Sei sulla pagina 1di 23

Atmospheric Environment Pergamon Press 1973. Vol. 7, pp. 527-549. Printed in Great Britain.

CYCLONE PERFORMANCE AND DESIGN


DAVID LEITH* a n d DILIP MEHTA~
Ambient Purification Technology, Inc., P.O. Box 71, Riverside, California 92507, U.S.A.

(First received 1 June 1972 and in final form 14 November 1972)

Al~traet--Cyclones are probably the most widely used industrial dust collecting devices.
Although many procedures for calculating cyclone pressure drop and collection efficiency
have been developed, current design practice emphasizes past experience rather than an
analytical design procedure. In this paper, five pressure drop theories and four efficiency
theories are evaluated against experimental data taken from the literature. One pressure drop
and one efficiency theory are judged "best" and used with an optimization procedure to
develop a cyclone design method. This design method gives a means of calculating the dimen-
sions of a theoretically optimum cyclone for any set of design criteria. A cyclone designed by
this procedure is compared with a standard high efficiency cyclone, and is found theoretically
to have better efficiency under identical operating conditions.

NOMENCLATURE
a gas inlet height
,4 inside surface area exposed to spinning gas
Ac particle collecting surface of cyclone
b gas inlet width
B dust outlet dia.
C cyclone geometry coefficient
CP modified cyclone geometry coefficient
dp particle dia.
d~ size of particle collected with 50 per cent efficiency
D cyclone dia.
Dc cyclone dia. cm.
De cyclone gas exit duct dia.
f see equation (7)
g acceleration of gravity
G friction factor, 0.005
h cylinder height
H overall height
K constant, see equation (3)
l distance vortex extends below gas exit duct
n vortex exponent, see equation (6)
Ne equivalent number of turns gas makes in cyclone
P pressure drop coefficient, see equation (31)
Qg gas throughput
r radial distance to vortex centerline from point in vortex
S gas outlet duct length
SP modified cyclone inertia parameter
T absolute temperature
ut see equation (14)
v¢ gas velocity through cyclone inlet
v~ see equation (14)
vt tangential gas velocity
w particle migration velocity to cyclone wall
wj, particle terminal settling velocity
wt* terminal settling velocity of critical particle
* Present address: Harvard School of Public Health, 55 Shattuck Street, Boston, Massachusetts
02115, U.S.A.
t Present address: Chemspec, 32 Morvi House, 28/30 Goa Street, Bombay-l, India.
527
528 DAVID LEITH and DILIP MEHTA

Y constant, see equation (4)


AH pressuredrop in inlet velocity heads
Ap pressuredrop in static pressure head
~e entrance loss factor, see equation (12)
q exit loss factor, see equation (13)
•/ fractional collection efficiency
,~ friction factor, 0.02
vc gas viscosity
pG gas density
pL liquiddensity
pe particle density
particle relaxation time
see equation (9)
~v cyclone inertia parameter.

1. INTRODUCTION
CYCLONES are probably the most widely used dust collecting devices. They are
relatively simple and inexpensive to fabricate, operate with moderate pressure losses,
and if properly applied, give long troublefree service. Efficiency is generally good for
dusts where particles are larger than about 5 ~m in dia. Cyclones are frequently
used as final collectors where large particles are to be caught. They can also be used
as precleaners for a more efficient collector such as an electrostatic precipitator,
scrubber or fabric filter.
To design a new cyclone system, or upgrade an existing one, it is necessary to
predict cyclone performance accurately. Over the years, a number of methods have
been developed to predict cyclone pressure drop and efficiency. All make assumptions
regarding the characteristics of gas and particle motion within the cyclone. The
accuracy of the performance equations depends upon how well the assumptions made
in their development reflect the actual operating conditions within the cyclone.
In this paper, several frequently used methods for predicting cyclone pressure drop
and efficiency will be reviewed. One pressure drop and one efficiency equation will
be selected for further analysis. An optimization of cyclone geometry will be made,
and the results will be presented to enable the selection of a theoretically best cyclone
design for a dust collection problem where certain design criteria have been set.
Although many different types of cyclones are used, the present discussion is
limited to cyclones of the cone-under-cylinder layout with tangential gas inlet as
shown in FIG. 1. This is the type most frequently used for industrial gas cleaning.
In this design, particle-laden gas enters the cyclone tangentially through the gas inlet
duct and forms a vortex within the cyclone body. In the region below the gas exit
duct, but above the bottom of the vortex, the spinning gas stream moves radially
inward into a "central core" of low pressure which extends below the gas exit duct.
Upon reaching the central core, the cleaned gas moves up and out of the cyclone
through the gas exit duct. Collected dust moves along the cyclone walls to the dust
exit due to the downward movement of gas at the cyclone walls, aided somewhat by
gravity if the cyclone is installed vertically. Tangential gas velocity, vt, has been found
by SHEPHERDand LAPPLE (1939); FIRST (1950); ALEXANDER(1949) and others, to be
related to radial distance r from the vortex centerline raised to the power n by:

vt r" = constant. (1)


Cyclone Performance and Design 529

I-- U --I

FIG. 1. Cyclone, with dimensions.

Here, n is called the vortex exponent. This relationship does not hold within the
cyclone's central core. ALEXA~qDEg (1949) has stated that the distance the vortex
extends into the cyclone depends only on gas inlet and outlet dimensions. Theoretically
then, the vortex need not reach the bottom of the cyclone.
A cyclone's geometry is described if each of the eight dimensions shown in FIG. 1
is known. However, it is often more convenient to state dimensions by expressing
them in dimensionless form, as a multiple of dia., D. The dimensions of the cyclone
are then specified by dia., D and 7 dimension ratios, a/D, b/D, De~D,S/D, h/H, H/D
and B/D. This permits a comparison of the geometric similarity of several cyclones
through a consideration of their dimension ratios, without introducing absolute
magnitude to the cyclone comparison.

