вход по аккаунту


The Influence of Changes to Mixing on the Sooting and NOx Emission Characteristics of Unconfined Turbulent Jet Diffusion Flames.

код для вставкиСкачать
Dev.Chem. Eng. Mineral Process., 7(3/4),pp.361-374,1999.
The Influence of Changes to Mixing on the
Sooting and NOx Emission Characteristics
of Unconfined Turbulent Jet Diffusion
G.J.R. Newbold* and G.J. Nathan
Department of Mechanical Engineering, The University of Adelaide,
Adelaide, SA 5005, Australia
Results are presented fiom new measurements of verticallyfired unconfined d i m i o n
flames issuingfiom precessing jet nozzles and new calculations of existing data fiom
simple jet jlames. These results provide quantitative information about large-scale
mixing, NOx emissions and sooting tendencies of precessing jet flames. It is shown
that a characteristic large e&y strain rate at the tip of a precessing jet $ame is
reduced by an order of magnitude when compared to that in simple jet flames. The
reduction in large e m strain rate correlates with an increased sooting tenaenCy that
is consistent withfindings for simplejet flames presented elsewhere. The reduction in
strain rate correlates with decreased NOx emissions and is found to be contrary to
flamelet model predictionsfor simplejet flames.
Precessing jet flows and flames are the subject of ongoing research at the University
of Adelaide [ 13 because of the benefits that they offer to industries employing high
temperature gas flames. These benefits include improved flame stability, lower NOx
*Authorfor correspondence.
G.J.R. Newbold and G.J. Nathan
Figure 1. Schematic diagram of the fluidic precessing jet nozzle. The solid lines
represent a simplified description of the instantaneous streamlines of the precessing
(rotating)flow field and the dotted lines indicate the azimuthal precessing motion.
Combustion occurs downstream from the nozzle exit.
emissions and increased productivity and can be derived solely from changes in the
mixing [2,3].
Precession is a term used to describe the motion of the jet exiting the nozzle about
the nozzle axis (Fig. 1). Details of the naturally occurring fluid dynamic phenomenon
and the nozzle geometry that causes jet precession may be found in Nathan et al. [4].
When employed as a non-premixed gas burner the fluidic precessing jet (FPJ) burner
produces a lifted flame with low NOx emissions [5].
Changes to flame chemistry resulting from changes to mixing by jet precession are
investigated here. Mixing is considered in relation to global trends in NOx emissions
and soot formation in precessing jet flames. Comparison is made with published
empirical relations and models derived for simple turbulent jet diffusion flames.
Mixing, Sooting and NOx Emission Characteristics of Precessing
Jet Flames
Newbold et al. [6] presented a study of vertically fired unconfined diffusion flames
from fluidic precessing jet (FPJ)burners firing commercial grade propane. It has been
demonstrated that the far field of FPJ flames is dominated by the entrainment into, and
Characteristics of unconfined turbulent jet d i m i o n flames
S e a l e d Ball Bearings
Deflected N o z z l e
Stationary Sleeve
Figure 2. Details of the mechanical precessing jet nozzle.
mixing within, discrete large-scale buoyant structures, in a manner analogous to
simple jet flames [7] and pool fires [8]. The dynamic motions of the large-scale
structures causes air to be entrained into those structures and to mix with the fuel and
combustion products due to a buoyant instability attributed to fluid mechanics external
to the flame envelope. That study determined the size and number of flame structures,
the celerity of those structures, flame dimensions and residence times. From these
results it was demonstrated that the large-scale structures are characteristic of buoyant
“puffs” that form above pool fires. It was demonstrated, but was not quantified, that
the strain rate throughout the bulk of the precessing jet flame is reduced compared
with simple jet flames.
