Professional Documents
Culture Documents
www.fuelfirst.com
Abstract
This paper presents an experimental study on the flame properties of O2/CO2 combustion (oxy-fuel combustion) with focus on the
radiation characteristics and the burn-out behaviour. The experiments were carried out in a 100 kWth test unit which facilitates
O2/CO2 combustion with real flue gas recycle. The tests comprise a reference test in air and two O2/CO2 test cases with different recycled
feed gas mixture concentrations of O2 (OF 21 @ 21 vol.% O2, 79 vol.% CO2 and OF 27 @ 27 vol.% O2, 73 vol.% CO2). In-furnace gas
concentration, temperature and total radiation (uni-directional) profiles are presented and discussed. The results show that the fuel burn-
out is delayed for the OF 21 case compared to air-fired conditions as a consequence of reduced temperature levels. Instead, the OF 27
case results in more similar combustion behaviour compared to the reference conditions in terms of in-flame temperature and gas con-
centration levels, but with significantly increased flame radiation intensity. The information obtained from the radiation and temperature
profiles show that the flame emissivity for the OF 21 and OF 27 cases both differ from air-fired conditions. The total emissivity and the
gas emissivity of the OF 27 and the air-fired environment are discussed by means of an available model. The gas emissivity model shows
that the increase in radiation intensity (up to 30%) of the OF 27 flame compared to the air flame can partly, but not solely, be explained
by an increased gas emissivity. Hence, the results show that the OF 27 flame yields a higher radiative contribution from in-flame soot
compared to the air-fired flame in addition to the known contribution from the elevated CO2 partial pressure.
2006 Elsevier Ltd. All rights reserved.
0016-2361/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2006.08.013
K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668 657
Nomenclature
is instead mainly due to the energy penalty in the ASU. To from the surrounding walls in refractory lined furnaces
design, optimize and scale up furnaces in power plants with such as the one used in the present study as well as in
O2/CO2 combustion, models must be validated with exper- the studies given in [12,13]. The wall dominance on hemi-
iments under well defined conditions. spheric radiation measurements is obviously further accen-
Experimental studies on coal-fired O2/CO2 combustion tuated for moderate and non-sooty flames with low
have been performed in laboratory and semi-industrial emissivity of the flame envelope. In order to reduce the
scale test facilities of different size e.g. in [4–7] and more influence of the hot furnace walls on the radiation data,
recently in [8–10]. The experimental work performed has uni-directional measurements of the radiation intensity
to a large extent focused on the emission characteristics have been carried out in this study, which hence improve
for different coal types with an emphasis put on the mech- the possibility for interpretation of the difference between
anisms of NOx formation during flue gas recycling condi- the three test environments investigated. In addition,
tions (e.g. [8,9,11]). The study given in [10] also combines detailed profiles on temperatures and gas concentrations
experimental work with CFD modelling in order to evalu- have been measured to characterize the properties of the
ate different burner and combustor designs. Detailed stud- different flames to facilitate a discussion of the flame emis-
ies on the in-furnace combustion characteristics have been sivities in relation to the radiation profiles. The industrial
performed on semi-industrial scale under natural-gas-fired application of O2/CO2 combustion is most likely to be
conditions [12] as well as under coal-fired conditions [13]. based on pulverized coal plant technology, whereas
Little fundamental data is available on the radiative prop- instead this study presents results performed under gas-
erties of different O2/CO2 mixtures and fuels, but the tests fired conditions. By avoiding solids to take part in the
performed in [12] included measurements of the axial combustion the complexity is reduced in terms of e.g.
hemispheric heat flux distribution, from which it was con- the combustion reaction schemes as well as the radiative
cluded that a feed gas concentration of 28 vol.% O2 during heat transfer mechanisms. Hence, this improves the possi-
O2/CO2 combustion produced similar heat flux character- bility to better isolate the differences between air and O2/
istics compared to air-fired conditions. Profile measure- CO2-fired conditions and to determine the influence of
ments of the hemispheric radiative heat flux are also the CO2 on the gas emissivity versus the total emissivity
reported in [13], but for coal-fired conditions using an with the subsequent effect on the radiative heat transfer
ellipsoidal radiation probe. In general, the dominating in a moderately sooty flame without the dominance of coal
contribution to the local incident radiative flux originates particles and ash.