TABLE 1. CYCLONE STANDARD DESIGNS

Source Recommended duty D a/D b/D De/D S/D h/D HID BID

STAIRMAND (1951) High emciency 1 0.5 0.2 0.5 0.5 1.5 4.0 0.375
SWIFT (1969) High efficiency I 0.44 0.21 0.4 0.5 1.4 3.9 0.4
LAPPLE (1951) General purpose I 0.5 0.25 0.5 0.625 2.0 4.0 0.25
SWIFT (1969) General purpose 1 0.5 0.25 0.5 0.6 1.75 3.75 0.4
STAIRMAND (1951) High throughput* 1 0.75 0.375 0.75 0.875 1.5 4.0 0.375
SWIFT (1969) High throughput 1 0.8 0.35 0.75 0.85 1.7 3.7 0.4

* Scroll type gas entry used.


A.E. 7 / 5 - - D
530 DAVID LEITH and DILIP MEHTA

A number of cyclone "standard designs" or sets of dimension ratios have been


suggested in the literature. Several are listed in TABLE I. Comparing the designs
shows that the dimension ratios differ according to the purpose for which the cyclone
is to be used. High efficiency cyclones tend to have smaller inlet area (aid and b/D)
and exit area (De~D) than do the high throughput designs. Gas exit duct length (S/D)
is less in the high efficiency designs, probably because inlet height (a/D) is less. Exit
duct length is always greater than inlet height in order to avoid the gas short circuiting
the cyclone by passing directly from the inlet to the outlet without forming a vortex.
General purpose standard designs appear to be a compromise between high efficiency
and high throughput.
Higher efficiency can be traded off somewhat against higher throughput according
to STAIRMAND(1951) and SWIFT (1969). A high efficiency cyclone of standard design
will process only one-half or so as much gas as a high throughput design of the same
diameter and overall height, operating at the same pressure drop. Cyclone design
varies, then, according to the purpose for which the unit is to be put. The variability
in design emphasizes that there is no single optimum cyclone. That is, there is no
single cyclone design which will perform best for all dust collection problems.

2. C Y C L O N E PRESSURE DROP
Pressure drop in a cyclone will be considered as that between points 1 and 2 in
FIG. 1. A number of factors contribute to the pressure difference between these
points, and have been identified by SHEPHERDand LAPPLE(1939) as:
(1) Loss due to expansion of the gas when it enters the cyclone chamber.
(2) Loss as kinetic energy of rotation in the cyclone chamber.
(3) Losses due to wall friction in the cyclone chamber.
(4) Any additional frictional losses in the exit duct, resulting from the swirling
flow above and beyond those incurred by straight flow.
(5) Any regain of the rotational kinetic energy as pressure energy.
Most pressure drop theories consider factors two and three to be most important
in determining cyclone pressure drop. While it is not the purpose of this paper to
discuss the derivations of various pressure drop theories, the final working equations
reported by SHEPHERD and LAPPLE (1940); FIRST (1950); ALEXANDER (1949);
STArer,lAND (1949) and BARTrI (1956) will be presented. All express cyclone pressure
loss as a number of inlet velocity heads, AH. Inlet velocity heads can be converted to
pressure drop in terms of static pressure head, AP, by:

AP - - I)02 PG AH
2g PL (2)

SHEPHERDand LAPPLE (1939, 1940) Pressure drop equation


Shepherd and Lapple correlated experimental pressure drop results with the
following equation:
ab
An = K-- (3)
De z
Cyclone Performance and Design 531

where K = 16 for a cyclone with standard tangential inlet, and K = 7.5 for a cyclone
with an inlet vane, i.e. where the inner wall of the tangential entry extends past the
cyclone inner wall to a point halfway to the opposite wall.

Fmsr (1950) Pressure drop equation


First presented the following expression for cyclone pressure loss, which while
similar to the Shepherd and Lapple equation, takes into consideration the additional
factors of cyclone diameter, D, plus cone and cylinder height, ( H - h) and h,
respectively.
ab { 12[_Y
an = ~ [h(n- h)lD2]'/3J" (4)
Here, Y, a constant, is 0.5 for a cyclone with no inlet vane, 1.0 for a neutral inlet
vane, and 2.0 for an inlet vane that expands the entering gas stream and touches the
gas outlet duct.

ALEXANDER(1949) Pressure drop equation


Alexander derived the following expression for pressure drop, assuming all pressure
losses are due to kinetic energy losses in the gas stream. His equation is"

46 ab 1} ( D~ 2hI

where
n=l--(1 (0"394D~)°'l')(--~-T~ °'3
2.5 \283/ (6)
and

0.8 _In(1 1 1 -- n

(7)
Here, Dc is the dia. of the cyclone in cm, and T is the absolute temperature in degrees
K.

STAIRMAND(1949) Pressure drop equation


Stairmand has presented a method of calculating the pressure drop through a
cyclone by:

6 H = 1 + 2¢ 2 \ De 1 q- 2 \-~-ee2] (8)

where
De ~ 112 ( De 4GA] x12
-- 2(/)---- b)/ q- "2(1)~ b) q "~]
4' = 2GA (9)
ab
532 DAVIDLEITHand DILIP MEHTA

Here, G is a friction factor approximately equal to 0.005 and A is the inside surface
of the cyclone exposed to the spinning gas. A value for A can be calculated from:

A = -~
~r (D 2 _ De2) %- rrDh %- zrDeS %- -~ (D %- B) (H -- h) 2 -+-

(10)
Stairmand assumes that the total energy loss is due to losses in the cyclone vortex,
plus entrance and exit losses.

