Escolar Documentos
Profissional Documentos
Cultura Documentos
ID: 14LF-0188
Abstract: The majority of theories associated with the variety of flame acceleration scenarios are based on the
"geometrical formulation: namely, the wrinkled to planar flame speeds ratio, Sw / SL , is evaluated as the scaled
increase in the flame surface area, while the entire combustion chemistry is immersed in the thermal-chemical
parameter S L , which is assumed to be constant. However, in the practical reality, S L may experience spatial and
temporal variations; at least, due to the associated pressure and temperature distribution within a combustor, and
their evolution during the combustion process. In contrast, in the present work, we initiate the systematic study of a
much more intriguing situation: when S L variations are externally imposed in a manner being a free functional of
the formulation. This is relevant, in particular, to multi-phase combustion in dusty environment, with a non-uniform
distribution of combustible and/or inert dust. Specifically, linear, parabolic and hyperbolic spatial S L - distributions
are incorporated into the theories of flame acceleration due to wall friction [Physical Review E 72 (2005) 046307]
and the finger flame shape [Combustion and Flame 150 (2007) 263]. In the case of dust, the Seshadri formulation
for the planar flame [Combustion and Flame 89 (1992) 333] is employed. Revisited formulae for the flame speed,
acceleration rate, flame tip position and velocity profile are obtained.
Keywords: planar flame speed; corrugated flame speed; flame acceleration; wall friction; finger flames; dust
combustion
1. Introduction
A planar premixed flamefront propagates with a speed, S L , which solely depends on the thermalchemical properties of the fuel mixture, irrespective of the configuration and hydrodynamics.
However, such a flame occurs very rarely in reality. Indeed, the majority of industrial and
laboratory flames are corrugated, due to turbulence, acoustics, shocks, combustion instabilities,
wall friction, in-built obstacles etc. A curved flamefront has a larger surface area relative to a
planar one, involving the same mixture and under the same conditions; therefore, it consumes
more fuel per unit time and releases more heat, thereby propagating faster than the planar
flamefront. Consequently, a flame should accelerate with a continuous increase in the flame
surface area.
There are various mechanisms of flame acceleration. Shelkin[1], associated flame acceleration
with nonslip boundary conditions of the walls. Namely, as a flamefront propagates from a closed
tube/channel end, the burning matter expands; it pushes a flow of the fresh fuel mixture; friction
ID: 14LF-0188
at the pipe walls makes the flow non-uniform, which bends the flame front, increases the flame
propagation velocity and reduces flame acceleration. On the basis of this idea, a criterion of
flame acceleration has been proposed, according to which any realistic flame with large density
drop at the front is expected to accelerate from a closed pipe end [2]. However, friction at the
viscous walls is not the only factor to increase in flame velocity. For instance, a flame shape also
gets corrugated due to the flame interaction with turbulent vortexes, intrinsic flame instabilities,
ignition peculiarities, flame acoustic interactions. etc.; and all these phenomena may play a
supplementary role for the Shelkin scenario.
One more mechanism of the flame acceleration was first suggested by Clanet and Searby [3], and
the idea was subsequently developed into a theory [4]. Here, a flame propagating in a cylindrical
tube of radius R, with ideally slip adiabatic walls was considered. One end of the tube is closed;
the flame is ignited at the symmetry axis at the closed end. In that case the flame front develops
from a hemispherical shape at the beginning to a finger-shape. Surface area grows
exponentially in time and growing is fast, but only for a short time, until the flame skirt touches
the side wall of the pipe.
The majority of theories associated with the flame acceleration scenarios, including the
formulations [2,4], are based on the "geometrical formulation: the wrinkled to planar flame
speeds ratio, S w/ S L , is evaluated as the scaled increase in the flame surface area, while the
entire combustion chemistry is immersed in S L , assumed to be constant. However, in the
practical reality, S L may experience spatial and temporal variations; at least, due to the
associated pressure and temperature distribution within a combustor, and their evolution during
burning.
Nevertheless, this phenomenon is an outcome of the combustion process and, in principle, it
may be predicted. In contrast, in the present work, we initiate the systematic study of a much
more intriguing situation when S L variations are externally imposed in a manner being a free
functional of the formulation. For instance, this is relevant to the case of spatial variations of the
equivalence ratio as well as to the multi-phase combustion in dusty environment, with a nonuniform distribution of combustible and/or inert dust. In particular, to account for the inert dust
particles, the Seshadri formulation [5] is employed. According to Ref. [5], the upstretched
laminar flame speed in a dusty-gaseous premixture S L,d can be expressed as
S L,d S L ,0
1
C 4 rs2 s
1 S
ns
CP 3
(1)
where S L,0 is associated with the respective gaseous combustion, and are heat capacity of
gas and particles, respectively, is the mean particle radius, is the concentration defined as
the number of particles per unit volume. For the spatial variation of the flame velocity, various
non-uniform dust distribution gradients can be employed into concentration term, , such as a
linear, hyperbolic, parabolic etc. functions as a function of tube in radial direction, (). We do
not discuss these effects in the present paper because of limited space.
