Professional Documents
Culture Documents
Department of Fire Protection Engineering, Worcester Polytechnic Institute, Worcester, MA 01609, USA
Department of Mechanical Engineering, Indian Institute of Technology Madras, Chennai 600 036, India
a r t i c l e
i n f o
Article history:
Received 19 October 2011
Received in revised form 3 January 2012
Accepted 8 February 2012
Available online 12 March 2012
Keywords:
Coal dust interaction
Laminar premixed ame
Equivalence ratio
Laminar burning velocity
Shadowgraph image
Cone angle method
a b s t r a c t
This study investigates the interaction of micron-sized coal particles entrained into lean methaneair
premixed ames. In a typical axisymmetric burner, coal particles are made to naturally entrain into a
stream of the premixed reactants using an orice plate and a conical feeder setup. Pittsburgh seam coal
dust, with particle sizes in the ranges of 025 lm, 5363 lm, and 7590 lm, is used. The effects of different coal dust concentrations (10300 g/m3) entrained into the mixture of methaneair at three lean
equivalence ratios, /, of 0.75, 0.80 and 0.85, on the laminar burning velocity are studied experimentally.
The laminar burning velocity of the coal dustmethaneair mixture is determined by taking high quality
shadowgraph images of the resulting ames and processing them using the cone-angle method. The
results show that the laminar burning velocity reduces with the addition of coal dust having particle sizes
in the ranges of 5363 lm and 7590 lm, irrespective of the equivalence ratio values. However, burning
velocity promotion is observed for one case with particle size in the range of 025 lm at an equivalence
ratio of 0.75. Two competing effects are considered to explain these trends. The rst effect is due to volatile release, which increases the overall equivalence ratio and thus, the ame temperature and burning
velocity. The second is the heat sink effect that the coal particles take up to release the volatiles. This process reduces the ame temperature and accordingly the burning velocity also. A mathematical model is
developed considering these effects and it is seen to successfully predict the change of laminar burning
velocity for various cases with different dust concentrations and equivalence ratios of the gas mixture.
Furthermore, the implication of this study to coal mine safety is discussed.
2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
1. Introduction
Many materials, which are virtually non-ammable in their
bulk form, become highly reactive and even explosive, if dispersed
as a cloud of ne particles in air, due to signicant increase in the
surface area for enhancing heat and mass transfer processes. From
a combustion viewpoint, this can be treated as both a benet and a
hazard. In industries that manufacture, transport, process and/or
use combustible dusts, accidental dust deagrations represent a
real hazard to both personnel and equipment. An example is the
recent coal mine explosion in West Virginia (April 5, 2010), which
resulted in the killing of 29 miners and considered as one of the
most disastrous mining accidents in US history. Interestingly, most
coal mine explosions often involve both a methane deagration
combined with fugitive coal dust that is collected by the combustion wave as it progresses through the mine.
The physical and chemical processes involved during the travel
of a combustion wave through a ammable gasdustair mixture
0010-2180/$ - see front matter 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.combustame.2012.02.013
2450
Nomenclature
A
B
b
Cp
Cs
Ctotal
cs
E
h
k
Lv
M
m000
CH4
m000
air
m000
fuel
m000
v
n
ns
P
n_ air
Q_
q_ 00
R
r
Su
Tb
2. Experimental apparatus
Figure 2a presents the schematic of the experiment setup used
for the measurement of laminar burning velocity and dust entrainment rate. The design is based on the concept of a Bunsen burner
having side openings to entrain coal dust particles into the ow
of the reactant mixture. Specic details of the dust injector in such
a burner are discussed by Xie et al. [9]. The burner is made of a
steel tube with an inner diameter of 10.2 mm and wall thickness
of 1.2 mm. A 1 mm thick acrylic plate with a 1 mm diameter orice
Tf
T 0f
T 00f
tr
U
V_ air
V_ CH
V
x
Ze
Greek symbols
a
cone angle
q
density of the mixture
qs
density of the particle
d
thickness of devolatilization zone
U
original gaseous mixture equivalence ratio
/s
equivalence ratio of coal particles and air assuming
complete gasication
Subscripts
b
adiabatic condition based on original gas phase condition
f
ame
s
solid particle
u
conditions in the ambient condition
v
vapor
is installed inside the steel tube, 150 mm away from the burner
exit, to allow the ow streamlines to become parallel well before
the burner exit. Dust is fed to the orice plate through three openings of dimensions 7.5 mm wide and 9 mm long, located on the
tube in an axisymmetric fashion above the orice plate as shown
in Fig. 2b. A brass jacket tube of inner diameter 0.1 mm larger than
the outer diameter of the steel tube is secured by two socket head
screws. This is used to adjust the opening size and therefore the
dust entrainment. Pittsburgh seam coal dust (with no additives)
is used in the present study. Few separate experiments have been
2451
Fig. 2. Schematic of the experimental apparatus: (a) dust burner and the weighing assembly; (b) dust entrainment assembly.
carried out with sand particles as well. The sanddust particles are
used to analyze the inuence of an inert on the burning velocity.