BARTH (1956) Pressure drop equation


Barth expresses cyclone pressure drop as a function of the sum of two factors,
Ee and ~l. The factor ~e reflects losses due to entrance to the cyclone and internal
friction and kinetic energy losses. The factor ~l results from losses at the gas outlet.

¢ (11)

where

De{[1--(u,/v,)(~-
"e = ---~ S) (2~De)a] 2 -- 1} (12)

4.4
eI = (ujvi)2/3 %- 1 (13)

u, (De/2) (D -- b)~r
-- = (14)
vi 2aba %- (H -- S) (D -- b) ~rh

a g 1 -- 1.2 b/D. (15)

Here, ~ is a friction factor approximately equal to 0.02.

Comparison of pressure drop equations


Determining which pressure drop equations work best would be done by comparing
experimental values of pressure drop with calculated values for each of the pressure
drop equations. For proper comparisons to be made, it is necessary to have all eight
dimensions and an experimental value of AH. SaEPrmRD and LM'PLE (1940); STERN,
CAPLAN and BusrI (1955) and STAmMAND (1949) present cyclone pressure drop data.
In many cases, complete measurements of the cyclone geometry are not available.
For those cases where all dimensions are listed (except dust outlet dia., B, which was
assumed equal to 0.25D) and for cyclones without inlet vanes, the experimental
values of AH were compared with those calculated according to the methods of
Shepherd and Lapple, First, Alexander, Stairmand and Barth given above. The data
and results are shown in TABLE 2.
To evaluate which method for calculating pressure drop is most compatible with
the data, correlation coefficients between the data and computed results were
calculated according to the method of MILLER and FREUND (1965). From these data,
it appears that the Barth, Stairmand, and Shepherd and Lapple methods are best.
Cyclone Performance and Design 533

v~

Z m

<

0
Z

<

e4
n

~ , ~°~

~.1 •
c,,t

• + - m
534 DAVID LEITH and DtLIP MEHTA

TABLE 3. CORRELATION COEFFICIENTS FOR THEORETICAL


CYCLONE PRESSURE DROP EQUATIONS BASED ON CONTENTS
OF TABLE 2

Method Correlation coefficient

BARTH (1956) 0.80


STAIRMAND (1949) 0.77
SHEPHERD and LAPPLE (1940) 0.77
ALEXANDER (1949) 0.55
FIRST (1950) 0.53

The Alexander and First methods do not appear to be as good. The correlation
coefficients calculated are given in TABLE 3.
The Barth and Stairmand methods are complex and require knowledge of all cyclone
dimensions. The Shepherd and Lapple approach is simpler; it does not include all
dimensions. However, for these data, the Shepherd and Lapple method appears to
give results about as good as those produced by the more complicated calculation
methods.

3. C Y C L O N E COLLECTION EFFICIENCY
Cyclone fractional efficiency is the weight of particles of a stated size collected in
the cyclone, divided by the total weight of particles of that size going to the cyclone.
Experience in dealing with cyclones has shown that collection efficiency increases with:
(1) Increasing particle size and density.
(2) Increasing speed of rotation in the cyclone vortex.
(3) Decreasing cyclone diameter.
(4) Increasing cyclone length.
(5) Drawing some of the gas from the cyclone through the dust exit duct.
(6) Wetting the cyclone's walls.
A cyclone's grade efficiency curve relates size of particles going to the cyclone to the
cyclone's efficiency on particles of that size. The grade efficiency curve for Stairmand's
high efficiency cyclone (see TABLE 1 for dimension ratios and TABLE4 for operating
conditions) is shown in FIG. 2. Note that efficiency continuously increases with
increasing particle diameter, and approaches 100 per cent asymptotically for sufficiently
large particles.
Various ways to calculate collection efficiency from theoretical considerations have
been developed, and will be discussed briefly in the following sections.

Critical size approach to cyclone efficiency


The efficiency of a cyclone is sometimes characterized by its particle "critical size"
(the size of particles calculated to be collected with 100 per cent efficiency), or by its
"cut size" (the size of particles calculated to be collected with 50 per cent efficiency).
As discussed above, actual collection efficiency approaches 100 per cent asymptotically
for larger particles. The critical size particle then, while calculable from theory, is not
observed experimentally.
Cyclone Performance and Design 535

I00

4-
§ 8o
STAIRMANO (1951)
6O
:>, ptimized design

u
40
uJ

20

I I
0 5 I0 15

Particle dio., /~m

Fxo. 2. Cyclone grade efficiency curves, high efficiency standard design and optimized design.

TABLE4. DZMENSZONSANDOPERATINGPARAMETERSFOREXPERIMENTALCYCLONES

Dimension ratio STAIaMAND0951) PETE~ONand WmTaY (1965)

a/D 0.500 0.584


biD 0.200 0.208
S[D 0.500 0.584*
De/D 0.500 0.500
hid 1.500 1.333
HID 4.000 3.167
BID 0.375 0.500
Operating conditions
dia., D 0.203 m 0.305 m
particle density, pp 2.0 g c m - 3 1.6 gcm - a
throughput, Qg 0.0627 m a s- 1 0.288 m a s- 1

* Assumed value.

