Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Powder Technology, 52 (1987) 227 - 232 227

The Effect of Particulate Solids on Pressure Drop Across a Cyclone

J. M. BEECKMANS and B. MORIN


Faculty of Engineering Science, The University of Western Ontario, London N6A 5B9, Ont., (Canada)
(Received January 2, 1987; in revised form April 29, 1987)

SUMMARY the cyclone was deep and nearly empty.


On the basis of visual observations in a
Using relatively coarse (over 0.1 mm) transparent cyclone, they attributed this
particles, it is shown that a cyclone is capable phenomenon to “a small quantity of solids
of trapping an inventory of solids, which leak that remained suspended in the vortex”.
to the dipleg and also, more slowly, to the gas This observation is not concordant with Yuu
outlet. The inventory of solids has a et al. ‘s conclusion that deposited particles are
maximum value, which depends on the size the cause of the observed reduction in pres-
and shape of the particles. It was found that sure drop across the cyclone, and it is difficult
at the limit of zero solids inventory the to understand how particles can remain
cyclone pressure drop reduction due to the suspended in the vortex without a simulta-
presence of solids is proportional to the neous process of deposition and reentrain-
projected area of the trapped solids, the ment from the cyclone surfaces.
proportionality constant being a function of The objectives of the present study were to
particle shape. confirm the conclusion of Yuu et al. that
particles deposited at the wall cause a consid-
erable reduction in pressure drop across the
INTRODUCTION cyclone, and to obtain additional information
concerning this phenomenon.
It is known that the pressure drop across a
cyclone is lowered by the presence of solids. EQUIPMENT AND PROCEDURES
The reduction in pressure drop was shown by
Yuu et al. [l] to be due primarily to a reduc- A sketch of the 0.30-m diameter cyclone
tion in tangential gas velocity within the used is shown in Fig. 1, and Fig. 2 gives an
cyclone, caused by increased friction at the overall view of the experimental apparatus.
walls due to the presence of deposited parti-
cles. Yuu et al. demonstrated that the tangen-
tial gas velocity could also be reduced, in the
absence of particles in the gas, by coating the
internal surfaces of the cyclone with a layer
of sticky liquid and subsequently depositing
particles on to this layer. They concluded that
the dust layer on the wall, not the dust in the
gas, caused the reduction in pressure drop.
Littlejohn and Smith [2] observed that the
flow rate across a cyclone at constant pressure
\I
1
drop increased when solids were introduced
in the entering air; however, the increased
Vl
air flow continued when solids flow into
the cyclone was discontinued and was only
restored to its original clean air value +d v2

following a brief interruption in the flow, Fig. 1. Schematic of the cyclone: I, Inlet; 0, gas out-
and only provided that the hopper beneath let; D, dipleg; Vl, V2, valves.

@ Elsevier Sequoia/Printed in The Netherlands


caused by solids was maintained when the
flow of solids into the cyclone was shut off
was confirmed, and it was obvious from the
sound emanating from the cyclone that solids
were trapped and were responsible for the
reduced flow resistance. This provided an easy
method for studying the effects of solids on
flow resistance, since the solids could be
collected and weighed by shutting off the air
Fig. 2. Schematic of the apparatus: B, Blower; C, flow to the cyclone. Two types of experiment
cyclone; D, air flow control valve; S, air and solids were performed, one type designed to deter-
entry point; V, venturi; P1, Pz, Pa, pressure taps.
mine the rate of loss of trapped solids over
time, and their fate, the second type to mea-
Two valves, separated by a 15-cm length of sure flow resistance as a function of the
tube, were located in the dipleg at the base weight and nature of the trapped solids.
of the cone. Air was drawn through the sys- Experiments were made using various solids
tem by a fan, the flow rate being measured by whose properties are shown in Table 1. The
a venturi meter placed downstream from the air velocity at the inlet to the cyclone was
cyclone. The gauge pressures at the cyclone maintained at 14.4 m/s in all runs, the
inlet, and at the inlet and throat of the pressure drop across the empty cyclone being
venturi were measured using water manom- 2.56 kPa.
eters. Note that the pressure drop across the
cyclone was taken as the difference in
readings of pressure taps PI and Pz (Fig. 2). RATE OF DECAY AND FATE OF TRAPPED
Although losses in an elbow and some pipe SOLIDS
are included in the overall pressure drop, this
does not affect the results, which are A measured quantity of solids was placed
concerned with the reduction in pressure in a beaker which was then upended so that
drop, at constant air velocity, due to the the solids dropped into the air stream flowing
presence of solids in the cyclone. to the cyclone. The air flow rate was im-
All measurements on cyclone pressure drop mediately adjusted back to its original value
were made without solids in the entering air. using a valve placed between the venturi
The observation of Littlejohn and Smith [2] and the fan, and the upper valve under the
that the frictional resistance of the cyclone cyclone was shut off (these manoeuvres took

