Beruflich Dokumente
Kultur Dokumente
We consider properties of a steady two-dimensional isothermal low Mach-number counter-flow, into which a
dilute loading of small spherical particles is introduced at the local gas velocity, at a finite axial distance from
the stagnation plane for the axial velocity component of the gas. The particles are introduced on one side of that
stagnation plane only, and the consequences of any subsequent velocity slip of the particles with respect to the
local gas are examined. The self-similarity (planar symmetry in the axial coordinate, for most of the key
dependent variables), familiar from particle-free counter-flow, also holds for the two-phase flow under these
conditions. Results, obtained by Lagrangian tracking of the motion of a single particle, distinguish: the
nonoscillatory trajectory of that particle for relatively small strain-rate, large drag-rate conditions (the particle
does not cross the stagnation plane for the axial velocity component); and the oscillatory trajectory of that
particle under relatively large strain-rate, small drag-rate conditions (the particle may cross the stagnation plane
repeatedly). However, for the multi-particle scenario for self-similar two-phase flow, the results for both
conditions have commonalities. A single, densely particle-loaded, very thin slab region arises: one planar side
of the slab interfaces with a particle-free, purely gaseous counter-flow, and constitutes the axial stagnation plane
for that flow; the other planar side of the slab interfaces with a dilutely particle-laden region, in which the
particle behavior is unaltered by the presence of the close-packed thin slab. In fact, the thin slab effectively is
the stagnation plane for the axial velocity component of the gas, the value of the strain-rate/drag-rate ratio
characterizing whether the plane is displaced from its pure gas counter-flow position (and, if so, to what
different axial position). 2001 by The Combustion Institute
NOMENCLATURE
a
D
L
l
m
t
t*
t1
u
V
v
w
x
x(t)
x(t)
Y
y
y(t)
y(t)
Greek Letters
1
2
3
(t)
(t)
[1 4(/)]1/2
[1 4(/)]1/2
[4(/) 1]1/2
Stokes drag rate, 6a/m
multiple to which a line segment of
simultaneously entering particles is
stretched in the y direction at time t
time rate of change of (t)
dynamic viscosity of fluid
mass of particles per volume of space,
at the entry plane
COMBUSTION AND FLAME 126:1630 1639 (2001)
2001 by The Combustion Institute
Published by Elsevier Science Inc.
s
(t)
INTRODUCTION
Counter-flow configurations have played an important role in both the theoretical understanding and the gathering of experimental data for a
variety of combustion-related phenomena. We
analyze here the transport of initially sparse
distributions of particles in such flows. We deal
with steady counter-flows, and the phrase initially sparse distributions refers to the number
density of particles at their sites of introduction
into the flow; wherever there is no large variation from the initial sparseness, the consequences of particle collisions are negligible.
Although our analysis is broadly applicable,
our work is particularly motivated by phenomena such as soot formation in hydrocarbon/air
counter-flow diffusion flames [1] and metal oxide particle formation (as feedstock for ceramics) in counter-flow diffusion flames [25].
Wooldridge [4] cites the counter-flow-diffusionflame apparatus as one of six basic experiments
especially suitable for examining the gas-phase
combustion synthesis of particles. Scenarios in
which fuel-droplet sprays are evaporated to
form a combustible vapor, such that the droplets do not persist to the vicinity of the gas axial
velocity component stagnation plane, have been
discussed [6 8], and do not constitute the major
motivation for our work.
A particular relevant context concerns a proposed modification in the schedule (duty cycle)
for transverse injection, into the hydrogen-rich
flow in the single nozzle of Stage 3 of the
Minuteman III missile, of a helium-pressurized
aqueous solution of the oxidizer strontium perchlorate Sr(ClO4)2. The liquid-injection transverse vector control (LITVC) system, in generating side force for attitude control (and some
modest thrust augmentation as well), may incur
1631
1632
reasonable anticipation is that planar similarity
holds extensively (if not everywhere), and a
transversely invariant, temporally unchanging
flux of particles enters the counter-flow.