T h e Stokes' law t e r m i n a l settling velocity of critically sized particles, w~,*, can be


calculated according to BARTH (1956):

Qog
ws* = (16)
2~r(H - - S ) v, 2

where vt is evaluated at r = De/2. I f the Stokes' law settling velocity o f a particle o f


dia. dp is given by ws, t h e n the ratio o f ws to ws* will be:

wj, ~'(H - - S ) v, 2 Op d,, 2


(17)
w.r* = 9 Qg I~G
Barth states t h a t theoretically, cyclone efficiency should j u m p f r o m 0 to 100 per cent
as ws/wt* attains a value o f 1. Actually, efficiency increases gradually with increasing
536 DAVID L~rrH and Drtrp MEHTA

I00

o
75

/
/
50
u

I.d
25

n' t / We*
FIo. 3. Cyclone efficiency vs wflwI* , from BARTH (1956).

wflwt* as is shown by Barth in Fro. 3. This figure was developed from experiments
run on a variety of cyclones.
In a similar manner, LAPPLE (1951) showed that actual efficiency increases
gradually with increase in the ratio of dp/dpc, as is shown in FIa. 4. Here, cyclone cut
particle size, dpc, is given by:

r /.Lab i t / 2
dpc---- 3[2~r~-~pv~Ne ] •
(18)

Here, Ne is the effective number of turns the gas makes in the cyclone, a constant
reported by FRIEDLANDER et al. (1952) as varying from 0.5 to 10, with a typical
value of 5.

~
o

oJ

OA

04 06 08 I0 2 4 6 8 ~0

FIG. 4. Cyclone etficiency vs d,/d,,, from LAPPLE (1951).


Cyclone Performance and Design 537

SPROULL(1970) Efficiency equation


SprouU has presented a method for calculating cyclone efficiency by drawing a
parallel between cyclone collection and collection in an electrostatic precipitator:

= 1--exPL
r- Qg
w"°lJ"
Here, the migration velocity of particles to the cyclone wall, w, is:
dr, 2 pn vi 2
w = . - 18/~
- o
D (20)

The interior collecting surface of the cyclone, Ac, can be calculated from:

Ac = 7rDh q- ~ (D q- B) (H -- h) 2 q- . (21)

LEITH and LIcrrr (1972) Efficiency equation


Leith and Licht derived an equation for predicting cyclone efficiency from
theoretical considerations. Their equation is:
~7= 1 -- exp[-- 2(C7t) 1/c2n+2~] (22)
where C is a function of the cyclone's dimension ratios only:

C = --~-

(23)

and where
l De (D2~ ~/3
= 2.3 - ~ \ab] (24)

and
d D - - ( D - - B ) ( ( S q - 1 - - h ) / ( H - - h))
= D (25)

Here, l is the farthest distance the vortex extends below the gas exit duct as given by
ALEXANDER(1949), and dis the dia. of the conical section at that point. 7tis a modified
inertia parameter, reflecting the nature of the gas/particle system to be treated:

7t = pp dp2 vg (n + 1). (26)


18/La D
The value of vortex exponent, n, can be calculated from equation (6).

Comparison of efficiency equations


LEITI-I and Lxcrrr (1972) have compared the four methods for calculating cyclone
efficiency described here with experimental data given in the literature by STAIRMArrD
538 DAVID LI~ITHand DILIP MEHTA

, 80 o experimental (1951)
,.~ A;/ / .... BARTH i'heory (1956)
oo /!// - - - L . . . . . -,
g
_° / 17/I -----SPRouLLI"heory (,970)

"
40 r /I,/,,,, / j
~ ---- LEITH and LICHT -I-heory (1972)
. . . . . . . . . .

oI//
/
/ f i
0 5 I0 15
Por'hcle dlo., ~m
Fro. 5. Cyclone efficiency vs particle dia., STAIRMAND (1951). Experimental results and
theoretical predictions.

I00

/~ /
~ / expertmen'fol (1965)
/,,~/ / .............
60 ~ .... BARTH theory (1956)
o~- / / ~ ~ - - - - LAPPLEtheory (,951)
- 40 I//iI / / SPROULLtheory (1970)
/J / / ',t and LICHT theory
LEITH ond fheor (1972)1
20 !
/

0 5 I0 15
Per-lqcle die, ,u.m

FiG. 6. Cyclone efficiency vs particle dia., PETERSON and WHITBY (1965). Experimental
results and theoretical predictions.

(1951) and by PETERSONand WHITBY (1965). Operating parameters for these experi-
mental cyclones are given in TABLE4. FIGURES5 and 6 compare computed efficiency
values with experimental values, and s h o w that at least for these data, the Leith and
Licht relationship appears to work best. A comprehensive comparison of the efficiency
equations with experimental data has not been possible due to a lack of sufficiently
detailed data in the literature.

4. O P T I M I Z I N G CYCLONE PERFORMANCE
With the exception of the Stairmand pressure drop equation, which applies to all
constructions, the pressure drop and efficiency equations discussed here have been
CyclonePerformanceand Design 539

