close

Вход

Забыли?

вход по аккаунту

?

The Impact of Scaling Criteria on the Characteristics of Pulverised Coal Flames.

код для вставкиСкачать
Dev. Chem. Eng. Mineral Process., 7(3/4), pp.301-331, 1999,
The Impact of Scaling Criteria on the
Characteristics of Pulverised Coal Flames
J. P. Smart* and W. L. van de Kamp**
* Fuel and Combustion Technology International, Copyground Lane,
High Wycombe HP12 3HE, UK
* * International Flame Research Foundation, PO Box 10000,
I 9 70 CA, IJmuiden, The Netherlands
Results are presented of a series of experiments where a 12 M ,
large-scale swirl
stabilised burner is compared with two 2.5 Mw, geometrically similar burners
designed around a constant-velocity and constant-residence-time scaling criterion.
High and low NOx flames were producedj?om all burners and selected flames were
extensively probed to determine in-flame thermo-chemical structures. Results show
that complex diflerences exist between flames. Results are also presented on the
effect of scale on the ignition properties of axial turbulentjet flames. Experimental
findings show an important eflect of ignition distance on NOx that has important
implications with respect to scaling eflects on these flame types. The results detailed
in the text are interpreted with reference to the eflect of scale on turbulent mixing,
two-phase interactions and ignition properties. Results confirm that, regardless of
the scaling criterion selected, similarity of the in-flame thermochemical features is
almost impossible to maintain as burner scale is changed irrespective of the scaling
approach employed.
*Authorfor correspondence.
301
J.P. Smart and W.L. van de Kamp
Introduction
New burner-design concepts are frequently evaluated at scales significantly smaller
than the scale at which they will ultimately be applied industrially. However, the
question arises: how representative of the full industrial scale unit is burner
performance at the reduced scale. This investigation addresses the important area of
scaling swirl-stabilised pulverised-coal burners under both high- and low-NOx
operation.
When such a burner is scaled down the ultimate aim is ideally to achieve similarity in
all fluid dynamic and thermo-chemical related processes in the scaled flame. If this
is achieved, the performance of the scaled burner is identical with the full-scale
situation. In reality, as reported by Spalding [l] and Beer[Z], comprises are
necessary because not all the physical and chemical processes scale in the same way.
The most important scaling considerations with respect to flames of pulverised coal
are related to the effect of scale on the gas-phase fluid dynamics (turbulent mixing),
and two-phase interactions between coaI particies and the main flow, and the
consequential effects on the resultant in-flame thermochemistry.
Background
Scaling Criteria
Traditionally, there are two practical criteria for the scaling-down of swirl-stabilised
pulverised-cd bumers.
These are constant-velocity and constant-residence
(constant mixing time) scaling, as discussed in works by Salvi & Payne [3], Lawn et
aZ[4] and Smart e?aZ[5,6,7]. Both of these scaling criteria rely on the scaling of the
large macro-scale turbulent-mixing process. Geometric similarity of the burner is
maintained in both cases, and swirl number is maintained constant. Constantvelocity scaling is the most commonly used by burner manufacturers; in this case,
302
Impact of scaling criteria on characteristics of PCjlames
velocity and momenttun ratios of co-axial combustion-air and fuel injection are
maintained constant with scale reduction. For constant residence-time scaling, the
bumer velocities scale with the burner diameter as scale is reduced, but co-axial
velocity and momentum ratios are still maintained constant.
To preserve chemical reaction rates, it is also important to preserve the coal particle
size when a burner is scaled down, as ultimately this affects the in-flame temperature
and gas-density distributions.
The principles behind constant-velocity and constant-residence-time scaling have
been published previously by the authors [5,6,7], and are included here for
completeness.
Constant-Velocity Scaling
The employment of this criterion allows the following relationships to be established:
Q, =P
...(1)
Do- Q25
...(2)
where the variable pois a characteristic inlet fluid density, normally the density of the
burner combustion air.
When a bumer is scaled down fiom a baseline throughput, Qo,buc, to a reduced scale
Qo,dd,
the following criterion is employed:
This allows the value of the characteristic burner dimension to be derived for a
303
J.P. Smart and W.L.van de Kamp
scaled-down burner when values for a large-scale burner are known. As geometric
similarity is maintained, all other burner dimensions can readily be calculated.
Constant-Residence-Time Scaling
When the constant-residencetime scaling criterion is used, the ratio (DJUJ remains
constant as scale is reduced. The ratio (DJVJ equates to the large macro-mixing
time-scalm,:
T
,
a DJU, = constant
...(4)
Qo
=P ouoD:
- POD?
...(5 )
Thus
and:
Do a Uo a Q,""
...(6 )
When a bumer is scaled down fiom a baseline throughput Qo.by. to a reduced scale
Qodd
by means of constant-residence-time scaling:-
...(7)
...( 8)
Thus, if values of Uoand Do are known for the large-scale case, values for the
reduced-scale bumer can be derived.