658 K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668
Direct O2 injection
Primary/secondary
register
FI TI
Pilot burner air/O2/CO2 fan mixing point
O2/flue gas
Air inlet
pre-heater
SC
FI
recycle recycle
FI
FI
Cylindrical
Measurement
furnace
ports R1, R2...R7
SC
2400 mm
Cooling water
For the experiments given in this paper, the oxygen is enables the detection of possible sources to air infiltration
mixed into the recycled flue gas up-stream the ID-fan in during flue gas recycle operation.
the recycle loop. To improve the mixing of the oxygen The gas-fired burner consists of two separate lances for
and the recycled flue gas, the oxygen/flue gas mixture (feed fuel and direct oxygen injection (as mentioned, the oxygen
gas) then has to pass plates that protrude perpendicular register is not used in these tests). The primary and second-
into the tube of the gas recycle tube. After passing through ary swirl registers are replaceable in order to facilitate a
the recycle fan the feed gas is divided into a primary and a change of the swirl number settings and the corresponding
secondary stream which enters the two burner registers. mixing conditions of the flame. Table 2 lists the settings
Approximately 40% of the total flow enters the primary used in the present work together with the overall test con-
register. The ratio between the primary and secondary ditions of the measurements.
stream is adjusted by individual valve settings and the vol- All tests were performed at a stoichiometric ratio of
umetric flows are continuously measured by turbine flow 1.15 resulting in a target O2 excess of 3.0 vol.% for the
meters. Both oxygen and propane is supplied from bundles air and OF 21 cases and 3.8 vol.% for the OF 27 case.
of gas tubes. The oxygen and propane flows are measured The higher O2 excess in the latter case is due to that the
and controlled by mass flow meters. volumetric flow of recycled flue gas is about 20% lower
In order to reduce the time for the combustor to reach than in the OF 21 case. For the OF 27 case the residence
thermal steady state conditions, cooling tubes with variable time is increased by the reduction in volumetric flue gas
length and an outer diameter of 50 mm are inserted from flow and to some extent also by the change in temperature
the top of the combustion chamber at four positions close distribution from air to oxy-fuel fired conditions. From
to the wall, equally distanced from the flame. The temper- initial experiences it was found that the in-flame tempera-
ature and flow rate of the incoming and outgoing cooling ture distributions of the OF 27 and the reference case are
water is continuously measured for heat balance and mod- rather similar, i.e. the difference in temperature will only
elling purposes. Wall temperature conditions are moni- moderately influence the relative change in residence time
tored by means of thermocouples (type K) measuring the from the air to the OF 27 case, whereas this change will be
temperature 20 mm from the inner wall surface at 2 · 7 more pronounced under the OF 21 conditions. All in-
positions at each side of the probe ports R1–R7 (Table furnace measurements have been repeated twice in order
1). The ports along the side of the furnace are used for to ensure satisfactory repeatability for each test con-
in-furnace measurements, by means of water-cooled dition.
probes. A number of ports are available down stream of Gas composition data consisting of O2, CO, CO2 and
the reactor exit and at various positions in the flue gas total hydrocarbon (HC) concentrations were measured
recycle pass, which allow for gas composition measure- by a gas suction probe in the probe ports, at the furnace
ments of the flue gas as well as the recycled feed gas. This exit and in the recycle loop. The total hydrocarbon con-
centration is measured as methane equivalents by means
of an FID (flame ionization detector) based instrument.