TABLE 1
Physical properties of particulate solids used

No. Material Shape Density pp Size range Mean size (ip


(kg/m31 (mm) (mm)

1 Sand Rounded 2650 0.18 - 0.25 0.22


2 Sand Rounded 2650 0.25 - 0.355 0.29
3 Sand Rounded 2650 0.355 - 0.425 0.38
4 Sand Rounded 2650 0.425 - 0.50 0.46
5 Sand Rounded 2650 0.50 - 0.60 0.55
6 Sand Rounded 2650 1.0 - 1.18 1.09
7 Sand Angular, 2650 0.36 - 2.8 2.12a
irregular
8 Sand Rounded 2650 0.11 - 0.60 0.247a
9 Glass Spherical 2800 1.0 - 1.7 1.13a
10 Salt Cubical 2130 0.425 - 0.50 0.46
11 Carbon Irregular 900 0.3 - 1.18 0.87*

aHarmonic mean diameter: _ 1


d,= -
%I/$ il
229

at most 30 s). The solids beneath the cyclone The rate of leakage of solids out the exit
were then removed by opening and closing duct was found to obey the following equa-
the lower valve, the upper valve was reopened tion:
slowly, and the pressure drop across the
cyclone was noted. The same sequence of dL, W
-=-
steps (except that of adding solids to the air dt r1
entering the cyclone) was repeated at regular
intervals, yielding data on the pressure drop Lo equals the cumulative quantity of solids
across the cyclone and on the loss of solids exiting the outlet duct, W equals the
via the dipleg, as functions of time. These inventory of solids in the cyclone at any point
experiments showed that some or all of the in time, and r1 is a time constant. The rate of
solids entering the cyclone were retained. loss of solids at any given instant is then
If the weight of solids entering the cyclone given by
WI exceeded a critical value I+‘(‘-‘)the excess
dW dL, + dL,
(i.e., Wr - WC”‘) immediately descended into (3)
the dipleg. The value of W@) depended on the -dt=dt dt
nature and sizedistribution of the solids The following equation was found by com-
(Table 2), with a trend towards larger values
bining eqns. (1) to (3) :
of W(O)with increasing size. However, shape
was also a factor since W(O)had by far the
highest value with the spherical glass beads. dW +
dt
w + Lu(f)
r1 r2 exp
= 0 (4)
TABLE 2
Parameters pertaining to trapped solids The solution of eqn. (4) is

No. Material &, 72 Iv(O) L,(f) 71


(mm) (min) (g) (g) @I jj,7 = w(o) - TILu(f) exp - 4

t
71 - 72 i 71 I
-

+
Sand 0.22 3.7 0 Large

( i
Sand 0.29 2.4 10.1 2.1 21.4 71 L,(f)
Sand 0.38 10.6 12.0 3.2 7.4 pexp - - (5)
Sand 0.46 9.2 15.1 1.1 3.7 7-l- 72 72
Sand 0.55 2.7 18.3 4.0 5.0
Sand 1.09 5.8 18.5 2.5 6.0
Sand 2.12 43 18.4 4.0 6.6 Equation (1) was verified by measuring L,
Glass 1.13 0.72 57.3 9.5 1.3 directly as a function of time, using the two
valves at the base of the cyclone to extract
leaked solids at intervals of time without
disturbing flow in the cyclone. W was inferred
Subsequently, particles continuously from the pressure drop across the cyclone,
leaked both into the dipleg and out the exit using data on the relationship between
duct of the cyclone. Leakage into the dipleg pressure drop reduction and solids inventory,
occurred over a relatively short time span, and which are reported below. It was found that
could be described by an equation of the r1 3 r2, so that for t 9 72 eqn. (5) reduces to
following form:

L=L,“‘[l -exp[- $)I (1)