304
Impact of scaling criteria on characteristics of PCjlames
Considerations of the Effects of Scale and Scaling Criteria on
Physical Processes in a Pulverised-CoaI Flame
The Effects of Scale on Turbulent Mixing
In swirl-stabilised pulverised-coal flames, the coal and air are initially segregated by
a distance proportional to the characteristic burner diameter Do. The turbulent
mixing process must overcome this segregation before the dominant combustion and
pollutant formation reactions can occur. This mixing process is scale-dependent.
The mixing rate between two segregated reactants can be described by:
T,
a DJU,
...(9 )
and the flow to be mixed is given as:
The mixing is proportional to the product of flame surface area and velocity; and, as
reported by Hawthorne [S], using standard mixing-layer theory, it can be readily
derived that the rate of mixing R, between initially segregated reactants at an axial
distance x is given by:
a (xDJ(l/DJ(U,DJa xUoDo
...(12)
where the parameters in parentheses are proportional respectively to the area over
which the mixing is taking place, the concentration gmdient and the turbulent
305
J.P. Smart and W.L. van ak Kamp
difisivity. This assumes, however, that the characteristic velocities associated with
the turbulent-diffusion mixing process always scales in fixed proportion to the mean
convective-flowvelocity.
Furthermore, burner scale dictates the amount to be mixed at the above rate. The
amount to be mixed is the burner scale that can be represented in terms of the burner
throughput Qoby:
Qo
a UP:
...(13)
The corollary of the above analysis is that the fractional degree of mixing X,at any
pomr downstream of the burner exit x is proportional to the ratio of the
aforementionedparameters; that is:
X, a (xDoUJ(Up;)a x/Do
...(14)
Implicit in this relationship is that independently of scaling criteria or burner scale
the fractional degree of mixing is constant at constant values of dD0. If all the
combustion and pollutant formation reactions are everywhere fast relative to the
mixing rate R,,,, the rate of reaction is dictated by the rate at which the reactants are
mixed. If this is the case, flames of any scale, or flames of identical scale but scaled
down by means of different scaling criteria, should show identical degrees of reaction
at identical values of dD0.
The Effects of Scale on Two-Phase Interactions
In a swirl-stabilised pulverised-coal flame, ignition and stability are governed by the
manner in which the incoming coal stream interacts with the internal zone (IRZ)
present in the near-burner field of these flows, see Figure 1. The IRZ constitutes a
region of reverse flow fed by hot recirculated combustion products. Coal particles
306
Impact of scaling criteria on characteristics of PCJames
penetrate this region of reverse flow if their momentum is sufficient to overcome
local particle-drag forces. The penetration depth I$ into such a region of reverse flow
is given by Field et a1 [9] as:
...(15)
I$ = UST p
3
particle paths
position 0
Figure 1. Particle dispersion in the new-burnerjleld.
where U,is the slip velocity between the injected coal particles (velocity K,UJand
the reverse-flow velocity in the IRZ (U,).
Therefore:
U S =K,U,- U,
...(16)
= d,Z pp/18p,
...(17)
and:
T,
wherep, and p, are respectively coal-particle density and gas viscosity. It has been
shown by Weber & Dupe [ 101 that the ratio of the inlet-flow velocity to the
307
J.P. Smart and W.L. van de Kamp
reverse-flow velocity in the IRZ is a constant (K4 for a constant inlet swirl number,
thus:
UJU,= K,
...(18)
and by normalising to the characteristicburner diameter:
...(20)
For constant-residence-time scaling U/D0 remains constant, and therefore O/Do
remains constant and shoould result in a similar degree of relative penetration with
scale change.
For constant-velocity scaling UJDo rises with the square root of the thennal input
ratio, and therefore 4/D0is larger at a smaller scale.
This indicates that the normalised degree of particle penetration of the IRZ for
particles of constant size is dependent on how the value of the ratio UJDovaries as
scale is changed and thus scaling criteria employed. This parameter rises as burner
scale is reduced when constant-velocity scaling is used, but it remains constant for
constant-residence-time scaling. This fact highlights a potential problem with
constant-velocity scaling of pulverised-coal bumers, in that if, to preserve the
chemical reaction rates, coal particle-size remains constant when scale is reduced
(rP= constant), the normalised degree of particle penetration of the IRZ is different.
The Effects of Scale on External Recirculation
When a swirling combusting pulverised-coal flame emerges into a furnace, the jet
308
Impact of scaling criteria on characteristics of PCjlames
entrains fumace fluid to conserve momentum. Since the furnace chamber restricts
the supply of fiesh entrainable fluid (furnace gases), an external recirculation zone
(Enis
) created. The fluid entrained in the flame as it emerges into the furnace has
been detrained h m the flame fiuiher downstream. The amount of fluid recirculated,
for flows that are not highly confmed (as in these experiments) is governed by the
furnace confmement according to the following relationship [I I]:
-
h4elMo= 0.4718 0.5
...(21)
9 = 2D,/Dxp,lpJoJ
...(22)
where
The factor 2 appearing in the numerator here as the appropriate diameter for
entrainment is the quarl exit diameter, which is twice the characteristic burner
diameter used in the previous analysis.