ID: 14LF-0188
In this work, various spatial distributions for the planar flame speed are incorporated into the
theories [2, 4]. As a result, the revisited formulas for the total flame speed, acceleration rate and
the flame-generated velocity profile are obtained.
2. Formulations
In this study, we consider three different variable velocity gradients, the linear, parabolic and
hyperbolic velocity distributions, which are given as
(2)
S L a0 a1 x ,
SL SL
max
x 2
1 ,
R
(3)
x
tanh(
a
(1
))
R ,
max
SL SL
tanh(a)
(4)
respectively, where a0 , a1 , a are constants. These three distributions are also shown in Fig. 1.
(-R,0)
(-R,0)
(0,0)
(0,0)
(SL,max)
r
(R,nmax)
(R,0)
(R,0)
(a)
(b)
(-R,0)
(SL,max)
(0,0)
(R,0)
(c)
ID: 14LF-0188
''burnt matter''
''fuel mixture''
z
R
(5)
t
f z
dx 2
The kinematic viscosity is coupled to the Reynolds number related to the flame propagation,
RS Lmean
R
.
Re
(6)
Pr L f
It is noted, that the conventional Reynolds number, associated with the flow significantly
exceeds the quantity (6). The pressure term in Eq. (5) is a function of time as K (t ) P / z . By
considering exponential regime of flame acceleration, Eq. (5) becomes
(7)
2 CK ,
where Re R / S L x , and we account for the velocity profile in the form
u z x, t x exp S L ( x)t / R ;
0 at x R .
cosh cosh( x )
R .
cosh 1
(8)
(9)
cosh cosh( x )
R .
cosh 1
(10)
(11)
cosh cosh( x )
R .
cosh 1 sinh
(12)
F
F
uz (0, t ) uz ( x, t ) S L ( x) S L ( x) 1 .
t
x
2
(13)
ID: 14LF-0188
Since flat region does not affect the analytical theory, Eq. (13) simplifies to
F
F
uz (0, t ) uz ( x, t ) S L ( x)
t
x
(14)
.
The burning rate is proportional to the flame amplitude and can be written as
u F ( R, t ) .
Then Eq. (14) takes the form
F
F
.
uz (0, t ) uz ( x, t ) S L ( x)
t
x
(16)
(17)
Substituting Eqs. (12), (15) and (17) into (16) we find the flame equation as,
( R)( 1)
x
( x) ( x)
(cosh
1).
1
S L ( x) R(cosh sinh )
R
In fact, Eq. (18) exhibit the form
(18)
Q( x)
P( x)
(15)
(19)
SL ( x ) dx R
SL ( x ) dx
( x)
(20)
e 0
0 AQ( x)e 0 dx.
( R)
While the analytical solution of Eq. (18) was not possible by any symbolic integration software
such as MATLAB and WOLFRAM, Eq. (20) has been solved by discretizing the integrants
in given intervals symbolically. Then, we further apply the trapezoidal method to calculate the
integration over the given intervals. The resulting nonlinear equation is further solved by
implementing Newton-Raphson method for .
''skirt''
z
''tip''
ID: 14LF-0188
1 (rur ) (ur )
0.
r (r )
( z )
(21)
The boundary conditions at the end-wall z = 0 is u z 0. Assuming the axial velocity at the flame
front which is labeled as 1 (fuel mixture),
(22)
u A (t ) z ,
z ,1
A1 (t ) R 2 r 2
.
2
r
(23)
A2 (t )
r and
2
u z , 2 A2 (t ) z .
(24)
drf
dt
ur ,1 S L (rf ), ur ,1 ur ,2 ( 1) S L (rf ), uz ,1 uz ,2 .
(25)
A1 A2 2( 1) S L (rf )
rf
(26)
),
(27)
R2
drf
dt
S L (rf )( ( 1)
rf
R2
After incorporating Eqs. (4), (5) and (6) for S L r f R f into Eq. (27), we can solve by using
numerical integration and developed method as mentioned in the previous section. We can also
find evolution equation for the flame tip by considering the flow along the axis, namely
(28)
dztip
rf z
dt
2( 1) S L (rf )
S L0 ,
43.7 cm/s is the maximal flame speed and 8 , providing realistically slow flame
where,
propagation as compared to the sound speed (the Mach number is about 10-3). We know from
the solution to Eq. (27) that t f (r f ) . In equation (28), the term can be replace with
S L0
2( 1) S L (rf )
rf
R
ztip S L0 f (rf ).