The analysis of O2, CO and CO2 were each performed
Table 1
by two different instruments with different measurement
Measurement positions on the Chalmers 100 kW O2/CO2 unit ranges, adapted for reference and O2/CO2 conditions,
Port number Distance from burner (mm)
respectively. The measurement principles of both O2
instruments are of paramagnetic type, CO is analyzed
R1 46
R2 215
based on NDIR (non-dispersive infrared absorption) and
R3 384 CO2 is measured by means of NDIR and TC (thermal
R4 553 conductivity). The suction probe consists of a heated inner
R5 800 tube with controlled temperature to prevent condensation
R6 998 inside the probe and a protecting water-cooled jacket. The
R7 1400
gas is dried and filtered before it is fed into the on-line gas
Table 2
Test conditions for the gas-fired tests
Heat input 80 kWth Fuel C3H8 Stoich ratio 1.15
Primary register Fin angle 45 Swirl No 0.79 Secondary register Fin angle 15 Swirl No 0.21
Test case Combustion media Temp feed gas [C] Feed gas composition (vol.%) Theoretic CO2 conc.
O2 N2 CO2 @ stack (vol.%)
surements in order to find the minimum flow applicable where q1 is the sum of contributions (i)–(iii) and the emis-
to minimize the influence from the purging on the sivity of the flame and combustion gases at a certain wave-
in situ measurement conditions. In the present probe setup length ek is treated as a general function of wavelength k
Argon was used as purge gas at a flow rate of 10 ln/min. whereas the background wall emissivity eb is assumed con-
The calibration of the narrow angle probe was performed stant. The function RkT gives, for a temperature T (K) and
in a black-body furnace before and after each mea- a wavelength interval k to k + dk, the black-body radiation
surement series to check the zero drift of the thermal RkTdk. The temperatures Tg and Ts are the gas and back-
detector. ground wall temperatures, respectively.
Collimating tube
Cooling
water
Fig. 2. Schematic illustration of the narrow angle radiometer (IFRF type) used for the radiation intensity measurements.
K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668 661
For the measurement conditions employed (operational scattering is neglected due to the small size of the soot par-
conditions and measurement levels in the furnace), the fur- ticles and hence the emissivity is assumed equal to the
nace wall temperature typically varies ±50 C (around an absorption. These assumptions result in a flame emissivity
average of 600 C) in all three cases. This results in an emis- independent of wavelength and, integrated over the spec-
sivity of about 0.9 of the refractory lining material. trum, Eq. (7) can be written:
However, with the low conductivity and considerable
q1 ¼ rem T 4mr þ rð1 em ÞT 4s ð8Þ
thickness of the furnace walls the conduction losses
through the lining are negligible at thermal steady state where em represents the mean emissivity and Tmr the aver-
conditions. Furthermore, on the wall surface, the convec- age radiation temperature of the gas layer concerned, i.e.
tive term is small compared with the dominating radiative along the viewing path of the probe. In order to determine
term. Thus, since the heat exchange is in principle only due em and Tmr, separate measurements of the radiation from
to radiation the energy balance in Eqs. (2)–(6) gives that the gas layer rem T 4mr and the background furnace wall
the effective emissivity eb is equal to unity. With Ts being rT 4s in Eq. (8) are required in addition to measurements
the temperature and Qthe total incident heat flux of the of q1, a procedure which is followed in the present work.