Thus the fraction of fluid recirculated and entrained into the flame is a function of the
furnace confinement (DJD,)
and can be maintained constant only if the burner and
the furnace are scaled simultaneously. If there is a significant difference in
Confinement across a scale change, the result is different compositions of the
entrained gases, regardless of whether the burner scaling criteria employed result in
'correct' burner scaling,as detailed in the previous sections.
Experimental
The main experiments detailed here explore, from a phenomenological viewpoint, the
effect on a generic swirl-stabilised pulverised-coal burner of down-scaling and
scaling criteria in the thermal input range 2.5
- 12 MW. The burners used for the
turbulent jet flame studies were rated at 2.5 MW.
309
J.P. Smart and W.L..van de Kamp
Experimental Furnaces
The 2.5 MW swirl stabilsed burner experiments were executed on Furnace No. 1 of
the International Flame Research Foundation (IFRF). The 12 MW experiments were
executed on the Babcock Energy Limited (BEL)Large Scale Test Facility, Renfkew,
Scotland. For the turbulent jet studies, the IFRF CEMFLAME cement kiln simulator
was used. The experimental furnaces for the main experiments are shown
schematically in Figures 2 and 3.
Figure 2.IFRFjmace No. I .
The furnace confmement of the 12 MW burner on the BEL test furnace was 0.25,
whereas the confinements of the constant-velocity burner and the constant-residencetime burner on the IFRF furnace were respectively 0.30 and 0.23.
Experimental Burners
As reported by Smart et aZ[12] and Smart & Weber [13], the experimental swirl
stabilised burner used in this work is capable of operating under high- and low-NOx
310
Impact of scaling criteria on characteristics of PCjlames
operation, dependent on the mean coal-particle trajectories in the near-bumer field
A range of coal-particle paths is possible in this burner, as shown in Figure 1.
-.--
Figure 3. Babcockjkrnace test facility
The predominant path has a marked influence on the NOx emission from this burner.
For example, if the paths are predominantly of Types 1 and 2 shown in Figure 1, the
devolatilsed volatile species, including volatile nitrogen, are oxidised on the internal
recirculation zone boundary in a region of high oxygen availability in the shear layer
that surrounds it. This results in efficient conversion of fuel nitrogen to NOx, and a
high resultant NOx emission in the furnace exhaust.
By contrast, if Types 3 and 4 particle trajectories predominate, devolatilisation occurs
primarily within the boundary of the internal recirculation zone in a region of low
oxygen concentration. This is conductive to the promotion of preferential reaction
pathways from volatile-fuelnitrogen to molecular nitrogen. It has been demonstrated
[12,131 that optimisation of the particle trajectories can lead to significantly reduced
NOx emissions from this burner.
311
J. P.Smart and W.L. van de Kamp
Low-NOx operation is achieved most simply by insertion of the coal injector
downstream off the burner throat; see Figure 1.
For the experiments, three geometrically similar burners were used (figure 4). The
first was a 2.5MW standard version, as used extensively in the development
programmes on this burner; it equates to a constant-velocity burner of any scale. The
velocities employed in this burner are typical of those used regularly in the burners of
large-scale utility boilers (40 ms-1 for the combustion air and 20 ms-1 for the transport
air). This burner can therefore be considered as a constant-velocity scaled version of
a larger-scale version.
The constant-velocity scaling criterion was used for the scaling-up of the 12MW
baseline burner up from the 2.5MW generic burner: the primary and secondary air
velocities were respectively 20 and 40 ms-1. The 2.5MW constant-residence-time
burner was designed simply by scaling down from the baseline 12MW burner using
the constant-residence-time scaling criteria. The velocities of combustion air and
primary transport air for the 2.5MW constant-residence-time scaled burner were
respectively 24 and 12 msl. Here, to prevent the coal settling out in the coal injector,
the coal was maintained at 15 ms-1 until close to the end of the coal injector, where
the annulus was expanded to give the required 12 ms-1 at the injector trip. Swirl in
the combustion air was generated by means of an IFRF movable-block swirleP. For
the analysis of experimental results, the burner throat diameter Do is selected as the
characteristic burner diameter. Schematics of the three experimental burners are
shown in Figure 4.
312
Impact of scaling criteria on characteristics of PCjlames
The coal used in the experiments was Gottelborn hvBb, a German Saar coal. The
coal was pre-dried, crushed, milled and pulverised to 75% ,75m before being
transported to all the bumers at a transport akfuel ratio of 2:l on a mass basis. The
excess-& factor was maintained at 1.15 for each bumer throughout the experiments.
Temperatures of combustion air and transport air were maintained at 300 and 7 K
respectively.
For each burner, the effects of coal-injector position and swirl level were evaluated.
For selected flames, extensive in-flame maps of gas composition and temperature
were performed.