The coal dust is lled in an inverted cone-shaped acrylic container,
which is also attached to the steel tube. The cone angle of the container is equal to 60, which represent the critical angle of repose of
dust particle size ranges used in this study. The repose angle is
determined using an experimental method discussed by Botz
et al. [10]. The adjustable burner and its attachments are secured
in a support frame and the entire assembly is kept over a Cole-Palmer Symmetry PR 4200 load cell. The load cell has a total weighing
capacity of 4.2 kg with a sensitivity of 0.01 g. The factory specied
uncertainty in the mass measurement is 0.03 g. A ring stand is
kept outside the load cell to support a collection pan as shown in
Fig. 2. Pittsburgh seam bituminous coal dust with particle sizes
in the range of 025 lm, 5363 lm, and 7590 lm are used in
the experiment. The size ranges are obtained by Retsch AS 300 sieve
shaker. Compressed air and methane (99.99% purity) cylinders are
used to supply the burner with an upstream pressure of 0.5 bar.
Each gas ow is controlled by a SIERRA Model 100 mass ow controller, which has an accuracy of 1% of its maximum ow capacity.
Direct shadowgraph technique is used to capture the ame cone
with or without coal dust injection. The schematic of the shadowgraph setup is shown in Fig. 3. A projector lamp of capacity 420 W
is converted into a point light source and placed at a distance of
one focal length from a double convex lens. A Canon EOS 5D single-lens reex (SLR) camera with a macro-lens (Canon EF100/2.8
Macro USM) with a minimum focal length of 31 mm is placed be-
hind the ame along the center axis of the parallel light beam. The
camera is manually adjusted (shutter speed of 1/4000 s, ISO of
1600, and aperture of 2.8) to obtain the sharpest image for post
processing.
3. Experimental results
A sample image of the actual ame and the shadowgraph image
obtained by the macro-lens are shown in Fig. 4. The main advantage of the shadowgraph is that it can capture clearly the ame
cone even when the ame is loaded with high concentration of
burning particles. As shown in Fig. 4a, it is hard to locate the edge
of the ame cone using conventional direct photograph technique,
while the ame cone is easily identied in the shadowgraph
(Fig. 4b).
For each dust concentration and equivalence ratio of the reactant mixture, a minimum of 15 images are captured and processed
by an image process algorithm programed in MATLAB. The corresponding average cone angle is used to estimate the laminar burning velocity. A sample of the processed image is shown in Fig. 4c.
The algorithm converts the shadowgraph into a gray-scale image
and detects the cone edge where a signicant change in the normalized intensity (a value from 0 to 1) is observed on each row
within the preselected boundary of the ame cone. The detected
cone edge is shown as two clusters of dots in Fig. 4c. Slopes that
connect each dot on one side of the cone are calculated. Then a
2452
Fig. 4. Photographs of (a) actual ame, (b) shadowgraph and (c) processed image.
best-t line for all the detected dots, shown as black solid line in
Fig. 4c, is used to obtain the averaged cone half angle (a). The standard deviation of the angle experimentally measured is within
1.5.
The laminar burning velocity is obtained by using cone angle
method as follows:
Su U sina;
4. Mathematical model
As mentioned earlier, a mathematical model is developed to explain the observation made in the experiments and to carry out
parametric studies to predict Su for particles of sizes not included
in the experiments. Figure 6 illustrates the interaction of the particles with the premixed ame and is used as a basis to develop a
model to estimate Su.