wall surface in the energy balance the total outgoing heat Hence, q2 is represented by the first term in Eq. (8). To-
flux Q+ can be written: gether with the net contribution of the background wall,
q3, the mean emissivity and the mean radiation tempera-
Qþ ¼ eb rT 4s þ ð1 eb ÞQ ð2Þ
ture of the gas layer can be calculated according to
According to the assumption above, all heat is exchanged q q2
by means of radiation and hence the rate of emission must em ¼ 1 1 ð9Þ
q
equal the rate of absorption, from which follows: rffiffiffiffiffiffiffiffi 3
q
T mr ¼ 4 2 ð10Þ
eb rT 4s ¼ eb Q ð3Þ em r
Q ¼ rT 4s ð4Þ Obviously, with Tmr and em from Eqs. (9) and (10) simpli-
þ þ
Q ¼ eb rT 4s þ ð1 eb ÞQ ð5Þ fications are introduced compared to real conditions for
which there is a wavelength dependant emissivity, ek, and
Qþ ¼ rT 4s ð6Þ
non-uniform concentration and temperature profiles within
Thus, it is clear that the effective emissivity eb is 1 regardless the measurement volume. The grey approximation is also
of the actual emissivity of the material of the wall surface. less accurate in the case of a gas-fired flame compared to
Therefore, the radiation from background reflection (iii) e.g. a highly luminous oil flame. This since the gas band
can be disregarded, and Eq. (1) can instead be written as radiation will have a stronger impact on the total emissivity
Z 1 Z 1 in a moderately sooty flame and cannot be neglected in
q1 ¼ ek RkT g dk þ ð1 ek ÞRkT s dk ð7Þ comparison to the continuous radiation due to soot and
0 0
especially so for O2/CO2 conditions. This implication on
Eqs. (1) and (7) consider the wavelength dependence of the the results will be further discussed in the results section.
emissivity of the flame/gas layer over which the total radia-
tion measurement is carried out. However, the measure- 3.1.1. Determination of total gas emissivity
ment data do not give information on the spectral As discussed above em will include emission from both
resolution of the radiation. Therefore a simplified treatment gas and soot. Therefore em can be expressed as
of the data is required in order to evaluate the difference in
em ¼ eg þ es Meg es ð11Þ
total emissivity between the air and the O2/CO2-environ-
ment, the latter being represented by the OF 27 case, due where eg and es are the gas and soot emissivities across the
to restricted measurement data for the OF 21 case. Hence, gas layer and the product eges represents the band overlap
for the air and the OF 27 case a method originally proposed of soot and gas. A reduction factor M should be intro-
by Schmidt, as discussed in [15], is used to determine a total duced, which depends mainly on the gas temperature and
mean emissivity of the gas layer, based on three separate to less extent on the soot volume fraction. This is because
measurements of the radiation intensity. Different versions the emission bands of CO2 and H2O are not randomly
of the method have been suggested, such as the two-path located in the spectrum and due to that the monochromatic
method as described in [16], as well as for the calculation soot emissivity increases with decreasing wavelength. A
of in-flame soot volume fraction in highly soot laden flames possible change in band overlap between soot and gas will
[17]. The latter based on the fact that the method is always however not be treated in this work, due to its small effect
exact monochromatically. In this work, however, the on the overall emissivity. There are no experimental data
method is used in its original form. The general basis of available on the soot volume fraction from the flames
the method is that the flame or gas layer is considered grey, investigated in this work. Instead, in order to reveal if the
so that the emissivity becomes independent of wavelength difference in mean emissivity between the two combustion
and equal to the absorptivity, independent of the relation environments corresponds to the difference in gas emissiv-
between the gas and the background temperature. The ity, including H2O and CO2, the gas emissivity is deter-
662 K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668
mined using the model by Leckner [14] running a Fortran and, finally, the total absorptivity of the gas mixture can be
routine provided in [18]. This model was chosen since it calculated from:
accommodates a high degree of generality for which it
Ts Ts
has been recognised (cf. e.g. [19]). This, since the CO2 ag ¼ aCO2 þ aH2 O De pH2 O L ; pCO2 L ð18Þ
and H2O emissivities are treated separately after which Tg Tg
the total emissivity of the mixture is determined by taking
a total band overlap correction term into consideration.
Hence, any arbitrary partial pressures of CO2 and H2O 3.2. Results and discussion
can be applied in the model. The polynomial correlations
of the model produce a maximum error of 5% for water va- Satisfactory experimental conditions were typically
pour and 10% for CO2 emissivities at temperatures above reached after some four to five hours with a CO2 concen-
400 K, compared to the spectral integration data upon tration level in the stack gas between 94 and 95 vol.%.
which the model is based [14]. The total emissivity for a Any discrepancy between the theoretical CO2 concentra-
mixture of CO2 and H2O is obtained from: tion level (96–97 vol.%) and the ones measured is due to
that N2 is still left in the recycling media and, to some
eg ¼ eCO2 þ eH2 O DeðCO2 þH2 OÞ ð12Þ
extent also due to limitations in the instrument precision
In the model a zero partial pressure emissivity is given by of the CO2 analyzer which has a maximum error of ±2%
the correlation: of the full measurement range (0–100%).