Finally, we take the particles to be introduced
in the counter-flow exclusively on one side of
the stagnation plane for the axial velocity component; of course, in the presence of slip, the
particles introduced on one side of the stagnation plane may or may not remain confined to
that side of the stagnation plane. Any crossing
of the stagnation plane is here not related to
thermophoresis, which is not present in our
isothermal flow; neither is any other diffusive
transport of particles taken to enter. Of particular interest is the possible existence, and location if it exists, of a dense (typically thin) layer
of particles separating particle-free regions of
the counterflow from only dilutely particleladen regions of the counter-flow; such a dense
layer may arise as a consequence of axial-flow
stagnation and finite interphase velocity slip.
Suggestions from theory of the existence of
singularly large particle number densities very
near the axial gas velocity stagnation plane for
smaller values of the Stokes number (the Stokes
number being here defined as the ratio of the
strain rate of the gaseous flow to a rate that
increases with the drag on particles) have not
been verified experimentally [9, 10]; further
study seems warranted to resolve the uncertainty [9, 10]. Furthermore, for larger values of
the Stokes number, single particle-tracking
models and experiments yield trajectories with
multiple reversals of direction [9, 10, 1517];
these results have been interpreted, for a multiple particle context, to imply the coexistence of
oppositely directed particles, so that analytic
complications arise [12]. Clarification again
seems warranted [17].
G. F. CARRIER ET AL.
Fig. 1. Sketch of the streamlines of a steady two-dimensional low-speed isothermal gaseous counterflow (solid
curves), with a stagnation plane (for the axial velocity
component u) located at axial-coordinate position x 0.
Small spherical particles are continuously introduced with
the local gas speed at the flow-domain boundary x l. The
trajectories of the particles introduced at the end points of
a differential line segment dY at x l are shown (dashed
curves), with the y-direction separation increasing to dY
when the particles subsequently lie at an (algebraically)
larger value of the axial coordinate x. No particles are
introduced in x 0, and no particles enter that region for
small-strain-rate, large-drag-rate conditions.
u x, v y in x l, y
(1)
where x,y denote Cartesian coordinates: denotes a constant strain rate; and l denotes a
positive number (Fig. 1). We denote the position of a single particle by x(t),y(t), where t
denotes time since the particle entered the flow
field defined by Eq. (1). We adopt the initial
particle position and initial particle velocity so
that initially there is no slip of the particle with
respect to the gas:
SINGLE-PARTICLE DYNAMICS IN A
COUNTER-FLOW
x0 l, y0 Y, x 0 l, y 0 Y
(2)
(3)
(4)
x
[exp ()] cosh(1)
l
1 2(/)
sinh(1)
1
l x xt 1,
cosh(1)
for
(5)
x
exp ()]
l
1
sinh(1)
1
(6)
y
[exp()] cosh(2)
Y
1 2 /
sinh(2)
2
(7)
y /Y
exp()] cosh 2
1
sinh 2
2
(8)
1 1 4/
1/ 2
, 2 1 4/
t/ 2 t/ 2//
1/ 2
,
(9)
1633
x
exp() cos 3
l
1 2/
sin 3
3
x
1
exp() cos 3
sin 3
l
3
(10)
(11)
where:
3 4/ 1 1/ 2
(12)
(13)
x
exp(2t] 1 2t
l
(14)
ol Y t x t Y t,
(15)
1634
G. F. CARRIER ET AL.
t
l
o
x t t
(16)
(17)
sV 2X Y olY
t 1,
(18)
sVX olY,
(19)
sV 2X ol
t 1 Y 2/ 2.
(20)
o lY o 2l
,
s V
s
t 1
or, alternatively:
(21)
(22a)
o X
t 1
.
s
2l
(22b)
The above integral-like statements concerning conservation of mass and momentum for the
dense layer would not be significantly altered if
there were fewer but larger particles in the
dense layer. The similarity of the flow in the
dilutely particle-populated region is what sets
the spatial rate of incrementation of the dense
layer, and the dilutely particle-populated region
is not likely to be altered by some aggregation
among particles already in the dense layer.