developed with the typical cyclone of FIG. 1 in mind. Their application to other
designs such as cyclones with scroll type gas entries, cyclones with swirl vane entries,
or "straight through" cyclones in which the gas does not reverse direction, is unknown.
Several factors known from experience to have an influence on cyclone pressure
drop and efficiency are not reflected in the pressure drop and efficiency equations.
No equation accounts for the apparent increase in efficiency as observed by BAXTER
(1968), VAN EBBENHORSTTEI~GBERGEN(1965), and others, and decrease in pressure
drop noted by BRIGGS (1946) and SPROOLL(1966) associated with increasing dust
loading to the unit. The SI~EPHERDand LAPI'LE(1939, 1940) and ALEXANDER(1949)
pressure drop equations do not include vertical dimensions (S/D, h/D or H/D) as
having an influence on pressure drop, despite the fact that increasing length should
increase gas stream friction losses at the cyclone walls. All efficiency equations
assume that once a particle reaches the wall of the cyclone, it is collected. This is in
contrast to the observations of MORII (1968) that large hard particles bounce off
the cyclone walls back into the gas stream. It is also probable that some collected
particles, primarily in the smaller size ranges, will be reentrained from the collector
wall before being discharged through the dust exit duct. Smaller particles may be more
susceptible to reentrainment as they are only collected with difficulty in the first place.
It has been noted by STAIRMAND(1951) that cyclone efficiency increases if the cyclone
walls are wetted, as wetting presumably hinders particle bounce and reentrainment.
No efficiency equation accounts for the increase in collection efficiency noted by
STAmMAND (1951) and CAI'LAN(1968) associated with drawing off a fraction of the
gas throughput from the dust exit. The increase in efficiency associated with the
base draw-off has been attributed to a reduction in dust reentrainment near the dust
exit. Another effect of the draw off may be to induce the vortex to run the full length
of the cyclone, rather than ending some distance above the bottom of the cyclone as
has been reported by ALEXANDER(1949). A disadvantage of this practice is that the
gas drawn off, typically 5-15 per cent of the gas throughput, has to be recycled to the
cleaning equipment. The recycle results in a somewhat larger cleaning system being
required, as well as the installation of otherwise unnecessary auxiliary fans.
Although the pressure drop and efficiency equations do consider many factors
influencing cyclone performance, there are other factors still to be accounted for.
The results calculated through the use of these equations remain open to question,
although fundamental trends can be pointed out. With these limitations in mind,
it is possible to explore improving cyclone performance through a theoretical optimiza-
tion of cyclone geometry. To do this it is necessary to select one pressure drop equation
and one efficiency equation for further analysis. The Shepherd and Lapple pressure
drop equation, equation (3) was selected as it is comparatively simple, yet yields
results about as accurate as the other more complex equations as discussed above.
The Leith and Licht efficiency equation, equation (22) was picked as it appears to
be the most accurate, and may best reflect actual conditions of particle and gas
motion within the cyclone.
The Shepherd and Lapple pressure drop equation does not consider any vertical
dimensions as contributing to pressure drop. From this equation, then, a tall cyclone
should have the same pressure drop as a short one, so long as inlet and outlet dimen-
sions are the same. On the other hand, the Leith and Licht efficiency equation predicts
that the efficiency of a cyclone can be increased ad infinitum if vertical dimensions
540 DAVID LEITH and DILIP MEHTA

H/D, hid and S/D are increased equally. To avoid obtaining unuseful results (make
cyclone as long as possible since this increases efficiency at no cost in pressure drop),
values of HID and hid must be fixed before cyclone geometry can be optimized.
This restriction does not compromise the fact that theory shows longer cyclones
should perform better. It does serve to anchor cyclone length so that the influence
of other geometric variables on performance can be examined.
It is helpful to understand the change in efficiency and pressure drop associated
with altering any one dimension of a cyclone. Stairmand's high efficiency, 8 in. dia.
cyclone was selected; its dimension ratios are listed in TABLE 1. The operating para-
meters for this cyclone are the same as given in TABLE4. The efficiency and pressure
drop this cyclone should exhibit when any selected dimension is decreased to 50 per
cent or increased to 150 per cent of its original value is shown in FIO. 7 for efficiency
and FIG. 8 for pressure drop. All dimensions other than the one being varied were
maintained at their normal (100 per cent) values. Although all dimensions enter the
efficiency equation, inlet height and width, a and b, exit duct dia., De and cyclone
dia., D, are the most important. FIGURE 7 shows the effect of changing a, b, De and
D only. Variations in other dimensions did not add or subtract more than 5 to the
87 per cent efficiency obtained with all dimensions at their original 100 per cent values.

iO0

o 90

80
c~

70
g
g
8
60

50
50 I00 150

Orlg,nal length, percent

FIG. 7. Effect of altering individual cyclone dimensions on theoretical collection efficiency.

The significant dimensions from the pressure drop standpoint are gas inlet height
and width, a and b, and exit duct dia., De, as these are the only dimensions in the
selected pressure drop equation.
As stated in equation (22) Leith and Licht show efficiency to be a function of the
product of C and ~g. Here C, equation (23) is a dimensionless cyclone geometry
coefficient, a function of cyclone dimension ratios only. kv equation (26) is a dimen-
Cyclone Performance and Design 541

40

E 20

a.

~ a

~ 6
13

I I I I I I I t I
50 I00 50
Origmol leng'lh, per cenl"

Fro. 8, Effectof altering individual cyclonedimensions on theoretical pressure drop.

sionless inertia parameter, reflecting the properties of the gas and particles. Efficiency
will increase as C, ~u, or both increase.
The factor W contains the term vg, cyclone inlet velocity. Inlet velocity is related to
gas throughput and cyclone inlet height and width by:

Qg
v~-~ab. (27)

Before optimizing cyclone geometry, cyclone inlet height and width must be removed
from ~ and moved to C to put all geometry terms together. This can be done by
multiplying ~u by the dimensionless quantity ab/D2 and C by its reciprocal, D2/ab.
Now for any fixed value of ab~/D 2, representing particles of any properties and
gas of any viscosity passing through a cyclone of any diameter with any throughput,
efficiency will be greatest when the cyclone's geometry factor of CD2/ab is greatest.
The quantity CD2/ab contains explicitly or implicitly all seven cyclone dimension
ratios. When varying dimension ratios to maximize this quantity, constraints must be
placed on some of the ratios. This is done to make sure the cyclones described are
physically feasible and within the region where the efficiency equation applies. The
two constraints are:

a S

S h
~ -~. (29)
542 DAWDLEIThand DILIPMEHTA

Equation (28) is necessary to keep the gas just entering the cyclone from short
circuiting to the gas exit duct without forming a vortex. Equation (23) for geometry
coefficient, C, was developed for only those cyclones where equation (29) is obeyed.
Combining equations (2, 3 and 27) yields:

ab = P (30)

where
gPL D 4 AP
P = 8p~ Qo2 " (31)