" j i # Radial temperature profiles are plotted in Fig. 3 for the
X M X N
Tg pa L
e0 ðpa L;p ¼ 1 bar;T g Þ ¼ exp cji log10 three different cases: air, OF 21 and OF 27 (the radiation
i¼0 j¼0
T0 ðpa LÞ0 intensity profiles shown in the same figure will be discussed
later on). It has been well established experimentally that
ð13Þ
O2/CO2 conditions with a similar volumetric feed gas con-
where Tg is the gas temperature, T0 = 1000 K, p is the total centration of O2 as for air (i.e. 21 vol.%), yields signifi-
pressure, pa is the partial pressure and L is the path length cantly lower peak as well as overall combustion
of the gas, either CO2 or H2O, (paL)0 = 1 bar cm and Cji temperature levels compared to air-fired conditions (see
are correlation constants as given in [18]. The emissivity e.g. [8,12]). This is also the general impression obtained
for higher partial pressures, i.e. if pa > 0, can then be calcu- from a comparison of the OF 21 case and the air case of
lated according to: this work (cf. Fig. 3). For example, the highest measured
combustion temperatures for the OF 21 and the air-fired
eðpa L; p; T g Þ ða 1Þð1 P E Þ
¼1 case in port R2 (Fig. 3a) are approximately 1370 C and
e0 ðpa L; p ¼ 1 bar; T g Þ a þ b 1 þ PE 1550 C, respectively. The lower temperature in the OF
2 ! 21 case is mainly a result from that CO2 has a higher spe-
ðpa LÞm
exp c log10 ð14Þ cific heat capacity than N2 and to a certain extent that there
pa L
is an increase in radiation losses caused by the CO2 which
where a, b, c and (paL)m are correlation parameters and PE increases the gas emissivity. The significance of the latter
is an equivalent pressure, all given in [18]. For calculation effect obviously depends on the radiative properties of the
of the total gas emissivity the last term in Eq. (12) needs flame considered, i.e. to what extent other components
to be considered, which is due to the band overlap of the such as moisture, soot and particles contribute to the radi-
two gases. This correction term is written ative heat transfer. The emissivities of the O2/CO2 and the
air-fired environments are further discussed below, and
f 10:4 related to the radiation intensity profiles.
DeðCO2 þH2 OÞ ¼ 0; 0089f
10:7 þ 101f A reduction of the recycled flue gas is an obvious way to
2:76
ðpCO2 þ pH2 O ÞL increase reactor temperature levels in order to improve the
log10 ð15Þ combustion intensity, in this case from the OF 21 to the OF
ðpa LÞ0
27 case. As a result (Fig. 3) the temperature levels of the
with the partial pressure ratio f defined as OF 27 case approach those of the air case in contrast to
p H2 O the OF 21 case which yields significantly lower tempera-
f¼ ð16Þ tures compared to air firing. Both the air and the OF 27
pH2 O þ pCO2
cases exhibit a strong negative gradient in temperature
The absorptivity is determined by multiplying a factor between 50 and 100 mm from the furnace centreline, which
(Tg/Ts)0.5 to the emissivity evaluated at the external surface represents the location of the flame front. In the OF 21 case
temperature Ts and a temperature scaled pressure path the temperature starts to decrease already between the
length according to: centreline and 50 mm away from the centreline (0 mm),
0:5 and no parabolic shape of the temperature profile can be
Tg Ts seen close to the centreline (the distributions are assumed
aðpa L; p; T g ; T s Þ ¼ e pa L ; p; T s ð17Þ
Ts Tg axi-symmetric). Although a weak trend of such a parabolic
K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668 663
45 1600 45 1600
Measurement port: R2 Measurement port: R3
1500 1500
40 40
Radiation intensity q [kW/m sr]
35 35
30 1200 30 1200
1100 1100
25 25
1000 1000
20 900 20 900
800 800
15 15
700 700
10 10
600 600
5 500 5 500
0 100 200 300 400 0 100 200 300 400
Distance from centreline [mm] Distance from centreline [mm]
Air: q1
OF 21: q1
OF 27: q1
Air: Tg
OF 21: Tg
OF 27: Tg
Fig. 3. Mean values of the measured radial radiation intensity profiles with hot refractory furnace wall as background (q1). The grey dotted profiles show
the corresponding gas temperatures.