Whilst the efficacy of the FPJ nozzle has been proven in practise, it produces a
flow field which is extraordinarily difficult to study at the fundamental level.
primarily this is because it is unstable, and it is this very instability which appears to
be the source of the efficacy. The instability manifests itself in several forms, the most
apparent of which is the precession of the emergent jet. As a step toward dissecting
the naturally occurring precessing jet flow into its component features, a fundamental
investigation of the effects of precession on a jet have been undertaken by
mechanically rotating about its axis a nozzle from which emerges an inclined round jet
with its exit on the axis of the rotating nozzle (Fig 2). Unlike the FPJ,the jet which
emerges from the mechanical precessing jet (MPJ) nozzle has well defined conditions
G.J.R. Newbold and G.J. Nathan
at its origin that can be varied independently: viz, the exit diameter (4,velocity (UO)
and angle ( 0 ), and the precessional frequency (fp).
This flexibility allows the
controlled study of the separate and combined effects of these parameters.
Measurements of the cold flow mixing characteristics of a MPJ nozzle [9,10]have
highlighted the significance of a dimensionless Srrouhal number of precession based
on the mean exit conditions, s t d =fpd/uo. When the nozzle is rotated at a low Strouhal
number (Sr, = 0.001) and with 6 = 45" the path of the jet deviates indiscernibly from
the direction at which it leaves the nozzle and the jet structures resemble those in a
fully pulsed jet.
As the frequency is increased, strong local pressure fields are
established in the region near to the nozzle in which the jet assumes a helical path.
Within some ten nozzle diameters, the helix collapses and the jet width is an order of
magnitude greater than d.
Downstream from the helix region the original jet
completely loses its identity and mixing takes place on the scale of the helix diameter
In the high Szd regime (Sr, > 0.01) the combustion occurs within and
downstream from the collapsed spiral region and the flames have comparable
combustion characteristics to FPJ flames [12]. Jet precession in the (MPJ nozzle)
high Srd regime results in an increased rate of decay of the mean velocity by an order
of magnitude, relative to that in a non-precessing jet in the first 10 nozzle diameters,
as indicated above. This is consistent with the increased rate at which the jet spreads,
and with the finding that the Reynolds stresses are increased relative to the nonprecessing turbulent jet in the near nozzle region [9].
Species emissions, flame shape and radiant fractions have been measured for
unconfined flames of methane and propane issuing from a vertically fired unconfined
MPJ nozzle [13,14].
The wide range of differing types of mixing that can be
produced by a MPJ nozzle correspond to quite different radiant fractions and species
emissions. The low Std flames (Fig 3a) tend to be short and blue, in some methane
cases, with high CO, have a high ratio of NOz to NO, while the high Srd flames (Fig
3b), which are broader and orange, tend to have lower NOx emission indices than do
simple jet flames. They found that global characteristics of FPJ flames (Fig 3c) are
closely matched to those from high
MPJ flames. High
MPJ flames show an
Characteristics of unconfined turbulentjet diffusionJames
G.J.R. Newbold and G.J.Nathan
increase of about 15% in the radiant fraction and a reduction of about 25% in the NOx
emissions relative to a simplejet flame.
Mean Strain Rate in Precessing Jet Flames
Further insight into the effect of precession on the combustion characteristics of a jet
flame can be determined by comparison of the characteristicbulk strain rates in simple
and FPJ flames. For jet flows whose turbulence characteristics are generally similar,
jet turbulence can be characterised by the exit strain rate, udd. However other bulk
values are more relevant when comparing simple jet flames with precessing jet flames
since the mixing characteristics are not similar [ 151.
Table 1 shows the strain rate at the jet throat, udd, for both FPJ flames
investigated by Newbold et al. [6] and the flames investigated by Mungal et al. [7].
The FPJ burner is characterised by its chamber diameter, D, and flow conditions at the
upstream throat where the flow is jet-like, 4 and d. Table 1 shows that the D = 21mm
FPJ flames span a similar range of Reynolds number and strain rate at the jet throat to
that of the simple jet flames. The jet flame buoyancy parameter developed by Mungal
er al.
[7],4, provides a criterion by which the effects of buoyancy can be quantified
in simple jet flames. A value of
dominated flame and
c 2 is considered to indicate a momentum
& > 10 is considered to indicate a buoyancy dominated flame.