Ild
Figure 4. ExDerimental burners
313
J.P. Smart and W.L.van de Kamp
Experimental Procedure
For each flame the experimental factors were allowed to come to thermal
equilibrium. When the steady state had been achieved, gas composition, temperature
and burnout were measured in the furnace exhaust.
Standard IFRFBEL gas-sampling probes and suction pyrometers were used for
measuring flue and in-flame gas species and temperatures [15].
The suction
pyrometer was equipped with a Pt-Rh6o/o/pt-Rh30% thermocouple. Gases were
analysed by means of standard gas meters - NDIR for CO and CO,, paramagnetic for
0, and chemi-luminescence
for NOx. In-flame measurements concentrated on the volatile flame.
Results and Discussion
General Flame Features and Observation
For all flames studied in this work, the total combustibles burnout, as measured by
ash tracer in the furnace exhaust, was greater than 99.0%.
Baseline Flames :Type 2 Flames
Baseline Type 2 flames were generated with the coal injector positioned at the burner
throat, dD, = 0; see Figures 1 and 5. Under these conditions the highest NOx
emissions in the furnace exhaust (corrected to 0% 0,) are observed for the constantvelocity burner; see Figure 6. The constant-residence-time burner gave the lowest
NOx levels, and the 12MW baseline burner produced intermediateNOx
314
Impact of scaling criteria on characteristics of PCj7ame.s
levels; this indicated that the mixing rate between fuel and air is highest for the
constant-velocity burner.
Effect of Ignition Pattern :Type 2 Flames
The baseline Type 2 flames studied in this work ignite in a well-defined localised
region where the incoming coal has mixed with sufficient recirculated combustion
products, and has devolatilised sufficiently to create a flammable fuevoxidant
mixture at a temperature high enough to ignite. As reported by Lawn et a1 [4], this
ignition must be expected to take place at a lean flammability limit. The ignition
fiont is on the IRZ side of the shear mixing layer located between the secondary
combustion air and the IRZ; see Figures 1 and Sa. For baseline flames, most of the
coal-particle trajectories are in this mixing layer, thus the majority of devolatilisation
also takes place in this region. Consequently flame ignition takes place on the IRZ
side of the coal jet.
la1
(bl-penetrorion flame
nearhmerfldd
-1
Figure 5. Coaljet tragectories and ignitionpatternsfor type I and type 2 flames
Recent work by Smart [I61 suggests that after ignition, the flame fiont initially
propagates into the fuel stream into an increasingly rich mixture. This is due to
devolatilisation having occurred ahead of the flame front as a result of radiation from
the flame front itself and hot refractory surfaces local to the ignition region.
315
J.P. Smart and W.L.van de K m p
The continuous entrainment of oxidant into the shear mixing layer subsequently
causes the flame front to propagate into a progressively leaner mixture. The flame
front is then ultimately convected downstream, ppagating at the lean limit.
In a constant-velocity scaled flame, the mixing of oxidant with the fuel stream is
more rapid than for the constant-residence-time case, because the value of U p e is
Iarger. This indicates that flame propagation in the shear mixing layer of a constantvelocity scaled flame is globally leaner than the corresponding residence-time case.
The corollary of the aforementioned discussion is that the constant-velocity scaled
flame is propagating giobally leaner than a constant-residence-time flame of the same
scale. Thus, a higher global conversion of volatile nitrogen species to NOx occurs in
the former case. Moreover, the potential for more coal particles to penetrate the IRZ
in the constant-velocity case (see section on "the effects of scale on two-phase
interactions") would augment this effect, as less coal devolatilises on the IRZ
boundary, and the local gaseous environment on the IRZ boundary is significantly
leaner.
Effect of the Coal Injector Position
The effect of coal-injector position on NOx emissions is also shown in Figure 6,
where the coal-injector insertion is seen to result ultimately in significant NOx
reduction. However, an interesting observation is that for both the swirl numbers
studied, the normalised insertion distance at which NOx falls is less for the constantvelocity scaled flame. A phenomenological explanation of this observation can be
given by expanding the arguments presented in previous sections, to take account of
the effect of inserting the coal injector forward of the burner throat on the ease of
penetration of the IRZ. To obtain low NOx emissions from the experimental burner,
a specific fractional or normalised degree of IRZ penetration by the coal jet
316
Impact of scaling criteria on characteristics of PCjlames
is necessary, as reported by Smart & Weber [ 131. This is independent of burner scale
or the scaling criterion used.
normalised cod injector padt&mtx/Dol
Figure 6. Efect of scale on NOX emissions
The two main parameters that govern the degree of IRZ penetration are (UJD,,) and p.
The latter parameter was essentially constant, as the coal particle-size distribution
was invariant for all experimental burners. However, for burners of the same scale
the value of (UJD,,) is larger for a constant-velocity scaled burner relative to a
corresponding constant-residencetime case. To take account of the coal injector
being inserted downstream of the burner throat, the following can be written:
4/Do= K3(UJD& + (IJD,,)
...(23)
where K3 = (K,- l/KJ and (IJD,,) is the normalised position of the coal injector
relative to the burner throat. If a minimum normalised degree of IRZ penetration
317
J.P. Smart and W.L. van de Kamp
for low-NOx operation is defined as:
...(24)
this is achieved at a shorter noxmalised coal-injector insertion distance for the
constant-velocity scaled burner, because of the value of (VJD,,) is larger; see Figure
6. This analysis assumes that the turbulent structure of the flow in terms of macro
and micro mixing lengths and velocity scales are scaling with Do and U,through the
whole flow/flame domain [17].