Equation (31) defines a cyclone pressure drop coefficient, P. Equation (30) shows how
the terms in P, pressure drop, dia. and throughput, interact with inlet and exit dimen-
sions. To isolate the influence of dimensions from the influence of other factors, it is
necessary to keep P constant. The relationship between inlet and exit dimensions is
then fixed by equation (30).
The length of the vortex in the cyclone, S - k / , need not necessarily reach the
bottom of the cyclone. If it does not, the space between the bottom of the vortex
and the bottom of the cyclone will not be utilized for particle collection. If, on the
other hand, the vortex theoretically extends through the bottom of the cyclone,
excessive turbulence and dust pickup from the dust exit may result. Total cyclone
space utilization calls for the vortex just to reach the bottom of the cyclone:
H S l
= ~ q- ~ . (32)

Solving equation (32) for the dimension ratio S/D with the aid of equations (24 and 30)
yields:

=-~--2.3 P (33)

Equation (23) for geometry coefficient, C, shows that C and hence CD2/ab will
increase as both inlet height, a i d and the product of inlet height and width, ab/D 2,
decrease. Inlet height, a/D, appears alone in equation (23) while width, b/D, does not.
For a fixed product ab/D 2, then, C will increase by making a i d smaller and hence
b/D larger. To prevent a sudden contraction in the gas stream entering the cyclone
which might interfere with vortex formation or increase pressure drop, inlet width
b i d should be less than the space between the cyclone wall and the gas exit duct.
Making b/D as large as possible without overstepping this limit leads to:

~) = ½ -- (34)

With biD fixed, a/D can be found by combining equations (30 and 34):

a 2
-D = P(De/D) z [1 -- (De~D)]" (35)
Cyclone Performance and Design 543

The substitution of equations (30-35) into equation (23) yields an expression for
CD2/ab which is a function of P, De~D,H/D, h/D and B/D:

_ 1 B

-4-(1- (~_e)2)(2.3 p1/3 (~_e)S/3)A- h H]. (36)

Examination of equation (36) shows that CD2/aband hence collection etficiency can
be made as large as desired by increasing HID and h/D, as pointed out above. Some-
times cyclone height will be limited by available space for installation. If space is not
a problem, when applying the present means of analysis, it is prudent to limit H/D
and hiD to values where the pressure drop and efficiency equations have been shown
to be valid by comparison with experimental data. Reasonable limits might be:
H
= 5 (37)

h
= 3. (38)

These values will be used here. Now let:


B
= 0.375 (39)

as a typical value. Alternatively, BID might be fixed as a fraction of De/D.


Substitution of equations (37-39) into equation (36) yields an expression for
CD2/ab which depends on De/D and P only. Values of CD2/abwere maximized on a
digital computer for a range of values of P, by a trial and error procedure for De/D.
With P fixed, the value of De/D leading to the greatest value of CD2/abwas found.
The other variable dimension ratios, S/D, b/D and aid were calculated as a function
of De/D from equations (33-35) respectively. The consequent value of "optimized C"
was then determined from equation (23).
It was found for P greater than about 140, the constraint given in equation (28)
was violated. Similarly, for P smaller than about 5, the constraint of equation (29)
was exceeded. The optimization results are therefore limited to P values between
5 and 140.
The results of this optimization are shown in FIGS. 9 and 10. FIGtn~ 9 is a plot of
optimized C against P. FIGURE 10 shows optimized C against its corresponding
dimension ratios, De~D,S/D, b/D and aiD. For any selected values of pressure drop,
Ap, dia., D, gas density, oo and throughput, Qg, a value of P can be calculated from
equation (31). FIGURE 9, the plot of optimized C against P, will then provide the
maximum value of C obtainable for this value of P. The dimension ratios for the
cyclone can be found from FIG. 10, the plot of dimension ratios against optimized C.
The efficiency of this cyclone on dust and gas of any properties can be found from
544

4oo!
200
DAVID LEITHand Dn.ir MEHTA

1130

~o

o
2o

,oi i I I I I I II i 1 i J I I I I I
5 ~0 20 50 tO0 200

Cyclone pressure drop coeffiment, "P"

FIG. 9. Cyclone pressure drop coefficient, P, vs geometry coefficient, C, for optimized


cyclones.

! i ]",, F T't-,LI 8,,D-- o37~1


I ; o/o I s/o , ]
I
i t iS , '
o O8
--~ -c-i
O6
I i oe/o I IN I / k ~ i~
f 0,4

E
c3
It I~
02
, I I'/

0 I
oo8

0 06
tO 20 50 I O0 200 500

Cyclone geometry coefhclenf, "C"

Fro. 10. Cyclone geometry coefficient, C, vs dimension ratios for optimized cyclones.

equation (22). Gas inlet velocity, vg, is found as a function o f throughput, Qg, and
inlet height and width, a and b, through equation (27). FIGURES 9 and 10 are for
values of HID = 5, h/D = 3 and BID = 0.375. Had different values for these dimen-
sion ratios been chosen, different plots would have resulted.
Cyclone Performance and Design 545

It is possible to rearrange the cyclone efficiency equation (22) into:


~ / = 1 -- exp[-- 2 CP SP] (40)
where

[Pe dp2] 1/(2n+2) ~--- [,r]l/(2n+2) (41)


SP =- L1 - ' ~ J
and
Qg I/(2n + 2)

CP -- [
LCDu
a-'# D a (n + 1) (42)

Here the effect of particle and gas properties have been combined into SP, a power of
particle relaxation time, ~-. Relaxation time is that time necessary for a particle to
reach 63 per cent of its terminal settling velocity in a quiescent gas. Note that SP
is not totally independent of dia., D, as the value of n in the exponent is related to D
through equation (6). For most cyclones, n can be taken as about 0.7 without too
much error.
The quantity CP contains the terms relating to optimized cyclone geometry,
CDZ/ab, dia., D and throughput, Qg. Cyclones with the same value of CP will have
the same efficiency on identical particles (particles with the same SP). FIGURE 11
shows different combinations of throughput, Qg, and dia., D, which can be used to
accomplish a given value of CP, for a pressure drop of 7.5 cm of water column.

