temperatures were 2900e and 3200e com-
pared to 332°e and 3800 e by Sobiesiak et aI. (1998). The fuel port angle was
significantly different for the two investigations. It was 15 to 50 degrees smaller
IOZ B.A. FLECK et at.

than in Sobiesiak et al. (1998). Also, the absolute value of the angle separating
adjacent fuel and air jets was only 5° compared to 20° to 55° in Sobiesiak et al.
(1998). One could try to attribute the lower furnace temperature to the smaller
fuel port angle. This is however contradictory to the theoretical postulates of the
strong-jet/weak-jet model presented in (Grandmaison et al., 1998). This model
predicts that, for a larger fuel port angle, flames are more diluted, longer, and
thus of lower temperature.
The measured NOx emissions in this investigation were 6 ppm @3% Oz. This
value is very low; however it is twice that reported by Sobiesiak et al. (1998),
despite the lower exhaust gas and air preheat temperatures and the same excess
air level. It was noted in the previous work that the NO x emissions changed little
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

between fuel port angles of 30° and 42.5°, but increased by 15-25% at a 35°
angle. Wiinning and Wiinning (1997) reported that the NO x emissions for their
FLOX burner which operates in a flameless oxidation regime, ranged from 10 to
100 ppm @3% Oz at a furnace temperature of around IOOO°C. The FLOX
burner, however, delivers all the fuel in a single jet located on the burner axis and
is surrounded by an array of air jets. They did not provide details of the burner
and it is unknown what the air jet angle was. The results of this investigation and
those reported by Sobiesiak et al. (1998) and Wiinning and Wiinning (1997) sug-
gest that in FOOl the fuel jet angle is irrelevant as far as NOxemissions are con-
cerned.
Interestingly, the mean temperatures near the burner axis at x = 900 mm do not
attain the ambient furnace level and the absence of any temperature peak here
imply very low reaction rates. For these steady conditions, the combined jet flow
could not be below the furnace temperature if all the reactants are consumed.
Indeed this is corroborated by the contour plots of carbon monoxide and methane
content at this distance from the burner, shown in Fig. 7. It indicates that a signif-
icant concentration of methane is present at x = 900 mm, spread in a zone which
extends from the floor to the ceiling of the furnace. The peak in carbon monoxide
content occurs at the peak in the methane content. Higher amounts of CO are
measured on the front side of the burner axis (y < 785) due to recirculation after
the combined jet stream encounters the blind wall (the wall directly across from
and facing the burner). Flux measurements for the methane done for similar con-
ditions (Fleck, 1998) indicated that as much as half the feed fuel leaves the near
field unburnt. The CH4 concentration around I% and close to the furnace blind
sidewall were previously reported by Sobiesiak et al. (1998), where 3 m Z of heat
sink was exposed at an air preheat of 299°C and estimated furnace gas tempera-
ture of 970°C. Under the conditions in this work and the coolest furnace by
Sobiesiak et al. (1998), the combustion zone extended to the furnace boundaries
(refractory walls). This conclusion is supported by the contours of the ceiling
INVESTIGATION OF CGR! BURNER 103
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

....-----=:;::==r::=?=.:<......., ... Methane 1%)

... ...
"'"
..."'"
E..
~

...
...
'"
... ... ,.... ... ,... '''l.li
''''

...
...
100

...
E..
~

...
....
'"

FIGURE 7 Contours at x = 900 mm of content of carbon monoxide (a, ppm) and methane (b, mole
%) on a dry basis

refractory temperature shown in Fig. 8. which indicate that the highest internal
surface temperatures in the furnace are in the neighbourhood of the blind wall
opposite to the burner. These high temperatures are either a result of high heat
104 B.A. FLECK et al.