Consequently, the parameter that should govern the noxmalised coal-injector
insertion distance required for sufficient penetration of the coal jet into the IRZ to
achieve NOx reduction is (VJDJ, assuming that the reverse-flow velocity in the IRZ
also scales with U,;
see Section 4. Relative to the baseline 12MW case,the value of
(UJDJremains constant by definition for the constant-residence-timescaled flames,
but becomes larger for the constant-velocity scaled case.
Effect of Ignition Pattern :Penetration Flames
For the penetration flames,the ignition pattern is significantly different from baseline
Type 2 flames. As the coal jet penetrates the IRZ on the burner axis, the centreline
region of the IRZ does not contain recirculated combustion products; see Figures 1
and 5. In this case the coal jet ignites at the lean limit but on the outside of the coal
jet [16]. The flame h n t then propagates into the coal stream in a progressively
richer mixture limited by devolatiliion, until the rich limit is reached. At this point
the flame front is convected downstream at the rich limit, constrained by oxygen
availability.
Because the flame-propagation process in penetration flames takes place under
predominantly fuel-rich conditions, the result is reduced NOx emissions.
318
Impact of scaling criteria on characteristics of PCj7ames
c
d
am
0.0
01
1.0
1J
U
W
m.
( d o
-
U
9.
Figure 7.In-frame measurementsfor the baselinejlames
319
J.P. Smart and W.L.van de Kamp
Detailed In-Flame Measurements :Type 2 Flames
This section presents a comparison of the constant-velocity, constant-residencetime
and 12MW baseline flames. Figure shows flue-gas output conditions and in-flame
measurements between dD0 = 1.O and 5.0. At the quarl exit d D o = 1.O) the shear
mixing layer is identified by the peaks in CO concentrations for both flames.
Corresponding temperature profiles confm ignition of these flames on the IRZ side
of the shear mixing layer. NOx is also observed to be forming in this shear-layer
region. The CO concentrations here are generally higher m the constant-residencetime flame than in the constant-velocity or the 12MW baseline case. This may well
confm a richer flame propagation in the former case,and suggests that the turbulent
mixing process is not scaling directly in accordance with the macro (convective)
mixing rates.
From observation of the in-flame profiles between the quarl exit, .dDo = 1.O, and .dDo
= 5.0 the following
comments can be made. Between the quarl exit and dDo= 2.0 the peak CO
concentrations are higher for the constant-residence-timeflame, because of the effect
of a richer global flame-propagation pattern. In this region, NOx levels in the IRZ
are highest for the constanwelocity burner and lowest for the constant-residencetime burner; the 12MW baseline burner produces intermediate levels. At dDo= 5.0
all radial variations in concenrration and temperature are negligible, and values are
broadly representative of those found in the fumace exhaust; this indicates the
completion of turbulent macro-mixing. The observation that NOx does not change
significantly between the completion of macro-mixing and the fiunace exhaust
suggests that NOx is essentially 'frozen' in concentration after the completion of
macro-mixing at the termination of volatile combustion.
320
Impact of scaling criteria on characteristicsof PCflames
01
1
!-OD
-t
Figure 8. In-flame measwesments for the penetrationjlames
321
J.P. Smart and W.L.van de Kamp
Detailed In-Flame Measurements :Type 1 Penetration Flames
For the constant-velocity and constant-residence-time flames, the measured Type 1
penetration flames were produced with the coal injector inserted 0.4 throat diameters
downstream of the burner throat (dDo = 0.4).
For the 12MW reference case,
significant penetration of the IRZ could not be achieved at d D o = 0.4, and a further
insertion to / D o = 0.6 was required. This suggests that in comparison with the
constant-velocity and constant-residence-time burners, the reverse-flow velocity in
the IRZ of the Type 1 penetration flame is stronger relative to the coal jet velocity.
However, in order to obtain a specific degree of penetration into the IRZ, the physical
size of the IRZ is important. Insertion of the coal gun into the IRZ has to be
relatively higher for the larger-scale bumex, because the remaining distance is larger
for the same slip velocity between the coal particles and the reverse flow within the
IRZ.
Figures 8a-8d show the detailed in-flame measurements for the penetration flames
between /Do of 1.0 and 5.0. Qualitatively there is broad similarity between all the
profiles, and especially temperature at /Do = 2.0. At d D 0 of 1.0 the oxygen and
temperatureprofiles clearly indicate external ignition of the coal jet, as seen in Figure
5 ; the oxygen profiles show the paths of both the secondary air and the primary coal
jet. Temperatures indicate that the flame fimt of the 12MW flame has propagated
onto the bumer centreline. This is not the case for the constant-residence-time flame
or the constant-velocity flame. Here the temperatures are less than 2 0 K , and
clearly in these flames the flame-fiont has not propagated through the coal jet.