temperature distribution may exist close to the centreline, tions, have been reported by Kimura et al. [6] and Liu
the tendencies in ports R2 and R3 clearly show a shift in et al. [8]. It should be noted that the gas concentration
position of the flame front (Fig. 3a and b). In addition, and temperature data from intrusive measurement instru-
as seen from the measurements in Port R2 (Fig. 3a), ments such as the ones used in this work rather represent
already between 50 and 100 mm the temperature of the average values over a certain measurement volume than a
OF 21 case decreases down to a level of about 700 C, value at an exact position. This applies to the gas suction
which corresponds to the gas temperature near the wall probes and suction pyrometers used here, especially the lat-
(400 mm) of both the OF 27 and the air-fired case. This ter probe due to the rather high suction velocity employed.
shows that for the OF 21 case, the physical volume of This is especially important to keep in mind for regions
the high temperature zone is smaller compared to the two with sharp gradients. For example, this is the case when
other cases. As will be discussed in the following section, comparing the HC profiles in Fig. 4a and the temperature
this influences the burn-out behaviour of the OF 21 case. profiles in Fig. 3a. The high concentration of unburned fuel
Radial gas concentration profiles (O2, CO, CO2 and at 50 mm from the centreline can be explained by the fact
HC) for ports 2 and 3 are given in Fig. 4, since these two that it corresponds to temperatures represented within a
ports represent the most intense combustion zone with distance from 50 to 100 mm from the centreline, i.e. some-
the highest measured temperatures for all cases. In Fig. 4 where in the range between 700 and 1200 C, rather than
it is seen that the gas composition of the flame envelope the exact same position as the temperature measurement
of the OF 21 case is clearly different from the two other at 50 mm. Compared to the OF 21 case the fuel burn-out
cases. The amount of unburned hydrocarbons is signifi- of the OF 27 case is favoured by the increased temperature
cantly higher for the OF 21 case and, as a consequence, level and improved mixing conditions between fuel and O2,
O2 is still present in the flame core as shown in Fig. 4a due to the reduced amount of recycled flue gas. The hydro-
and b. The suppressed development of the OF 21 flame is carbon levels are low and the oxygen profile at the centre-
further seen from the CO profiles (Fig. 4c and d) which give line exhibit a similar distinct zero-level as for air-fired
significantly lower concentration levels compared to the air conditions.
and OF 27 cases. The decline in combustion intensity of the The total intensity signal in Fig. 3 corresponds to q1 in
OF 21 case can to a large extent be attributed to the Eq. (8) with the furnace wall emissivity equal to unity.