It is apparent that the simple jet diffusion flames investigated range from momentum
dominated to buoyancy dominated.
The characteristic strain rate at the tip of vertical unconfined simple jet flames can
be calculated from the characteristic vertical velocity divided by the flame diameter at
the tip of the flame [16]. R ~ k k eet al. [16] estimated a strain rate at the flame tip
based on the nozzle exit conditions and used empirical relations to obtain values for
the variables at the flame tip. If the characteristic vertical velocity is taken instead as
the celerity of the large-scale structures in the flames, S, then a large eddy strain rate
can be calculated from measured data for the FPJ flames and compared directly with
comparable data in free turbulent jet diffusion flames from non-precessing jets, using
the findings of Mungal et al. 171.
Characteristics of unconfined turbulent jet diffusionflames
Mungal er al. [7] found that the celerity of the burning structures remains constant
along the majority of the flame length with a value of S
0 . 1 2 ~and
~ to be
approximately independent of fuel type (ethylene and acetylene). This implies that S
is lower for buoyancy dominated flames than for momentum dominated flames, which
is to be expected. The gas velocities at the visible flame tip, uxAW, were found to
vary 0.084IU X ~ V EI 0.25u0, so that S is a reasonable estimate of the gas velocity at
the flame tip. They demonstrate that the fluctuation in flame length, AX, resulting
from the burnout of successive large-scale structures, scales with the large-scales in
the flow, namely UXAVE and the jet diameter at the flame tip, 1. Thus, the length-scale
of fluctuations in simple jet flames, AX, was determined to scale in direct proportion
with 1 at the mean position of the flame tip, determined from the relation 1 = 0.44XAm.
Consequently, the large eddy strain rate for a simple jet flame can readily be
determined from the relation S/AX. They have shown that the flame length asymptotes
to XAVE= 230d for momentum-dominated flames. Thus, these relations show that S is
linearly proportional to UO,and AX is linearly proportional to d, so that S/AX will scale
directly with udd for momentum-dominated simple jet flames. Consequently, trends
in the jet throat strain rate scale in proportion with the large eddy strain rate, and these
terms can be treated as interchangeable.
The celerity, S, and fluctuation length, AX, for each of the FPJ flames has been
determined [ 6 ] . It was shown that S is not strongly dependent upon burner diameter or
flow rate at the nozzle throat for the FPJ flames and its measured values are of the
order of S = 2 d s . The celerity of simple jet flames [7] was found to increase with
exit velocity over the range 2 IS I30ds so that S is consistent with results for
buoyancy-dominatedjet flames. The fluctuations in the FPJ flame length are roughly
a constant proportion of the flame length and do not change significantly with exit
velocity. Relative fluctuations in flame length, MA^, in FPJ flames are roughly
twice that in simplejet flames.
Table 1 shows that the large eddy strain rate in all of the FPJ flames examined here
have a value of about S/AX = 5s-' which is comparable with that calculated for the
(d = 4.4mt1)
D =: 21mm
(d = 2.3mm)
Q (kw)
45 400
73 300
100 000
140 900
27 300
38 OOO
49 400
61 400
74 200
87 600
101 500
5 700
21 400
43 500
5 380
9 830
19 600
39 400
78 800
86 500
37 OoO
59 500
81 500
115 000
6 100
8 400
11 000
13 700
16 500
19 500
22 500
5 300
10 OOO
20 OOO
40 000
5 300
10 OOO
20 OOO
40 000
76 700
83 300
udd (5')
Jet Throat
Strain Rate
5 .O
Flame Tip
Strain Rate
S I M (s-')
Table I . Flame parameters. Data for FPJ flamesfrom Newbold et al. 161. Data for jet flames from Mungal et al. [7].