322
Impact of scaling criteria on characteristics of PCJames
Between d D o of 2.0 and 5.0, the NOx and CO levels are markedly different for the
three flames, and indicate that the actual turbulent fueVair mixing rates are different
despite the expected effects of scale on the macro-mixing rates.
Observation of on-flame data for d D 0 = 2.0 shows a complex difference in the
thermochemical structure of the flame. Temperature measurements clearly indicate
that flame ignition has taken place for all flames. However, the 0,and CO profiles
(in terms of carbon monoxide production and oxygen consumption) show clear
indications of ignition for the reference 12MW penetration flame and the constantvelocity scaled penetration flame. In contrast, the constant-residence-time scale
penetration flame shows that flame ignition has not occurred.
Effects of Scale on Turbulent Intensity and Macro-Mixing
It has been shown previously was shown that the degree of mixing (and thus reaction
for mixing-rate limited combustion), irrespective of the burner scale or scaling
criteria used, scales with dD,. This assumes, however, that the magnitude of the
fluctuating turbulent velocity that drives the turbulent mixing process between fie1
and air always scales in fured proportion to the characteristic flow velocity U,. Thus
the turbulent intensity (u '/Ubremains constant. Here we expand this idea to suggest
a method for a filler analysis of the results observed in the experiments.
From the in-flame measurements it is clear that the degree of overall reaction, as
indicated from the significantly higher CO levels in the near burner field of the
2.5MW baseline constant residence-time burner, is qualitatively not as advanced as
for the baseline 12MW or the constant-velocity scaled burner (at corresponding
values of dD,,).
In the present analysis it is assumed that the inlet flow conditions to all burners are
fully developed and that this drives the fueVair mixing process. (We ignore here
323
J.P. Smart and W.L. van de Kamp
combustion-induced modifications to the Reynolds Number, and assume that the
characteristic Reynolds Number in the shear mixing region of the experimental
flames can be represented by, and its proportional to, the inlet Reynolds Number).
Here a potential scaling problem is highlighted. Both the 12MW baseline flames
have inlet Reynold Numbers well into the region of fully turbulent flow, whereas the
constant-residence-time burner has a characteristic Reynolds Number falling in the
transitional regime. This indicates that the fluctuating
inlet velocity component zd does not scale with the characteristic inlet velocity U,,
and in effect has a reduced value.
This would suggest, as a consequence, that the turbulent mixing rate between fuel
and air is also reduced for the constant-residence-time bumer, and that the dD,,
criterion does not hold. Because the mixing rate is reduced, the rate of reaction is
reduced; higher CO levels are observed relative to the baseline and constant-U,
bumers, which (as previously mentioned) exhibit Reynolds Numbers well into the
completely turbulent regime.
The above analysis makes some broad assumptions about the nature of the flow in
the shear layer for the constant-residence-time burner, but the suggestion is of a
fundamental difference in the turbulent structure for the constant-residence-time
flame, and that this is responsible for the observed differences in in-flame depths.
The phenomena highlighted above are much more significant for smaller-scale
constant-residence-time scaled flames (and for constant-velocity scaled flames at a
less scale than 2.5MW), and should be borne in mind when any scaling attempt is
made. (It is worthy of note that at very small scale, flow laminarisation occurs, and
any attempt to scale the results to industrial scale bumer will be fraught with
problems).
324
Impact of scaling criteria on characteristics of PCjlames
The conclusion of the above discussion is that representative industrial information
can be obtained only at above a minimum scale, which is dependent on the size of the
original burner. In the set of experiments described here, a 12MW burner is scaled
down with a scaling factor of circa 5 to 2.5MW; significant differences are already
observed in the near-burner region for both high- and low-NOx flame types, and also
further downstream for the penetration low-NOx flames.
Conclusions
This study compares the effects of burner scale and scaling criteria on the
performance of a generic swirl-stabilised pulverised-coal burner in the thermal input
range 2.5 to 12MW, operating under baseline high-NOx operation and low-NOx
aerodynamically staged firing conditions. Throughout the study, the work was
performed on the same coal (Gottelborn hvBb, a Saar coal) and with the same
particle-size distribution. The bumers studied included 2.5MW constant-velocity and
constant-residence-time scaled versions of a 12MW baseline burner.
Results showed that complex differences exist between all flames. For the baseline
(high-NOx) firing configuration, significant differences exist in flue-gas NOx
emission levels. For the baseline Type 2 flames, the NOx emissions in the furnace
exhaust were higher for the constant-velocity burner than the constant-residence-time
case, with the 12MW baseline flame giving intermediate values.