The path of a coal particle, assumed to be along a ow streamline, is shown in Fig. 6. The particle absorbs the heat from the ame
while it travels though the devolatilization zone as illustrated by q_ 00
in the inset to Fig. 6. The initial temperature of the particle is assumed to be equal to the mixture temperature. Due to heat transfer
from the ame, once the temperature of the particle reaches a va-
2453
lue of Tv, the gasication process is initiated, thereby releasing gas_ 000
eous volatiles at the rate of m
v (Fig. 6). In this study, for simplicity,
the volatiles are assumed to be constituted by methane only as
suggested by Seshadri et al. [12]. This additional fuel released from
the coal particle can increase the equivalence ratio of the reactant
mixture locally in the devolatilization zone. For a lean mixture
ame, increased equivalence ratio promotes Su. This is the rst effect of the particle interaction. In Fig. 6, Tf is the temperature of the
ame and the thickness of the zone within which the temperature
increases from Tv to Tf is denoted by d. It is assumed in this study
that coal particles will pyrolize and release volatiles when it traverses this distance d. The height and the width of the ame cone
are represented by h and b, respectively. The cone half angle is designated as a. In addition, the particle also acts as a heat sink, since
it takes heat from the ame. As a result, the ame temperature and
thus Su is reduced. This is the second effect of particle interaction in
premixed ames. These two effects are competing in nature and affect the ame speed simultaneously; based on the combined effect
the ame speed will either increase or decrease due to particle
injection. To simplify the theoretical model, the two processes
are analyzed initially in a decoupled manner, and then re-coupled
later. The change in the ame temperature due to each effect is calculated separately and then superimposed to obtain the estimated
correct ame temperature.
d T
dx
qUC total
kT b T v
qUC total T b T u
cs ;
,
_
_
U V air V CH4
pb2
4
!
:
d
sina
tr
U:
_ 000
m000
v m
v tr :
With this amount of fuel added to the supplied reactant mixture, the new equivalence ratio can be calculated. The consolidated
amount of the gaseous fuel per unit volume in the mixture, de000
000
000
noted by m000
fuel , can be estimated as: mfuel mCH4 mv , where
000
mCH4 is mass of the original methane present per unit volume of
the mixture. Accordingly the new equivalence ratio is calculated
as,
_ 000
where m
v is mass of gaseous fuel evolved per unit volume per second (g/m3s). The quantities A and n are constants for a given coal
type. The particle temperature is denoted as Ts and it is assumed
that Ts = 0.5(Tb + Tv), where Tb is the adiabatic ame temperature
based on the methaneair equivalence ratio. For coal particles, devolatilization initiates at a temperature (Tv) of around 600 K [13]. The
value of Tb is calculated using an equilibrium solver, based on the
minimization of free energy (GASEQ) [14]. The total volume of particles present per unit volume of the reactant mixture is estimated
by dividing the dust concentration cs (g/m3) by the particle density
qs (g/m3). The number of particles per unit volume (ns) is then equal
to ns = (cs/qs)/Vs, where Vs is the volume of a single particle, assumed
to be spherical.
2 n
_ 000
m
v Ans 4pr T s ;
q
Volatiles are released from the coal particles into the gaseous
mixture as the result of heat transfer from the reaction zone into
the devolatilization zone. In order to estimate the amount of volatiles present in the gaseous mixture, it is necessary to explore the
rate of the volatile release and quantify the same as a function of
heat transfer.
There are four possible methods to estimate devolatilization.
The rst method is based on a much simplied assumption that
all volatiles are released when the particle reach a certain temperature. However, such a method excludes the fact that devolatilization is a transient process. The second method is to treat the
particles as liquid droplets and the gasication rate can be estimated as discussed by Suard et al. [2]. This method, however, does
not capture the non-linear behavior of devolatilization of a solid
particle such as coal, since the droplet evaporation is fairly rst order and linear. In the other two methods, devolatilization can be
solved by a set of equations based on Arrhenius rate law as described by Solomon and Colket [11], or it can also be estimated
based on a temperature dependent power-law relation as proposed
by Seshadri et al. [12]. The last method is employed in this study as
it is simplied yet an accurate method.
Seshadri et al. [12] used the expression given below to predict
devolatilization rate of coal particles,
dT
0;
dx
/ 9:52
m000
fuel
M CH4
m000
air
;
M air
where the coefcient 9.52 is the ratio of numbers of moles of methane to air when / equals 1. For the supplied mixture, the mass of
methane and air present per unit volume are given as,
m000
CH4
1
_ CH P
V
M
CH
4
4
A;
@
Ru T u V_ air V_ CH
m000
air
1
_ air P
V
M
air
A;
@
Ru T u V_ air V_ CH4
2454
where P is the atmospheric pressure (101,325 Pa), Tu is the unburned gas temperature (293 K) and Ru is the universal gas constant. M denotes the molecular weight of each species and V_ air
and V_ CH4 are the supply volumetric ow rates of air and methane
respectively.