(Tl:;;;‘o’]
+-)) =ln[ -; f6)
Thus, a plot of ln(W/W(‘)) uersuS time
L, equals the cumulative weight of solids has a slope equal to (-l/ri) for t S r2.
which had leaked into the dipleg after time t, Figure 3 gives an example of such a plot,
L (f) equals the total amount of leaked solids showing the fit of eqn. (5) to the data.
(buoth exclusive of any excess solids shed in Table 2 gives values for the time constants
the first few seconds after solids were ad- 71 and r2, as well as values for Lutf) and
mitted) and r2 is a time constant. W(O), for several solids.
230

It was found that the data for a given solid


could be fitted reasonably well by a second-
order polynomial:
AP,=aW+bW’ (7)
The zero-order term is deleted because the
definition of the pressure reduction implies
0 loo ml 300 400 AP, = 0 when W = 0. The curve shown in
t hnl
Fig. 4 was obtained by least-squares analysis.
Fig. 3. Typical plot of W/W(O) (ratio scale) versus
time (0.36 - 0.43 mm sand particles). Solid line:
The constant a equals the slope of the
eqn. (5). curve at the origin; a plot of u&p, uersus
harmonic mean particle size (or arithmetic
mean size for solids with a narrow size range)
is illustrated in Fig. 5.
RELATIONSHIP BETWEEN PRESSURE DROP
ACROSS THE CYCLONE AND QUANTITY OF
TRAPPED SOLIDS

In these experiments, a known weight of


solids, suspended in the entering air, was
admitted to the cyclone. The air flow rate
was then immediately adjusted to its predeter-
mined value, the upper valve beneath the 1 I,,,,,
I
I

0.2 0.5 IO - 2.0


cyclone was closed, the pressure drop across dplmm)

the cyclone was noted, and the fan was shut Fig. 5. ad&., versus dp. 0, rounded sands; 0, angular
off. The solids found above and below the sand; m, salt; V, carbon; A, spherical glass beads.
upper valve were then collected separately
and weighed. The weight of solids found
above the upper valve equalled W, the weight DISCUSSION
of solids circulating in the cyclone at the time
the air flow was shut off. The rate of decay of trapped solids is
A typical plot of AP,, the cyclone pressure clearly governed by two mechanisms: leakage
drop reduction (i.e., the pressure drop across into the dipleg, characterized by a time
the cyclone in the absence of solids, minus constant which was usually smaller than 10
the pressure drop in the presence of trapped min (although in one case it equalled 43 min),
solids, at the same air inlet velocity), uersus and loss through the gas outlet, characterized
weight of trapped solids W is shown in Fig. 4. by a time constant measured in hours.
The mechanisms by which particles are
carried to the solids outlet do not appear to
be fully understood. It is known that cyclone
performance is relatively unaffected by its
orientation; in fact a cyclone can even be
operated satisfactorily upside down (ter
Linden [ 3]), which proves that dust which
has been collected at the walls is not carried
down primarily by gravity.
Two other mechanisms suggest themselves.
One is that particles are carried towards the
base of the cone by drag of the gas, which at
the wall has an axial component in the direc-
tion of the solids outlet [3]. The axial
Fig. 4. Typical plot of pressure reduction Al’, versus
component of gas velocity is, however, small
weight W of solids in the cyclone (0.5 - 0.6 mm sand
particles). Solid curve is a plot eqn. (7) fitted by least- in the vicinity of the wall, and is zero at the
squares. wall. Furthermore, in the cone, the particles
231