BURNER ROOF

..
E

._._._-_._-_.,
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

I
I
I
I
I
I

4.5
Y,1Nn

FIGURE 8 Contours of the refractory temperatures on the ceiling. Broken line shows where the water
cooled floor (I m 2) is exposed. Temperatures in degrees Celcius

transfer rates as the combined jet impinges on the blind wall, or due to combus-
tion taking place near the hot blind wall and neighbouring ceiling. The latter
hypothesis is supported by previously measured refractory temperatures in the
CAGCT furnace for a conventional burner done by Bindar (l996). In the cited
work, the high temperature gases impinging on the blind wall caused no signifi-
cant hot spots in the refractory.
Measurements of Uz and ax for two downstream locations are shown in Figs.
9a and 9b. At x = 350 rom (Fig. 9a), two velocity peaks are caused by the high
momentum air jets, while further downstream at x = 1250 mm (Fig. 9b), there is
a single high velocity peak where the burner jets merge into a single combined
jet. Also visible is recirculation (u z < 0) on the front side (y < 580 mm at
= =
x 350 mm and y < 260 mm at x 1250 rom) of the burner axis.
Outside the jets the turbulence is essentially isotropic (ax"" az). It is anisotropic
within the jets and there is no downstream tendency to isotropy. This behaviour
is similar to that of a single isothermal jet (Hussein et al., 1993). Note how the
velocity is relatively high near the burner centreline at x = 350 mm (within the
ring of air jets) but the turbulence has magnitude and isotropy similar to the gases
in the surrounding furnace. This observation supports the hypothesis that the
gases within the jet ring are carried there by entrainment which moves from the
INVESTIGATION OF CGRl BURNER 105

40.------------------,
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

20r--------------------,

FIGURE 9 Velocity measurements nn traverses through the furnace mid plane (z =475 mm) at
x=350 rom (a) and x = 1250 mm (b). The burner axis is at y =785
106 B.A. FLECK .1 al.

ambient furnace environment through the gaps between the air jets (where the
fuel jets are) toward the burner axis. This is further illustrated by the numerical
modelling results in Fig. 10 which shows predicted fuel jet trajectories near the
burner face which arc toward the burner axis and merge together there.
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

FIGURE 10 Predicted streaklines in the near field initiated from the fuel jet inlets

The fuel jet entrainment to the burner centreline is also demonstrated in Fig. II
in the contours of vertical velocity ii,. A pronounced upward stream of gas is
measured at y = 785 mm and Z= 375 mm between the two strong air jets (which
have a downward component of velocity due to the air port orientation away
from the burner axis) It is also evident as a downward velocity to either side of
the top air jet (around y =785 mm and z =650 mm). Two of the jets (on the
burner horizontal axis at z = 475 mm) do not appear in the vertical velocity con-
tours because their burner ports do not have a significant vertical orientation.
Note also the vertical entrainment pattern outside the jet ring (at 7 and 11
"o'clock"). This entrainment pattern is asymmetric about the z axis due to the
entrained crossflow from the furnace rear.
Figure 11 also shows contours of mean temperature at this downstream loca-
tion. The lower temperature zones caused by the seven air jets are distinct and
INVESTIGATION OF CGRIBURNER 107

900 . - - - - - - - - - - - - - - - - - - - - - - ,
...••..................1······ ....

•••••• ..- -v., -2 -v-,


.. .....
"",.
.:.... .... -. ':
,
3, "\

v..
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

1100

AGURE II Contours of mean vertical velocity Uz (a) and mean temperature f' (b) at x = 350
mm).The broken velocity contour lines indicate negativevalues
108 B.A. FLECK eI at.

clearly visible. In this and all other temperature measurements, no evidence of


the fuel jets (that might conceivably be observed as low temperature zones near
the burner face) was observed. This is explained by the rapid dilution of the fuel
jets and by the fact that ~nlike the air jets. the fuel gas absorbs significant
amounts of thermal radiation and thus could be expected to rapidly heat up as it
enters the furnace environment.
The efforts to numerically model this system were only partially successful due
to the complexity of this system and the simple radiation and combustion model
available in TASCflow. In general it may be said that the modelling was capable
of accurately simulating the momentum and mass transfer processes but tended
to overpredict reaction rates. This led to model predictions where the fuel was
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