~ - ~ . 4 m 3s-I l
] 3.2 AP --7.5cm

E I

. 0.8

"o
/ IO/ ' 4 X\ !)" \
0.6
o

U 0.4
• V 0.1

02

0.!
20 4.0

Cp I S- I / ( 2 n + 2 )

FIG. 11. Cyclone dia., D, vs CP with gas throughput and inlet velocity as pammetcxs, and
pressure drop of 7.5 cm water column.
A.~.7/5----n
546 DAVIDLEITHand DILtp MEHTA

FIGURES 12 and 13 are similar figures, but where pressure drop is 12.5 and 17.5 cm
of water column, respectively. A gas density of 0.012 g cm -3 was used in these three
figures. At any point on FIGS. 11-13, sufficient information is pinpointed to enable
a value for pressure drop coefficient P to be calculated. With P known, the geometry
of the cyclone can be determined as outlined above.
FIGURES 11-13 show the interplay of gas velocity, cyclone diameter, and pressure
drop with collection capability. As an example, consider the design of a cyclone to
have a CP value of 20 s -~/(2"+2). This corresponds to an efficiency of 67 per cent

5 [ ~ I ! I :

08
E
06
o

o O4

O2

O.t

0 20 40

CP, S -I/(2nt 2)

Fio. 12. Cyclone dia., D, vs CP with gas throughput and inlet velocity as parameters, and
pressure drop of 12.5 cm water column.

on a 3/zm dia. particle of density 2 g c m - 3 , using equation (40). For a gas throughput
of 0 . 4 m 3 s -1, TABLE 5 lists three sets of design data, taken from FIGS. 11-13,
representative of cyclones operating at 7.5, 12.5 or 17.5 cm of water column pressure
drop, respectively. A comparison of the three sets of data reveals that a higher pressure
drop cyclone will have a smaller dia. The designer can now exercise his economic
judgement. With the cost of fans and power as well as material costs at hand, he can
determine which of the three cyclones is most economical to build and operate for
his application.
Reentrainment of collected dust from the cyclone wall may become a problem if
the velocity of the gas at the wall depends on the velocity of the gas in the inlet to
the cyclone. Reentrainment may not be a problem if the inlet velocity is below
3-4 m s -1. As shown in FIGS. 11-13, cyclones with smaller dia. can achieve higher
values of CP and therefore higher collection efficiency, than can larger dia. cyclones
with the same inlet velocity. Where collection efficiency requirements are high, it may
Cyclone Performance and Design 547

3 I
.6.4 rn3s"1

/
A3.a =17.5cm

t L,~ V \
r I A !\Xo.8
o.a tl X \ , ~
'/'\ ) 4
.
._o 0.2
0.4
{ :

0.08
0 20 40
CP, $-t/(2n+2)

Fro. 13. Cyclone dia., D, vs CP with gas throughput and inlet velocity as parameters, and
pressure drop of 17.5 cm water column.

TABLE 5. CYCLONEDESIGNSFOR EQUALEFFICIENCY

Pressure drop dia. Inlet velocity Geometry coefficient


(cm of water) (m) (m s- ~) (dimensionless)

7.5 0.63 3.1 205


12.5 0.49 3.3 135
17.5 0.41 3.5 105
For all designs:
CP = 20 s -1/(2"+2)
Qg = 0.4 m a s -1

be better to use a number of smaller dia. cyclones in series to process a given gas
throughput, rather than risk reentrainment from a single, larger cyclone.

5. C O M P A R I S O N OF STANDARD DESIGN AND OPTIMIZED DESIGN


The theoretical efficiency of a cyclone dimensioned through the use of the cyclone
o p t i m i z a t i o n p r o c e d u r e can be c o m p a r e d with existing, o r state o f the a r t design
p r o c e d u r e s . This c a n be d o n e b y c o m p a r i n g the g r a d e efficiency curve o f a well-
established " h i g h efficiency" cyclone o f s t a n d a r d design, with the curve f o r a n opti-
m i z e d cyclone o f t h e s a m e diameter, o p e r a t i n g at the s a m e pressure d r o p , with t h e
s a m e gas t h r o u g h p u t on d u s t o f the s a m e density. A s an e x a m p l e o f a s t a n d a r d high
548 DAVID LEITH and DILIP MErrrA