reduced temperature levels compared to the other two The q1-profiles shown in Fig. 3 are a result from the probe
cases, as shown previously in Fig. 3. Similar results with being traversed through the combustion chamber and
poor combustion intensity and ignition stability in tests directed towards the opposite hot refractory lined furnace
with OF 21 feed gas mixtures, but under coal-fired condi- wall (not shown in the figure). Fig. 3a shows that the max-
664 K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668
a 14 10
b 7 3
Meas. port: R2
Air: O2 9
12 6
OF 21: O2 2.5
8
OF 27: O2
O2 concentration [vol %]
O2 concentration [vol %]
10 Air: HC 5
HC concentration [vol %]
7
OF 21: HC
HC concentration [vol%]
2
OF 27: HC
8 6 4
5 1.5
6 3
Meas. port: R3
4 Air: O2
2 OF 21: O2 1
4 3
OF 27: O2
Air: HC
2 1
2 OF 21: HC 0.5
OF 27: HC
1
0
0
0 0
0 100 200 300 400 0 100 200 300 400
Distance from centreline [mm] Distance from centreline [mm]
c d 8
14 95 95
7
Air: CO 90 6
CO concentration [vol %]
Air: CO 90
CO concentration [vol%]
6 3
10 10
4 2
5 1 5
2
0
0
0 0
0 100 200 300 400 0 100 200 300 400
Distance from centreline [mm] Distance from centreline [mm]
Fig. 4. Measured radial O2 and HC (a and b) and CO and CO2(c and d) mean concentration profiles at positions R2 and R3, respectively.
imum measured radiation intensity 100 mm from the centr- increase in CO2 partial pressure (approximately eight times
eline is recorded close to the flame boundary as indicated the pressure level of normal air-fired conditions), from
by the temperature profiles. Moving towards the probe which an increase in gas band radiation is anticipated.
wall to the right the intensity drops, as the absorption The difference in total emissivity and in gas emissivity is
becomes higher than the emission. therefore compared between the O2/CO2 and air cases in
The information from the temperature distributions order to evaluate if the increase in overall emissivity can
shown previously together with the radial profiles of radia- be attributed to an increase in CO2 band radiation only.
tion shows that the emissivities of both O2/CO2 environ- The O2/CO2 environment is represented by the OF 27 case;
ments differ from those of reference conditions in air. As since the experimental radiation data is restricted for the
seen the radiation intensity is higher in the OF 27 case than OF 21 case (i.e. no measurement data is available on q2).
in the air case despite the fact that the radial temperature It should be noted that significant differences in emissivity
levels are generally lower or similar to those of air. Further- may exist also between OF 21 and OF 27 conditions since
more, it can also be seen that in spite of the overall lower the soot volume fraction may change from one O2/CO2
temperature levels of the OF 21 case compared to the air atmosphere to another. In fact, by comparing the shape
case, the OF 21 radiation intensity profiles approaches of the radiation intensity profiles in Fig. 3a it is seen that
those of the air case. The obvious physical difference the peak intensity 100 mm from the centreline is clearly
between the air case and the O2/CO2 conditions is the accentuated for the OF 27 condition, which is not the case
K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668 665
a 650 900
850
625
Wall temperature [oC]
750
575
700
Air, TBR
550
Air, TBL Air, Tgmin
650
OF 21, TBR Air, Tgmax
OF 21, T OF 27, Tgmin
525 BL
OF 27, Tgmax
OF 27, TBR 600
OF 21, Tgmin
OF 27, TBL OF 21, Tgmax
500 550
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400
Axial distance from burner inlet [mm] Axial distance from burner inlet [mm]
Fig. 5. (a) Axial distribution of the furnace wall temperature, 20 mm from the inner wall surface at the right (TBR) and the left (TBL) hand side of the
measurement ports. (b) Axial distribution of the minimum (Tgmin) and maximum (Tgmax) gas temperatures at the inner surface of the furnace wall.
in the two other test environments. This may indicate local suction pyrometer. The furnace wall temperatures are mea-
differences in emissivity between the tests conditions, which sured at the right (TBR) and left hand side (TBL) of the
could be caused by changes in soot volume fraction. probe ports on the same axial distances from the burner
In order to determine the total mean emissivity, the as shown in Table 1. As seen, the distribution of gas and
background wall temperature needs to be determined. wall temperatures correlate well for all three cases with
Fig. 5 shows the furnace wall temperatures for the three the highest gas and wall temperatures obtained in the OF
cases, measured with thermocouples at a distance 20 mm 27 case at a distance of 800 mm from the burner, where
from the inner furnace wall surface (a) together with the the temperature drops faster than for both air and OF 21.