experience a centrifugal force which, when The data suggest that particle shape plays
resolved in directions parallel to and normal a significant role in the observed phenomena;
to the cone, has a component in the direction indeed in comparing results obtained with
opposite to the narrow end, and which would materials 6 and 9, which have similar mean
tend to counteract drag by the gas. This force particle size and density but have dissimilar
is, in fact, the most obvious reason for the shapes, it is seen that the parameters 71, To,
relatively long-term stability of the solids W(O)and q(O) all differ very significantly
inventory. in the two cases.
For these reasons, an alternative mecha- The rate of leakage of particles to the gas
nism is proposed, namely, that particles outlet is smallest with the smallest particles
circulating in the cone constitute a kind of (Table 2), and appears to reach a maximum
two-dimensional gas which ‘overflows’ into value at approximately 0.5 mm mean size
the dipleg when the concentration of particles with rounded, non-spherical particles. This
exceeds a maximum value. An estimate of result is consistent with the findings of Mori
the degree of coverage of the cone of the et al. [ 41, who observed that the efficiency of
cyclone by particles at the maximum value of their cyclone decreased with increasing size
W may be obtained using the following of particles, in the size range from 74 to 246
formula: pm. They attributed this phenomenon to
irregular bouncing of particles against the
(8) internal surfaces of the cyclone. The higher
leakage rate observed with glass spheres
where 17(O)equals the fraction of the surface may be due to the fact that they roll more
area of the cone occupied by particles at rapidly, due to their shape, and may therefore
W = WC"), and A, equals the area of the cone bounce further when they hit an imperfec-
of the cyclone. Equation (8) was derived by tion, such as a seam, in the wall of the
assuming that particles are spherical and that cyclone.
each particle occupies an area equal to its own The effect of q on AP, is illustrated in
projected area. Values of Q(O)expressed as a Fig. 5, since it may be shown that
percentage are given in Table 3 for various
d@r
-=_ 2ad,p,
sands and for glass spheres. These figures (9)
suggest that with rounded particles a surface dr) 34
coverage of a little over 5% is normal. The where rl equals the fractional surface coverage
figure for the spherical glass beads is larger, in the cone of the cyclone due to a mass W of
whereas with the very angular particles of trapped particles. The data in Fig. 5 thus
crushed sand (material 7) the figure is much show that in most cases, and over a broad
lower. The coverage is also somewhat smaller range of sizes, pressure drop reduction at the
for materials 1 and 6. limit of near zero solids is approximately
proportional to surface coverage. The propor-
tionality constant is significantly smaller for
the spherical glass beads, and is significantly
TABLE 3 larger for the irregular, angular crushed sand
Estimated maximum surface coverages in the cone of particles with average size 2.12 mm, and for
the cyclone the smallest particles. With the exception
of these three cases, the values of ap,i&,
No. Material (s,
r)(o) are confined to the span 2.4 X lo5 to 3.1 X
(mm) (%I lo5 Pa/m2 over a range of mean size from
0.22 to 1.09 mm.
Sand 0.22 2.7
Sand 0.29 5.5
Sand 0.38 5.0
Sand 0.46 5.2
Sand 0.55 5.3 CONCLUSIONS
Sand 1.09 2.7
Sand 2.12 1.4
Glass 1.13 7.5 The following physical phenomena were
observed:
232

(1) A cyclone is capable of holding an L


u
(0 maximum value of L,
inventory of circulating coarse solids for t time, s
significant periods of time after feeding of mr pressure drop reduction due to solids,
solids has ceased. Pa
(2) The inventory of solids decays over W inventory of solids in the cyclone, kg
time, partly to the underflow, with a rela- w(o) maximum inventory of solids in the
tively short time constant, and partly to the cyclone, kg
gas outlet, with a much longer time constant. WI mass of solids fed to the cyclone, kg
(3) There exists a maximum value for the
inventory of solids, any excess material in the Greek symbols
feed immediately descending to the under- 77 nominal fractional surface coverage in
flow. The maximum solids inventory depends cone, -
on the size and shape of the particles. n(o) maximum nominal fractional surface
(4) With particles of similar shape and for coverage, -
W < W(O), the pressure drop reduction is 71 overhead leakage time constant, s
approximately proportional to the degree of 72 underflow leakage time constant, s
nominal surface coverage in the cone, regard- PP particle density, kg/m3
less of size or density of the particles. How-
ever, the pressure drop reduction may be sig-
nificantly larger or smaller for the same ACKNOWLEDGEMENTS
degree of surface coverage with particles
having different shapes. The authors wish to express their apprecia-
tion to the Natural Sciences and Engineering
Research Council of Canada for financial
LIST OF SYMBOLS support for this work.

constant in eqn. (7), Pa/kg REFERENCES


: surface area of cone of cyclone, m2
bc constant in eqn. (7), Pa/kg2 1 S. Yuu, T. Jotaki,
Y. Tomita and K. Yoshida,
d* mean particle size, m Chem. Eng. Sci., 33 (1978)1573.
cumulative weight of solids escaping 2 R. F. Littlejohn and R. Smith, Proc. Inst. Mech.
LO
Engrs., 192 (1978) 243.
with the air up to time t, kg
3 A. J. ter Linden, Proc. Inst. Mech. Engrs., 160
JL cumulative weight of solids descending (1949) 233.
through the dipleg up to time t (ex- 4 Y. Mori, A. Suganuma and T. Shu, J. Chem.
clusive of material shed initially), kg Engrg. Japan, 1 (1968) 82.

You might also like