efficiency cyclone, Stairmand's high efficiency design was selected. Its grade efficiency
curve was given in FIG. 2. Operating parameters at which this curve was developed
were given in TABLE4.
For identical operating conditions, equation (31) shows that the pressure drop
coefficient, P, of an optimized cyclone will be 31.5. FIGURE9 shows that for this value
of P, a cyclone with geometry coefficient of 100 should be used. Dimension ratios
for a cyclone with this C value can be taken from FIG. 10. The value of C calculated
using the dimension ratios of Stairmand's high efficiency cyclone is 55. The optimized
cyclone has a C value 82 per cent greater than that of the standard design.
The grade efficiency curve for the optimized cyclone can now be calculated from
equation (22) after determination of IF values for a number of particle diameters.
FIGURE 2 shows the theoretical grade efficiency curve of the optimized cyclone, as
well as the experimentally determined curve of the Stairmand cyclone. There is an
improvement in efficiency through use of the optimized cyclone. Note, however,
that the overall height of the optimized cyclone is 5 D, while that of Stairmand is 4 D.
An alternative route to the grade efficiency curve for the optimized cyclone can
be taken through use of FIG. 11. Here, for dia. and throughput equivalent to
Stairmand's values, a CP value of 27 S - 1 / ( 2 n + 2 ) c a n be identified. Values of S P
calculated from equation (41) can be used with this CP to obtain the same optimized
cyclone grade efficiency curve, using equation (40).
The theoretical pressure drop calculated for the Stairmand cyclone using the
Shepherd and Lapple pressure drop equation is 9.71 cm of water column. This is
considerably above the experimental value of 7.60 cm, which was used above to
calculate P and determine the design of the optimized cyclone to be compared with
Stairmand's. The Stairmand cyclone and its optimized counterpart might also be
compared at the same theoretical pressure drop of 9.71 cm of water. Had the com-
parison been made at the same theoretical pressure drop, the collection efficiencies
calculated for the optimized cyclone would have been better than those calculated
above. The comparison at 7.60 cm pressure drop, illustrated in FIG. 2, therefore gives
a conservative estimate of collection efficiency for an optimized cyclone design.

6. CONCLUSION
Several means for calculating cyclone pressure drop and collection efficiency are
available. Based on a comparison between experimental data drawn from the literature
and theoretical calculations, the pressure drop calculation methods proposed by
BARTH (1956); STAIRMANO(1949) and by SHEPHERDand LAPPLE(1939, 1940) appear
superior to the methods proposed by ALEXANDER(1949) and by FIRST (1950). The
Shepherd and Lapple approach is the simplest to use of all the methods considered
here, and yields results about as accurate as any. Based on a limited comparison
between literature data and theoretical calculations, the LEITH and LICHT (1972)
method for predicting cyclone collection efficiency appears to be more accurate than
the methods of BARTH(1956) ; LAPPLE( 1951) or SPROULL(1970).
In the past, cyclones have been designed on the basis of past experience rather than
through use of an analytical design procedure. No pressure drop or collection
efficiency theory takes into account all the factors known to influence cyclone
performance. However, bearing these limitations in mind, the Shepherd and Lapple,
Cyclone Performance and Design 549

and Leith and Licht theories were used with an optimization method to yield an
analytical method for designing cyclones to meet any set of specified design criteria.
For one set of criteria, any of several optimized cyclones may be used. The designer
must choose between the alternatives by deciding which is most economically feasible
for his application. The efficacy of the present design method is checked by comparing
a cyclone designed by this method with a standard cyclone design. The unit designed
by the present method has better efficiency characteristics, while operating under the
same conditions.

REFERENCES
ALEXANDERR. McK. (1949) Fundamentals of cyclone design and operation. Proc. Australas Inst.
Min. Met. (New Series) 152-3, 203-228.
BARTH W. (1956) Design and layout of the cyclone separator on the basis of new investigations.
Brenn. Warme Kraft 8, 1-9.
BAXTERW. A. (1968) Source Control by Centrifugal Force and Gravity. K. J. CAPLAN.In Air Pollution,
Vol. III (Edited by STF.RNA. C.). Academic Press, New York.
BRIOOS L. W. (1946) Effect of dust concentration on cyclone performance. Trans. A.LCh.E. 42, 511.
CAPLAN K. J. (1968) Source control by centrifugal force and gravity. In Air Pollution, Vol. III
(Edited by STERN A. C.). Academic Press, New York.
Fms'r M. W. (1950) Fundamental factors in the design of cyclone dust collectors. Doctoral thesis,
Harvard University.
FRIEDLANDERS. K., SILVERMANL., DRINKERP. and FIRST M. W. (1952) Handbook on Air Cleaning,
U.S.A.E.C. Washington (AECD-3361 ; NYO-1572).
LAPPLE C. E. (1951) Processes use many collector types. Chem. Engng. 58, 144-151.
LEITH D. and LICHT W. (1972) Collection efficiency of cyclone type particle collectors, a new
theoretical approach. A.L Ch.E. Symposium Series: Air-1971.
MILLER I. and FREUND J. E. (1965) Probability and Statistics for Engineers. Prentice-HaU, New
Jersey.
MORII Y., SUGANUMAA. and TANAKAS. (1968) On collection efficiency of gas cyclone on coarse
particle range. J. Chem. Eng. Jap. 1, 82-86. (In English.)
PETERSONC. M. and WHrrav K. T. (1965) Fractional efficiency characteristics of unit type collectors.
A S t t R A E J. 7(5), 42-49.
SHEeHERD C. B. and LAPPLE C. E. (1939) Flow pattern and pressure drop in cyclone dust collectors.
Ind. Eng. Chem. 31, 972-984.
SHEeHERD C. B. and LAPPLE C. E. (1940) Flow pattern and pressure drop in cyclone dust collectors.
Ind. Eng. Chem. 32, 1246-1248.
SPROULL W. T. (1966) Effect of dust concentration upon the gas-flow capacity of a cyclonic collector.
A.P.C.A.J. 16, 439--441.
SPROULL W. T. (1970) Air Pollution and Its Control. Exposition Press, New York.
STAIRMANDC. J. (1949) Pressure drop in cyclone separators. Engineering, 16B, 409-411.
STAmMAND C. J. (1951) The design and performance of cyclone separators. Trans. Instn. Chem.
Engrs. 29, 356-383.
STERN A. C., CAPLANK. J. and BUSH P. D. (1955) Cyclone Dust Collectors, Am. Petrol. Inst. Report.
SWIFT P. (1969) Dust control in industry--2. Steam Heat. Engr. 38, 453-456.
VAN EnnENHORSXTENGnERGENH. J. (1965) Comparative studies with cyclones. Staub 25(11), AA A9"

Potrebbero piacerti anche