PARTICULATE REMOVAL 835
almost always result in debits to another. An example is shown
in Figure 4 of three cyclones studied by Van Ebbenhorst
Tengbergen.^24 They are drawn to the same scale and sized
to have the same gas throughput and the same pressure drop.
The relative costs of A : B : C are estimated as 4 : 3 : 9, but the
difference in cost is reflected in the grade efficiency curves.
Pressure Drop Although pressure drop is much easier to
measure than overall or grade efficiency, prediction of pres-
sure drop from theoretical principles is not in good agree-
ment with experiment. Stern^10 has analyzed the pressure
drop predictions of several authors 26 – 28 and has concluded
that a constant value of X 16 in the equationhXHW
dn
= ii
02 , (3)where
h n pressure drop is expressed as inlet velocity
headsH i inlet height
W i inlet width
d 0 outlet pipe dia.gives as good agreement with experimental data as does any
equation for X which includes geometric factors. Measures
values of X do indeed cover a wide range of values, and use
of a constant value of X is merely a best guess in absence of
data on the specific cyclone. Fortunately, because measure-
ment is straightforward and simple, most vendors supply reli-
able information on pressure drop for their own cyclones.
For most cyclone applications the only pressure drop of
interest is that from gas inlet to gas outlet. In some appli-
cations the pressure drop from gas inlet to the dust outlet
at the bottom of the cone is also important. Such is usually
the case for cyclones above fluid beds where collected dust
is returned directly to the bed through a fluidized standpipe
or dipleg. Too great a pressure drop to the dust outlet will
preclude a satisfactory pressure balance in the dipleg and
prevent proper discharge of particles. As a first approxima-
tion pressure drop from inlet to dust outlet can be taken as
13.5 inlet velocity heads.
Cyclones in Series Cyclones are often installed in series
either with other cyclones or as pre-cleaners ahead of more
efficient dust collecting equipment. As pre-cleaners they
are especially attractive because efficiency increases with
dust loading and because they seem able to handle almost
any dust loading that a gas stream can carry to the inlet.
Some cyclones in fluid catalytic crackers operate at inlet
loadings of over 1lb dust per actual cubic foot of gas. Two
stages of internal cyclones in a cat cracker often operate at
an overall collection efficiency of 99.997% on catalyst with
a particle density of 75–80 lbs/ft^3 and a mass median dia.
of 60 m. A third stage of cyclone in this application may
be expected to operate at about 85% efficiency, however,
because that stage sees much finer particles at much lower
loading. Cat cracking cyclones typically operate at pres-
sure drops of about 1 psi/stage, higher than is tolerable in
many applications.
Optimum Cyclone Size and Parallel Operation The
choice of one or a few large cyclones as opposed to many
smaller cyclones operated in parallel represents a balance
between the lower cost of a few large cyclones combined
with easy manifolding against the inherently higher effi-
ciency of the smaller cyclones. Each application is different,
but the considerations that apply in determining size of cat
cracker cyclones may prove instructive:1) Cracking catalyst is highly erosive necessitating
periodic repair of internal metal surfaces or refrac-
tory linings. The cyclones must permit access for
such repairs, which sets a minimum barrel dia. of
about 3½ ft.
2) Cyclone length is proportional to diameter, and
headroom plus dipleg pressure balance often lim-
its diameter (a 5 ft dia. cyclone is typically 17 ft
long). Reactor openings also limit diameter as
does reduced efficiency.(^246810)
0
20
40
60
80
100
PARTICLE DIAMETER, MICRONS
COLLECTION EFFICIENCY, %
C
A
B
ABC
Relative Cost
A : B : C = 4 : 3 : 9
FIGURE 4 Relative sizes, costs, and grade efficien-
cies of different cyclone designs having same through-
put and pressure drop.^21
C016_002_r03.indd 835C016_002_r03.indd 835 11/18/2005 1:06:46 PM11/18/2005 1:06:46 P