LOW STRAIN-RATE, HIGH DRAG-RATE
CASE
For the high drag-rate, low strain-rate case
(/) (1/4), with the aid of Eqs. (59), we
anticipate that the dense layer lies contiguous to
the stagnation plane for the axial velocity
component, x 0; i.e., the dense layer lies in
x(t1) x 0, where: x(t1) 0;
X/l xt 1/l;
(23)
t 1 V t 1
,
,
, and t 1,
Y
s
(24)
1635
1636
G. F. CARRIER ET AL.
(25)
Fig. 6. The normalized time *(t*/2) to the first stagnation of axial motion of a single particle vs. the Stokes
number (/) for the large-strain-rate, small-drag-rate
circumstances [(/) (1/4)]. A particle that sooner
travels into the region of stronger opposed flow is stalled
earlier.
(26)
3 * h*
(27)
then:
tan h* 3, h* arc tan 3
(28)
so (Fig. 6):
* arc tan 3/ 3
(29)
x *
1/ 2
exp(*)
(30)
yt 1
t 1 y t 1 t 1 V
X xt 1 x t 1
l,
l , l , t 1 Y , Y , s , Y , and t 1
from the eight relations Eqs. (7, 8, 10, 11, 16, 21,
22a, 25). As previously, for convenience, we
operationally regard ( o / s ) to be deduced and
t1 as specified, and replace Eq. (22a) with its
equivalent, Eq. (22b).
Results, again obtained by assigning a succession of values to (/) (1/4), for each of
which a range of values for t1 is adopted, are
given in Figs. 8 to 10. As is to be expected, much
of the behavior of the dependent variables is
seen to be about the same as in the previously
treated case for small strain rate and high drag
rate. In fact, the key difference is that the
gas-flow stagnation plane (for the axial velocity
component) is now at [x(t1) X], not at x 0
as before (Fig. 11). This translation of axial
coordinate is to be accounted for in writing
expressions for the purely gaseous flow in the
domain x [x(t1) X]. Because the dilutely
particle-laden domain extends over the expanse
l x x(t1), which interfaces with a thin
1637
(31)
dense layer, the scenario holding for the multiparticle counter-flow for (/) (1/4) is, in
many respects, an axially displaced version of
the scenario holding for the multi-particle
counter-flow for (/) (1/4).
For an experimental study of the multi-particle counter-flow described in this manuscript, it
is noted, concerning the x-momentum flux law,
that the momentum of gas and particles crossing
x l is equal to momentum of gas crossing
x L{[x(t1) X]}. However, any momentum balance-motivated adjustment of pressure,
to account for the dilute particle loading of only
one of the two opposed streams constituting the
counter-flow, probably would be of modest
magnitude.
1638
G. F. CARRIER ET AL.
CONCLUSION
We have considered particle transport in a
steady two-dimensional low Mach-number
counter-flow. More specifically, we have addressed scenarios in which the planar symmetry
(similarity) familiar from purely gas-phase contexts is extended to a two-phase context, such
that particles persist to the stagnation plane of
the axial velocity component. Examination of
single particle trajectories in a counter-flow
provides a convenient, informative introduction
to the phenomena; such studies distinguish: 1)
small strain-rate, large drag-rate circumstances
(in which no reversal of direction of a relatively
small particle is observed), from 2) large strainrate, small drag-rate circumstances (in which
typically multiple reversal of direction of a
relatively large particle are observed). However,
description of the transport of multiple particles
in a self-similar counter-flow goes beyond Lagrangian tracking, and identifies the existence of
a densely particle-laden layer(s) even in a
counter-flow in which particles are introduced
with dilute number density at the local gas
velocity holding at a flow-domain boundary.
We have considered nonvaporizing monodisperse spherical particles introduced at modest
number density and with no slip with respect to
1639
2.
3.
in a laminar counter-flow. However, we reiterate our belief that the reason that the dense
layer is challenging to detect seems related
primarily to its thinness, rather than to other
cause, such as instability phenomena. If the
dense layer of particles were unstable in the
counter-flow, the resulting thicker (if somewhat
less dense) layer would be expected to be more
readily detectible. The fact that the dense layer
has not been detected is consistent with the
layer being thin and remaining stable. Also, the
time of residence of a laterally flowing particles
in the dense layer (within a counter-flow apparatus of feasible dimension in the presence of
Earth gravity) is to be compared with the time
for any instability to grow to finite amplitude,
and thus be of practical consequence. Hence,
even were an instability to arise, the experimental evidence suggests that the instability is of no
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.