rapidly consumed and a traditional high temperature distinct flame was pre-
dicted.
To compare the numerical and experimental results. maximum values of meth-
ane mass fraction' and the downstream x velocity component were found at a
number of x locations. The experimental data were compared to predictions for
the uniform refractory wall temperatures of 800°C.
Figure 12 shows the comparison of the numerical and experimental results.
The mass fraction predictions for methane are similar to measured values up
until x = 900. From this point. the model predicts a rapid bum-out of the remain-
ing fuel. contrary to the slow rate of combustion which evidently occurs in the
furnace. The model overpredicts downstream velocity, probably owing to the
overpredicted reaction rate which would lead to faster gas expansion and higher
velocities. The measured peak velocity at x = 215 appears below the expected
trend and could be evidence of systematic error in the LDV measurement. possi-
bly due to preferential rejection of the highest velocity measurements that should
form the ensemble average velocity.
It appears that for both the modelled and experimental results, the methane
peaks are not largely affected by the furnace temperature. Thus in the near field,
the consumption of methane is a function of the turbulent mixing field and not
reaction kinetics. A more thorough examination of these results (Fleck, 1998)
indicates that furnace temperature affects the final bum-out of fuel gas outside
the burner near field.

• The mole % dry basis values were converted to mass fraction using the measured content of C~
and assuming the completecombustion of the oaturaJ gas fuel molecule. This gave an estimatefor the
amount of water removed andthe molar mass of the sample.
INVESTIGATION OF CGRI BURNER 109

10° .....- - - - - - - - - - - - - - - - - - - - .

• Gss ssmpllng
CFO

~
u

>-
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

1500

• LOV
CFO

FIGURE 12 Maximum of YCH 4 (a) and u~ (b) as a function of downstream x location (distance
from the burner inlet). Experimental (points) and numerical results (curves)
110 B.A. FLECK et at.

4 SUMMARY AND CONCLUSIONS

The near field and combustion zone of the CGRl burner were characterized
experimentally and by mathematical modelling. The CAGCT furnace conditions
gave the lowest furnace temperature at which the CGRl burner has been tested
and the fuel jet angle the smallest used to date. The research demonstrates that
the FOOl firing strategy can be successfully implemented even at low furnace
temperatures. The FODI technology is exceptionally effective in NO x reduction
by exploiting two characteristics present in most furnaces: the abundance of
cooled furnace gases, and the momentum of the oxidant (and to a lesser degree
that of the fuel) feed stream. Even at the unusually low furnace temperature, the
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

resultant combustion had the peculiar feature of being visually flameless, and so
the term combustion zone is used instead of flame.
The velocity field in the furnace was dominated by high momentum air jets and
the two large recirculation zones they generate. The furnaces gases which were
entrained toward the burner axis transport the fuel there. Combustion close to the
burner face was limited by delayed mixing due to the geometry of the multiple
jets. Further away from the burner (~0.45 m), the air jet ring was broken up by
the entrained crossflow, transforming the ring into a crescent shape, and then to a
single oval jet. At 1.25 m downstream from the burner, the majority of the reac-
tants were fully mixed (at the scale accessible to probing), however the high dilu-
tion by the furnace gases slowed their combustion process. Mass flux
calculations for the 1.25 m location indicated that only less than half of the fuel
was consumed up to this position. For the present conditions, the remaining fuel
burned outside the near field.
The mathematical modelling of the burner was capable of adequately predict-
ing jet trajectories and primary flow structures in the furnace. The predicted tem-
perature and species concentration depart from measured values thus raising a
question of how to effectively numerically simulate, with CFD, three-feed proc-
esses under strong non-adiabatic conditions.
The CGRl burner is a promising technique for natural gas combustion in a
world where emissions regulations continue to grow more stringent. Compared
to conventional designs, this burner and similar technologies using the FOOl
combustion strategy require a somewhat more holistic approach to fur-
nacelburner design and installation, due to the strong interaction of the velocity,
concentration and temperature fields they generate.
INVESTIGATION OF CGRI BURNER III

LIST OF SYMBOLS

E specific dissipation rate of turbulence kinetic energy. m4ts3


p density, kg/m 3
A'b'" B,bu eddy break up model coefficients (part of the eddy diffusivity combustion model).