Characteristics of unconfined turbulent jet dimion James
buoyancy-dominated jet flames. The value of large eddy strain rate in the FPJ flames
is independent of burner throughput in contrast to simple jet flames. These results are
consistent with the finding that the large-scale structures in FPJ flames are buoyancy
driven for the flames presented [6]. However, the large eddy strain rate for the simple
jet flames increases with exit velocity so that the buoyancydominated flames have a
large eddy strain rate that is an order of magnitude higher than that of FPJ flames.
These trends, taken together, are unambiguous and clearly demonstrate that the strain
rates at the flame tip produced by jet precession are lower than those in momentum
dominated non-precessing flames.
Note that this type of dimensionless comparison based on a direct measure of
actual flame properties also avoids any problems associated with attempting a direct
comparison between the exit flows produced by a FPJ (which are not well defined)
and those of a simple jet.
The Relationship Between Soot Formation and Strain Rate in
Precessing Jet Flames
The effect of mixing parameters on soot volume fraction and flame temperature in
acetylene turbulent jet diffusion flames have been investigated by Kent and Bastin
[17]. Soot volume fraction profiles were found to scale with respect to a characteristic
flame mixing time, d u o . This mixing time is the inverse of the jet throat strain rate.
That is, low strain rate correlates with high soot volume fraction for simple jet flames.
The large eddy strain rate for an FF'J flame is lower than in a simple jet flame and
is consistent with the high sooting tendency for FPJ flames [a] and also for high Std
MPJ flames [13].
Nathan et al. [ 131 present images of a low Sfd and a high s t d MPJ flame With the
same ~0 and d, and therefore constant jet throat strain rate, udd. However their
images visually demonstrate a trend in sooting characters and radiant fractions that
can only be attributed to the effect of jet precession on mixing, suggesting that the
characteristic strain rate is altered by precession for the constant udd and increasing
The low s f d flames have lower radiant fractions and tend to be dominated by gas
G.J.R. NewboM and G.J. Nathan
chromatic radiation ([13] and Fig 3a) while the high Srd flames are bright orange due
to soot luminescence ([13] and Fig 3b). The change in sooting character found in
MPJ flames correlates with the trends of reduced mixing rate trends for soot formation
in free turbulent jet diffusion flames.
This implies that combustion occurs
predominantly in a region characterised by increased strain in low s t d flames while it
predominantly occurs in a region characterised by reduced strain in high Srd flames.
Thus it is concluded that precession of a jet at high Std produces a flow in which the
strain rate in the main body of the flame is reduced, resulting in increased sooting
tendency, and that the strain rate at the jet throat (and hence also the jet exit) is not
representative of the local strain rate within a precessingjet flame.
The Relationship Between NOx Emissions and Strain Rate in
Precessing Jet Flames
The decrease in the NOx emissions which result from high Sr, mixing in MPJ flames
has been related to changes in the global residence time and characteristic flame
temperature of the gases in the flame [ 13,141. The flame temperature was calculated
based on flame volume and measured radiant fractions. The trend for MPJ flames is
consistent with similar trends measured in free turbulent jet diffusion flames [18,19] in
which the radiant fraction changed with the differing sooting propensity of different
fuels. The influence of flame radiation on NOx emissions can also be seen in the
work of Boerstoel er al. [20,21]. They perform a numerical simulation of combustion
in a glass furnace and show that flue NOx emissions are reduced by 40% by including
soot formation in their combustion model and better agree with furnace measurements.
This demonstrates that soot has a clear effect on flame temperature and thus on NOx
Rakke et al. [16,22] have characterised the NOx emissions from unconfined
turbulent jet diffusion flames of methane, propane and natural gas and partly premixed
flames of propane from a strained flamelet model of combustion. Their model
postulates that radiation effects on flame temperature are less significant than the
influences of finite rate chemical kinetics. This model relates the external strain rate
3 70
Characteristics of unconjined turbulentjet di&ion
imposed on the flamelets by the flow field to the characteristic strain rate at the tip of
the flame, and used empirical relations to obtain values for the variables at the flame
tip from flow variables at the jet throat. They predict that NOx elmissions scale as
with Fr being the Froude number (u?/g&> and Y, being mass fraction of the fuel. In
this model
can be considered to be the inverse of a modified exit strain rate.