With the new equivalence ratio, the new ame temperature T 0f
is estimated by using the equilibrium solver [14]. The calculated
results are shown in Fig. 7 for coal particles having sizes in the
range of 7590 lm, with supply methaneair equivalence ratios
of 0.75, 0.80, and 0.85. As shown in Fig. 7, the ame temperature
increases with increased dust concentration as it results in more
volatiles to be released and a higher effective equivalence ratio
obtained.
4.2. Heat sink effect of coal particles
In addition to a local increase in equivalence ratio, a coal particle will also act as a heat sink as it absorbs the heat from ame for
devolatilization. Two aspects are considered for modeling the heat
sink effect; (1) the heat used to raise the temperature of coal particles from ambient condition to the ame temperature and (2) the
heat used to gasify the coal particles. Three assumptions have been
made; (1) the heat release rate from the ame is assumed to remain constant, (2) the coal particles simply act as inert particles,
which are also able to devolatilize by absorbing energy from the
ame and (3) coal particles reach the ame temperature when they
reach the ame sheet. Based on these assumptions, the heat released to raise the gas temperature per unit time for ame without
dust will be equal to the sum of heat release required to raise the
temperature of gas and particles and the heat of gasication per
unit time. Therefore, each term should be determined to estimate
the ame temperature after accounting for the heat sink effect.
First, the heat released from ame without dust is calculated.
For a lean methaneair mixture, the global chemical reaction is given by,
/
/
CH4 O2 3:76N2 ! CO2 /H2 O 3:76N2 21 /O2
2
2
mixed ame without coal dust for a given ow rate of air and /, can
be calculated as:
h
i n_
X
air
Q_ T b T u
C p nproduct
;
4:76
10
where n_ air is the number of moles of air supplied per unit time.
For the ame with coal dust particles, the same heat is released;
however, it is also affected by the energy requirement for volatile
gasication and for rising the temperature of the particles. Therefore, the corrected ame temperature T 00f can be estimated using
energy conservation as shown in Eq. (11).
h
i n_
X
air
n_ s C s T 00f T u Lv =t;
C p nproduct
Q_ T 00f T u
4:76
11
12
Lv vm000
v V d DhCH4 ;
13
where the fraction, v, is assumed to be 0.01, as suggested by Seshadri et al. [12], DhCH4 is the heat of combustion of methane, and Vd
represents the volume of the devolatilization zone (conical space
with a thickness of d as shown in Fig. 6), given by,
Vd
2
2
1
b
1
d
b
d
ph p h
:
3
sina 2 cosa
4 3
14
9
The total heat released during the complete combustion, for (//
2) mole of methane or 4.76 mol of air, is solved by energy conservation as follows:
T b T u
C p nproduct ;
T 00f
n_ air
4:76
Q_ Lv =t r
C p nproduct n_ s C s
Tu:
15
Using Eq. (15), T 00f is calculated for several cases and is plotted in
Fig. 8. As observed in Fig. 8, the ame temperature reduces due to
heat sink effects, as the concentration of dust increases for all three
equivalence ratios.
Fig. 8. Corrected ame temperature considering the heat sink effect by coal
particles.
T f T 0f T 00f =2:
The corresponding ame temperature is then used to estimate
Su using the model developed by Seshadri et al. [12].
Su
s
2Bke2
E
;
exp
RT f
qC total
16
where
e 1=Z e ; and Z e
ET f T u
RT 2f
The values of B and E are chosen as 3.5 106 mol1 s1 and
88,800 kJ mol1 respectively to match the calculated burning
velocity with burning velocity obtained by experiments for ames
without dust. The constants, A = 0.034 g/m2K s and n = 1.1, were
introduced in eq. (2) and used to evaluate T 0f .
5. Results and discussion
5.1. Validation
The calculated burning velocities are compared with the experiment data in Fig. 9. Overall, the model captures the measured
experimental trends reasonably well. Considering the simplications in the model, an exact agreement is impossible, however, a
good qualitative agreement has been obtained. Further, for 0
25 lm coal particles the model predicts Su is promoted as shown
by the dotted line (theory) that intersects the solid gray triangular
symbol (experimental data point) in Fig. 9. Further, it should be
noted that for the same size range of 7595 lm, the reduction in
Su by injecting the coal particles is lesser than the case where sand
particles are injected. This observation demonstrates that the heat
sink effect is partially compensated by the increase in the overall
equivalence ratio due to gasication. However, in the case of the
sand particles only the heat sink effect plays a role.