In-flame
measurements confm that ignition of these flames takes place on the IRZ side of the
coal jet, as it mixes initially with combustion products from the IRZ in the
sumounding shear mixing layer. The difference in NOx is explained in terms of the
differences in flame-propagation pattern through the coal jet. Detailed in-flame
measurements indicate that a preferential penetration of the IRZ by the coal jet
occurs for the constant-residence-time burner, resulting in the observed lower NOx
emissions.
325
J.P. Smart and W.L. van de Kamp
In-flame measurements also confm that for the baseline fuing configuration, the
flow domain is not scaling in accordance with the scaling rules previously discussed.
For all burners, NOx was observed to fsll as the coal injector was inserted into the
burner quarl, and this can be explained by the different flame ignition and
propagation patterns. However, it was clearly observed that the NOx fell at a lesser
normalised coal injector insertion distance for the constant-velocity scaled flame than
for the constant-residence-time flame. For the 12MW baseline bwner a further
relative insertion distance was required before the low-NOx operation was reached.
The steepest fall in NOx with coal-injector insertion could be observed for the
constant-velocity scaled burner.
Low-NOx penetration flames were produced for each burner. For the constantvelocity and constant-residence-timescaled bumer, a coal injector insertion distance
of 0.4 throat diameters was required. By contrast, an insertion distance of 0.6 was
required for the 12MW burner. This indicated a significantly stronger IRZ in the
latter case. As a similar particle-size distribution (equating to a similar particle
relaxation time) was used for the different burners, this resulted in further penetration
being required for the large-scale burner to exploit the larger axial dimension of the
IRZ and achieve the degree of penetration required for low-NOx operation.
Detailed in-flame measurementsperfamed on the low-NOx penetration flames show
broad similarities in global flame structure downstream of the bumer; but significant
differences in detailed chemistry are seen. It is suggested in the main text that the
differences observed in in-flame thennochemistry arise because the flame
aerodynamics do not scale in accordance with the applied scaling rules.
326
Impact of scaling criteria on characteristics of PCflames
Neither constant-velocity nor constant-residence time scaled flames produce
adequate similarities in flame structure and thennochemical fields.
Differences in flame thennochemical features are observed between the constantvelocity and constant-residence-time flames relative to the 12MW baseline case.
However, the differences are not explicable on the basis purely of considering the
effect of scaling criteria on the turbulent mixing process.
Clearly, an important part is played by differences in coal-particle dispersion in the
near-burner region, and the actual effect of scale on the turbulent mixing process in
the near-burner region. The mixing process of fuel and air in the constant-residencetime scaled burner is less advanced for similar values of relative nonnalised axial
distance when compared with the other two burners. This may be due to the different
scaling of the turbulent velocity in the shear mixing layer caused by a flow pattern
for the transitional flow regime. This indicates that a minimum scale exists at which
representative full-scale information can be obtained when studies are made of
reduced-scale burners.
In conclusion, it is suggested that on the basis of evidence presented in the main text,
neither constant-residence-timenor constant velocity-scaled flames can produce total
similarity in flame structure and thennochemical fields.
Comments on the Effect of Ignition Distance on NOx emissions for
turbulent jet flames Scaling Implications
-
A series of experiments were also performed on coal fwed turbulent jet flames to
evaluate the effects of ignition behaviour on NOx emissions. The burners used were
of the mono-channel design as used fiequently in rotary kilns, see Figure 9. These
burners are fiequently associated with having a stand-off distance from the burner
32 7
J.P. Smart and W.L.van a'e Kamp
nozzle prior to ignition. The magnitude of this ignition distance markedly affects the
NOx emissions and has significant scaling implications. As the ignition of coal is
time dependent, generally the higher the injection velocity the greater the ignition
distance. During this ignition delay the coal jet entrains secondary air and results in
more air entrained into the coal jet at the point of ignition as ignition distance is
Figure 9. vpical axial turbulentjet flame
increased. The consequence of this is a more oxygen rich flame propogation pattern
and higher flame temperatures leading to more effective conversion of fuel bound
nitrogen to NOx and higher tennal NOx formation rates. In Figure 10, results are are
shown for three fuels; a high and medium volatile bituminous coal and a petroleum
coke.
NOx data is presented plotted against the mass of air entrained into the fie1jet at the
point of ignition relative to the stochiometric air requirement. The mass of air
emtrained was calculated using the following equation [181:
MJM,
= 0.32x/DO(pJp,
)0.5
Ignition distance was determined from visual obsevation.
328
Impact of scaling criteria on characteristics of PC flames
Implicit in the above discussion is an important scaling effect. Reitterating the time
dependent ignition delay of coal particles, the ignition distance is greater relative to
the burner diameter for a small scale burner compared to the full industrial scale for
the same injection velocity. Implicit here is an observation that small scale burners
will produce more NOx compared the full industrial scale therefore.
Results are quite dramatic and the effect of fuel reactivity on ignition delay is clearly
I
Mass of air e n h i n d into jet at p i m t nf i p i t i o n relative rn s t h i o n c h i c
Figure 10. NOx emissionsfor the turbulentjet flame
Acknowledgements
The authors would like to thank the Executive Committee of the IEA Combustion
Sciences Programme Annex 2, the CEMFLAME Consortium, and the Joint
Committee of the IFRF for permission to publish this paper.