G momentum flux, N

k specific turbulence kinetic energy, m 21s2


R reaction rate, kg/(m3.s
time, S
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

TgQ.S' Tjunction temperature of measured furnace gas, andof the thermocouple junction. K or QC

r, T mean temperature and RMS of the fluctuations of temperature, K or ·C

Uz 1 11 z components of mean fluid velocity in the x and z directions respectively, mls

Ux , Uz components of RMS velocity in the x and zdirections respectively, mls


X mole fraction
.r, y, z positionsin Cartesian space(wherex is thedirections of the burner axis, y is the
otherhorizontal direction and z is the vertical direction, m
Y mass fraction

Acknowledgements
This research was supported by grants and scholarship from the Natural Sciences
and Engineering Research Council of Canada and British Gas pic.

References
R.J. Ancimer and R.A. Fraser. Flame-induced laser Doppler velocirnetry velocity bias. Meas. Sci.
Technol., 5:83-92,1994.
Yazid Bindar. Experimental and numerical investigations of a multibumer furnace operated with
various heat transfer boundary conditions. PhD thesis, Queen's University, 1996.
A. D. Birch and M. G. Dodson. Some aspects of velocity biasing in turbulent mixing flows resulting
from non-uniform seeding. Optica Acta, 27(1),1980.
Jacques Dugue and Roman Weber. Measurement equipment, LDV probe. Technical Report
C761y11l8, International Flame Research Foundation, Il-rnuiden, The Netherlands, March 1994.
Brian Andrew Fleck. Experimental and Numerical Investigation of the Novel Low NO, CGRJ burner.
PhD thesis, Queen's University, 1998.
E.W. Grandmaison, I. Yimer, H.A. Becker, and Andrzej Sobiesiak. The strong-jet/weak-jet problem
and aerodynamic modelling of the CGRI burner. Combustion and Flame, 114(3-4):381-396,
1998.
M.V. Heitor, S.H. StArner, A.M.K.P. Taylor, and J.H. Whitelaw. Instrumentation/or Flows with Com,
bustion, chapter Velocity, Size and Turbulent Flux Measurements by Laser-Doppler Velocime-
try. Combustion Treatise. Harcourt Brace & Company Publishers, 1993. Edited by A.M.K.P.
Taylor.
112 B.A. FLECK et al.

Hussein J. Hussein, Steven P. Capp, and William K. George. Velocity measurements in a high-Rey-
nolds-number, momentum-conserving, axisymmetric, turbulent, jet. J. Fluid Mech., 258:31-75,
1993.
J. P. Jessee, R. F. Gansman, and W. A. Fiveland. Multi-dimensional analysis of turbulent natural gas
flames using detailed chemical kinetics. Comb. Sci. Tech., 129:113-140, 1997.
P.S. Larsen and P. Buchhave. Flow measurements: why, what and how? Technical report, Dantee
Information, Measurement and Analysis, January 1986. Pan 2.
B. F. Magnussen and B. H. Hjenager. On mathematical modeling of turbulent combustion with spe-
cial emphasis on Soot formation and combustion. In 16 th Symposium (International) of the Com-
bustion Institute, 1976.
D.K. McLaughlin and W.G. Tiederman. Biasing correction for individual realization of laser ane-
mometer measurements in turbulent flows. Phys. Fluids, 16(12):2082-2088, 1973.
I. Nakamachi, K. Yasazawa, T. Miyahara, and T. Nagata. Apparatus or method for carrying out com-
bustion in a furnace. United States Patent, 1990. U,S. Patent Number 4,945,841.
North American Manufacturing Company. Twinbed lIn.. overview. Bulletin 4343, July 1995.
Downloaded by [Hernando Yepes] at 22:57 22 April 2013

Andrzej Sobiesiak, Shahrzad Rahbar, and Henry A. Becker. Performance characteristics of the new
low-NO, CGRI burner for use with high air preheat. Combustion and Flame, 115(1-2):93-125,
1998.
Wayne Nigel Owen Turnbull. Experimental investigation of sample probe effects on gases drawn
from flame zones. Master's thesis, Queen's University, 1995.
Charles K. Westbrook and Frederick L. Dryer. Simplified reaction mechanisms for the oxidation of
hydrocarbon fuels in flames. Comb. Sci. Tech., 27, 1981.
J. A. WOnning and J. G. WOnning. FlameJess oxidation to reduce thermal NO-formation. Prog.
Energy Combusl. Sci .. 23:81-94,1997.

Vous aimerez peut-être aussi