measured minimum and maximum gas temperature (b), Fig. 6a shows a comparison of radiation intensity pro-
measured close to the inner furnace wall surface with the files for air and OF 27 conditions in port R3, measured
OF 27 conditions - Ts [oC]
a b 660 670 680 690 700
45 0.7 1050
Meas Port: R3 Measurement port: R3
Air - q1 Air: ε m
40 Mean radiation temperature, Tmr = f(Ts)
OF 27 - q1 OF 27: ε m
Radiation intensity q [kW/m2sr]
35 OF 27 - q2 0.6 Air:Tmr
30
950
Mean emissivity
25 0.5
900
20
15 0.4
850
10
5 0.3 800
0 100 200 300 400 630 640 650 660 670 680
Distance from centreline [mm] Air conditions - Ts [oC]
Fig. 6. (a) Comparison of radiation intensity profiles, air and OF 27 conditions, measured against both cold black-body target (q2) and hot refractory
furnace wall (q1) in port R3. (b) Tmr and em based on the measurements given in Fig. 6a and calculated as a function of different furnace wall temperatures
(Air: 655 ± 20 C and OF 27: 680 ± 20 C).
666 K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668
a 0.36
Test case: Air
b 0.36
Test case: OF 27
ε CO2 ε CO2
0.32 0.32
ε H 2O ε H 2O
εg εg
Gas emissivity ε=f(Tg)
0.24 0.24
0.2 0.2
0.16 0.16
0.12 0.12
0.08 0.08
700 800 900 1000 1100 1200 1300 700 800 900 1000 1100 1200 1300
c 0.42
Test case: Air
d 0.42
Test case: OF 27
Total gas emissivity and absorptivity εg /α g
0.4
εg Total gas emissivity and absorptivity ε /α 0.4 εg
0.38 0.38 α g hot BBT
α g hot BBT
0.36 0.36 α g cold BBT
α g cold BBT
0.34 0.34
0.32 0.32
0.3 0.3
0.28 0.28
0.26 0.26
0.24 0.24
0.22 0.22
0.2 0.2
0.18 0.18
0.16 0.16
700 800 900 1000 1100 1200 1300 700 800 900 1000 1100 1200 1300
Fig. 7. The gas emissivity given as a function of average gas temperature of the furnace cross section for (a) air-fired conditions and (b) OF 27 conditions
with the hot furnace wall as background. The gas emissivity and absorptivity as a function of average gas temperature for (c) air-fired conditions and (d)
OF 27 conditions with both hot and cold black-body targets (BBT) as backgrounds.
Table 4
Input parameters, partial pressure (pa), background surface radiation equal to the furnace wall temperature or to that of a cold
temperature (Ts) and path length (L) to the gas emissivity model black-body target. The difference increases for the OF 27
Test case pCO2 pH2O Ts (K) L (m) case with a maximum difference of about 25% between
Air 0.10 0.14 928 0.80 emissivity and absorptivity for the corresponding gas and
OF 27 0.78 0.18 953 0.80 background temperatures, which to some extent affects
the calculation accuracy of the total mean emissivity, since
the combusting flow is not a prefect grey gas as assumed.
increases with approximately 20% in port R3 (Fig. 3b) Yet, for the mean radiation temperatures and total mean
compared to the air case despite similar temperature emissivities calculated (cf. Fig. 6b) the relative error is sig-
profiles. nificantly reduced and the results obtained from Eqs. (9)
Fig. 7c shows that the maximum difference between the and (10) show good agreement with the measurement data
total gas emissivity and absorptivity for the reference con- shown in Figs. 3 and 6. Hence, a comparison of the mean
dition in air is approximately 10% when the gas tempera- emissivity as obtained from Eqs. (9) and (10), in relation to
ture varies between 700 C and 1300 C with two different the total gas emissivities determined according to Eqs.
background temperatures, i.e. the background temperature (12)–(18) shows that the significant increase in radiation
668 K. Andersson, F. Johnsson / Fuel 86 (2007) 656–668