Nomenclature
dP
Df
Do
4
IRZ
diameter of particle, m
characteristicdimension of fumace, m
characteristic burner diameter, m
insertion distance of coal injector, m
Internal Recirculation Zone
329
J.P. Smart and W.L. van de Kamp
K,K,,K2 constants of proportionality
inlet mass flow-rate, kg s1
recirculated mass flow-rate, kg s-1
burner throughput, kg s”
radial distance, m
mixing rate, m3s-1
fluctuating velocity component, m s’1
characteristic burner velocity, m s-1
slip velocity, m s-1
reverse-flow velocity within IRZ, m s-1
axial distance, m
hctional degree of mixing, dimensionless
Thring-Newby parameter, dimensionless
dynamic viscosity of gas, Pa s
density of particle, kg m-3
characteristic density of inlet fluid, kg m3
density of fumace gases, kg m3
macro-mixing time-scale, s
particie relaxation time of a coal particle of diameter, d,,,s
constant residence time, constant velocity
penetration depth, m
M
O
Mr
Qo
R
Rln
U’
uo
us
ur
X
4
0
p*
PP
Po
P,
Tm
T
7, v
4
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
11.
330
Spalding, D.B. The art of partial modelling. 9th Symp ( I t ) on Combustion. The Combustion
Institute, 1962, p. 833.
Beer, J.M. Phenomenological models for flames in furnaces. US DOE Workshop on Modeling
of Combustion in Practical Systems, 1978.
Salvi, G.and Payne, R Combustion system scaling. International Flame Research Foundation
Report DO7/a/69,1978.
Lawn, CJ., Cunningham, A.T.S., Street, P.J., Matthews, KJ., Sargeant, M. and Godridge,
A.M. The combustion of heavy fuel oils 1n:The Principles of Combustion Engineering for
Boilers (ed Lawn). Academic Press, London, 1987. Chap 2, pp 61-196.
Smart,J.P., Morgan, DJ. and Roberts, P.A. The effect of scale on the performance of swirl
stabilised pulverised coal burners. 24th Symp (Int) on Combustion. The Combustion Institute,
1992 p. 1365.
Smart,J.P. and Morgan, D.J. Exploring the effects of employing different scaling criteria on
swirl stabilised pulverised coal burner performance. Combust Sci and Technology. 1994. 99,
1-2. p. 105.
Smart, J.P., van de Kamp, W.L. and Morgan, M.E. The effect of burner scale on NOx
emissions from a swirl stabilished pulverised coal burner. Fuel, 1990.70, p. 1350.
Hawthome, W.R., Wedell, D.S. and Hottel, H.C. Mixing and combustion in turbulent gas jets.
3rd Symp (lnt) on Combustion. The Combustion Institute, 1949, p. 266.
Field, M.A., Gill, D.W., Morgan, B.B. and Hawksley, P.G.W. Flow patterns and mixing. In:
Combustion of Pulverised Coal, BCURA. Leatherhead, 1967, Chap 2, pp. 25-79.
10.
Weber, R. and Dugue, J. Combustion accelerated flows in high confinements. Prog
Energy and Combust Sci, 1992,18, p. 349.
Thring, M.W. and Newby, M.P. Combustion length of enclosed turbulent jet flames. 4th
Symp (Int) on Combustion. The Combustion Institute, Pittsburgh, 1953.
Impact of scaling criteria on characteristics of PCflames
12.
13.
14.
15.
16.
17.
18.
Smart. J.P., h i l l , K.J., Visser, B.M. and Weber, R. Reduction of NOx emissions in a swirl
stabilished coal flame by particle injection into the internal recirculation zone. 22nd Symp (Int)
on Combustion. The Combustion Institute, 1988, p. 117.
Smart, J.P. and Weber, R. Reduction of NOx and optimisation of burnout with an
aerodynamically air-staged burner and an air-staged precombustor burner. J. lnsf .Energy,
1989,62, pp, 237-245.
Beer, J.M. and Chigier, N.A. Combustion Aerodynamics Krieger Publisher Company, Florida,
1983.
Chedaille, J. and Braud, Y. Measurements in flames. In: Industrial Flames, Vol. 1 (ed Beer &
Thring). Edward Amold, London, 1972.
Smart, J.P. On the effect of burner scale and coal quality on low NOx burner performance.
PhD Thesis, University of London, 1992.
Tennekes, H. and Lumely, J.L. A first course in Turbulence. MIT Press, Cambridge,
Massachusetts, USA, 1972 (Ch. 1).
Ricou. F. P and Spalding D.B. Measurement of entrainment of axisymetric turbulent jets. J.
Fluid Mech, 1960, 11, p21.
33 I
Документ
Категория
Без категории
Просмотров
0
Размер файла
1 251 Кб
Теги
flames, pulverised, characteristics, coal, criterias, scaling, impact
1/--страниц
Пожаловаться на содержимое документа