It is shown in Fig. 9 that the decreasing trend in the slope of Su
for 5363 lm coal particles is lesser than that with the particles in
the size range of 7595 lm. It indicates that as the size of coal par-
2455
Fig. 10. Burning velocity with coal particles of different sizes at different supply
equivalence ratios.
Fig. 11. Change of overall / due to particle injection calculated for three supply
equivalence ratios and different particle sizes.
2456
and heat transfer processes in the turbulent regime. The devolatilization process will also be enhanced due to the increased
residence time for the particle present within the turbulent eddies.
As a result, the ame velocity will be further promoted. A deagration to detonation transition will become likely.
6. Conclusions
Fig. 12. Average ame temperature calculated using combined effects of increase in
ame temperature due to increase in overall equivalence ratio and decrease in
temperature due to heat loss to coal particles as a function of dust concentration.
the smallest particle size range (Fig. 11). However, there is a notable decrease in the average ame temperature as shown in Fig. 12.
In general, higher gasication from smaller particles leads to an increase in the overall equivalence ratio. Based on the particle size
and the dust concentration, which affects the mixture specic heat
value, if the increase in the overall equivalence ratio is such that it
remains less than the slightly rich value where Su is the maximum
for gaseous ames, then the increasing trend in the Su can be preserved. However, for the supply equivalence ratio of unity, increase
in the overall equivalence ratio due to the injection of smaller particles is such that its value goes to the richer side, where the heat
release, the ame temperature, and hence, the laminar ame speed
show a decreasing trend. This is shown in Fig. 11, for / = 1 case
having smallest particle dust injected, where the overall equivalence ratio shoots up beyond 1.05 (approximate / at which Su is
the maximum for gaseous ames), around the dust concentration
of approximately 100 g/m3 and a notable decrease in ame temperature is also observed in Fig. 12 beyond this concentration.
Apart from this, the value of the mixture specic heat, which depends on particle size and dust concentration, also plays a vital
role.
The explosion in coal mine tunnels evolves through three
stages: laminar ame, turbulent ame and detonation. The Bunsen
burner type experiment setup in this study facilitates a better
understanding of interaction of coal particles and ame at the rst
stage of the explosion. As shown in this study, small coal particles
(025 lm) have the ability to enhance the laminar burning velocity
of methaneair mixture under certain conditions. The initial pressure blast of methaneair explosion (re-damp) in a coal mine can
activate the coal dust deposited over various surfaces. Small particles can be easily suspended in air as compare to the larger ones.
Therefore, a ame initiated due to hot spots will rst interact with
smaller particles coarsely suspended. This study shows that if the
ame interacting with these particles has been initiated by a lean
fuelair mixture, then there is a good possibility for the ame
velocity to be enhanced by the interaction of small coal particles.
This initial enhancement in the burning velocity could grow to a
turbulent condition. At this stage, more dust particles of various
sizes can be drawn into the ame due to enhanced mixing, mass
J.B. Greenberg, A.C. McIntosh, J. Brind, Proc. Roy. Soc. 457 (2001) 131.
S. Suard, P. Haldenwang, C. Nicoli, C.R. Mec. 332 (2004) 387396.
D.B. Graves, J.O.L. Wendt, Proc. Combust. Inst. 19 (1982) 11891196.
L.D. Smoot, M.D. Horton, Prog. Energy Comb. Sci. 3 (1977) 235258.
J.L. Krazinski, R.O. Buckius, H. Krier, Prog. Energy Comb. Sci. 5 (1979) 3171.
S.E. Slezak, R.O. Buckius, H. Krier, Combust. Flame 59 (1985) 251265.
R.K. Eckhoff, Dust Explosions in the Process Industries, third ed., Gulf
Professional Publishing, Boston, 2003.
K.L. Cashdollar, J. Loss Prev, Proc. Ind. 13 (2000) 183199.
Y. Xie, V. Raghavan, A.S. Rangwala, Powder Technol. 213 (2011) 199201.
J.T. Botz, C. Loudon, J.B. Barger, J.S. Olafsen, D.W. Steeples, J. Kansas Entomol.
Soc. 76 (2003) 426435.
P.R. Solomon, M.B. Colket, Proc. Combust. Inst. 17 (1979) 131143.
K. Seshadri, A.L. Berlad, V. Tangirala, Combust. Flame 89 (1992) 333342.
L.D. Smoot, P.J. Smith, Coal Combustion and Gasication, Plenum Press, New
York, 1985.
C. Morley, 2005, Gaseq <http://www.c.morley.dsl.pipex.com/>.