That is, decreased strain rate and the subsequent reduction of the scalar dissipation
acts to increase the importance of the Zeldovich mechanism relative to the prompt
mechanism by increasing temperatures and to increase NOx formation by making the
reaction zone thinner, resulting in a higher flame temperature. Their prediction was
shown to agree well With measurements for partly premixed flames.
As deduced from the large-scale flame characteristics, the characteristic strain rate
at the tip of FPJ flames is lower than in non-precessing jet flames. It is noted that the
reduced strain rate also deduced to occur in high St,MPJ flames relative to the simple
jet flames correlates with reduced NOx emissions shown in Figure 4 and increased
emissivity [13]. This trend is the opposite of that found in simple jet flames [16,22].
An explanation of this seeming contradiction may be found in the difference in the
mixing character of the two types of jet flows and the resulting effect on the dominant
combustion mechanisms. A lifted turbulent jet diffusion flame is typically stabilised
some ten to twenty jet throat diameters downstream from the nozzle tip, in a region
where the turbulence properties (eg. mean and RMS velocity profiles, energy spectra)
have reached or are approaching self-similarity. Thus it may be deduced that the
mixing characteristics (eg. the spectrum of mixing scales and local fine-scale strain
rates) which propagate throughout a simple jet flame are generally similar. For simple
jet flames the mixing characteristics depend principally on jet diameter and exit
velocity for a given type of fuel.
G.J.R. Newbold and G.J. Nathan
u = 12.6d~
0 d=3 m u=36.7d~
0 Simple jet flames [ 181
0 d = 10-
Strouhal Number (fdu)
Figure 4. NOx emission measurementsfrom unconfined propane jet diffusionflames
from a mechanical precessing jet nozzle ( 0 = 450) [13]. NOx emission indices for
simple jet flames [18] are indicated on the ordinate (experimental condition can be
found in Nathan et al. [13]).
By contrast the turbulence characteristics at the region of flame stabilisation in
precessing jet flows are far from self-similar and the turbulence properties there differ
drastically from those in simple jet flows [9,10,12].
The reduction of the
characteristic strain rate at the tip of the flame by jet precession, taken together with
the increased radiant fraction associated with high Std MPJ flames [ 131, suggests that
the effect of radiation on the temperature of methane and propane jet flames can
become significant when strain rate characteristics are modified sufficiently. This
finding is entirely consistent with the trend found by Kent and Bastin [17].
3 72
Characteristics of unconfined turbulent jet diffusionJames
Precession of a jet is shown to reduce the large eddy strain rate at the tip of precessing
jet flames compared with simple jet flames. It is argued here that the reduced strain
increases the sooting tendency and so results in the subsequent increase in the radiant
fraction that was measured by Nathan et al. [13]. The correlation is consistent with
trends found by Kent and Bastin [17] for simple jet diffusion flames, where the bulk
strain rate is characterised by the throat velocity and diameter. The decrease in the
NOx emissions has previously been found to correlate with an increase in global
residence time and reduction in the temperature of the gases in the flame [13]. The
present work demonstrates that the large eddy strain rate in precessing jet flames is
reduced by an order of magnitude relative to momentum dominated simple jet flames
and this difference is shown to correlate with reduced NOx emissions and increased
emissivity [13]. This trend is the opposite of that found by R0kke et al. [16,22] for
free turbulent jet flames. This strongly implies that the effect of radiation on the
temperature of jet flames can become significant when the similarity characteristics of
jet turbulence characteristics are modified to reduce the bulk strain rate in a flame.
The implication for burner design is that it may be possible to increase radiant heat
transfer and simultaneously to reduce NOx emissions by promoting soot formation
from reduced bulk strain in jet flows. Jet precession is shown to achieve this whilst
maintaining good burnout.