P ( rf )
(29)
K ( rf )
(30)
ID: 14LF-0188
ztip
K ( r )e
(r )
f
P ( rf ) drf dr
P ( r f ) dr f
e
(31)
f
15
Linear
Constant
Hyporbolic
10
0
0
20
40
Re
60
Fig. 4: The scaled acceleration rate versus the Reynolds number for constant, linear and hyperbolic cases
For the finger flame acceleration mechanism, by solving Eq. (27), we obtain evaluation of the
flame front in time for linear, parabolic and hyperbolic velocity gradients, which is illustrated in
Fig. 5. In this figure, the scaled flamefront radius increases in time as expected for three velocity
variations. However, this increase is stronger in the case linear velocity distribution. We also
consider three different radius length.
1
R=20lf
0.4
R=30lf
0.2
0.6
R=20lf
0.8
R=30lf
0.6
rf/R
R=10lf
0.6
0
0
1
R=10lf
0.8
rf/R
rf/R
0.8
0.4
0.2
0.5
1
t(s)
1.5
-3
x 10
0
0
R=10lf
R=20lf
R=30lf
0.4
0.2
2
t(s)
3
-3
x 10
0
0
0.5
1
t(s)
1.5
2
-3
x 10
Fig. 5: The scaled flame front versus time for tubes of radius for linear, parabolic and hyperbolic velocity
distributions at = ,
ID: 14LF-0188
Equation (28) can be solved by a developed MATLAB code and discretization method. In Fig.
6, we can see the scaled flame tip position versus time for the different tubes of radius and
velocity distribution gradients. Also by taking the derivative of the with respect to time, we
obtain flame velocity with respect to burnt matter which is an important parameter.
20
15
15
R=10lf
R=20lf
10
R=10lf
R=20lf
10
R=30lf
Ztip/R
10
Ztip/R
Ztip/R
15
0
0
0
0
R=10lf
R=20lf
R=30lf
0
0
0.5
1
t(s)
1.5
-3
x 10
t(s)
R=30lf
0.5
1
t(s)
-3
x 10
1.5
2
-3
x 10
Figure 6: The scaled flame tip position versus time for tubes of radius for linear, parabolic and hyperbolic
velocity distributions at = ,
12
10
30
20
R=10lf
10
R=20lf
0
0
R=30lf
0.2
0.4
0.6
0.8
t(s)
1
-3
x 10
dZtip/dt
dZtip/dt
40
12
R=10lf
R=30lf
8
6
4
2
0
R=10lf
10
R=20lf
dZtip/dt
50
R=20lf
R=30lf
8
6
4
2
t(s)
3
-3
x 10
2
0
0.5
1
t(s)
1.5
2
-3
x 10
Figure 7: Velocity of the flame with respect to burnt matter versus time for tubes of radius for linear,
parabolic and hyperbolic velocity distributions at = ,
The change of flame velocity, due to different velocity distributions and radius in time, can be
seen in Fig. 7. The flame surface area increases in time as a result of thermal expansion.
Increasing the surface area absorbs more fuel and increases flame velocity. It can be seen that the
increase of flame velocity over time is particularly strong for a finger flame mechanism.
4. Conclusions
In this study, the various spatial S L distributions (linear, parabolic and hyperbolic) are
incorporated into the Bychkov theories of flame acceleration due to wall friction [2] and finger
flame shape [4]. Revisited formulae for the flame speed, acceleration rate, velocity profile, flame
tip position and flame front equations are obtained.
5. Acknowledgements
This work is supported by the Alpha Foundation for the Improvement of Mine Safety and
Health as well as West Virginia Universitys Program to Stimulate Competitive Research
(PSCoR).
ID: 14LF-0188
6. References
[1] Shelkin, K., J. Exp. Teor. Phys. 10 (1940) 823.
[2] Bychkov, V., Petchenko, A., Akkerman, V., Eriksson L. E., Phys. Rev. E, 2005. 72: p. 1-10.
[3] Clanet, C., Searby, G., Combust. Flame, 1996. 105: p. 225238.
[4] Bychkov, V., Akkerman, V., Fru, G., Petchenko, A., Eriksson L. E., Combust. Flame, 2007.
150: p. 2632376.
[5] Seshadri, K., Berlad, A.L., Tangirala, V., Combustion and Flame, 1992. 89: p. 333-342.
[6] Qiao, L., Xu, J., Combust. Theory Modell. 2012. 16(5): p. 1-27.