Throat diameter (upstream orifice diameter FPJ nozzle)
Diameter of the chamber section of the FPJ nozzle
Gravitational constant
Local jet diameter
Flame thermal output
Celerity of large-scale structures
Velocity at the nozzle throat
Height of the fluctuating region of the flame
Average flame length
Mass fraction of the fuel species
Angle of the jet emerging from the MPJ nozzle
G.J.R. Newbold and G.J. Nathan
Density of the source fluid
Dimensionless Groups
Froude number
Reynolds number
Strouhal number
Buoyancy parameter [7]
1. Luxton, R.E., Nathan, G.J., and Luminis Pty. Ltd., 1991. Controlling the motion of afluid jet. USA
Letters Patent No.5.060.867.
2. Manias, C.G., andNathan, G.J., 1993. World Cement. March, 24(3), 4-11.
3. Manias, C.G., and Nathan, G.J.. 1994. World Cement, May, 25(5), 54-56.
4. Nathan, G.J., Hill, S.J. and Luxton. R.E., 1998. J. Fluid Mech., accepted for publication.
5. Nathan, G.J., Luxton, R.E., and SmarL J.P., 1992. Twenty-Founh Symposium (International) on
Combustion, The Combustion Institute, 1399-1405.
6. Newbold, G.J.R., Nathan, G.J., and Luxton, R.E., 1997. Comb. Sci. Tech., 126(1-6), 71-95.
7. Mungal. M.G., Karasso, P.S., and Lozano, A., 1991. Comb. Sci. Tech., 76, 165-18.5.
8. Cetegen, B.M., and Ahmed, T.A., 1993. Comb. Flame, 93,157-184.
9. Schneider, G.M., Hooper. J.D.. Musgrove, A.R., Nathan, G.J., and Luxton, R.E., 1997. Expt. Fluids,
10. Schneider, G.M., Froud, D., Synd,N., Nathan, G.J., and Luxton, R.E., 1997. fiptpt.Fluids, 23,89-98.
11. Nobes, D.S., 1997. Ph.D. Thesis, The University of Adelaide, Australia.
12. Nathan, G.J., Nobes, D.S.. Mi. J.C., Schneider, G.M.. Newbold, G.J., Alwahabi, Z.T.. Luxton, R.E.,
and King, K.D., 1997. Combustion and Emissions Control III, The Institute of Energy, 49-69.
13. Nathan, G.J.. Turns, S.R., and Bandaru, R.V., 1996. Comb, Sci. Tech., 112,211-230.
14. Turns, S.R., Bandaru, R.V., and Nathan, G.J., 1994. Annual Reporr, Dept. Mech. Eng., for the Gas
Research Institute. GRI Contract No. 5092-260-2596.
15. Newbold, G.J.R., 1997. Ph.D. Thesis, The University of Adelaide, Australia.
16. R@kke,N.A., Hustad, J.E., and SBnju, O.K., 1994. Comb. Flame, 97.88-106.
17. Kent, J.H., and Bastin, S.J., 1984. Comb. Flame, 56,2942.
18. TUXIS,S.R., and Myhr, F.H., 1991. Comb. F l ~ m e87,319-335.
19. Turns, S.R., Myhr, F.H., Bandaru, R.V., and Maund, E.R., 1993. Comb. Flame, 93,255-269.
20. Boerstoel, G.P., Wieringa, J.A., van der Meer, T.H., and Hoogendoom, C.J., 1994. EUROTHERM.
ENEA Research Centre, Saluggia, Italy, 5-7 October.
21. Boerstoel, G.P., van der Meer. T.H., and Hoogendoorn, C.J., 1995. Eighth International Symposium
on Transport Phenomena in Combustion. San Francisco, USA, July.
22. R0kke. N.A., Hustad, J.E., Smju, O.K., and Williams, F.A., 1992.
(International) on Combustion, The Combustion Institute, 385-393.
3 74
Twenty-Founh Symposium
Без категории
Размер файла
1 151 Кб
flames, nox, jet, characteristics, turbulent, sooting, change, unconfined, mixing, emissions, diffusion, influence
Пожаловаться на содержимое документа