9588ch21 frame Page 305 Wednesday, September 5, 2001 10:07 PM

Cyclone Design
21
Cyclones are very common particulate control devices used in many applications,
especially those where relatively large particles need to be collected. They are not
very efficient for collecting small particles because small particles have little mass
that can generate a centrifugal force. Cyclones are very simple devices that use
centrifugal force to separate particles from a gas stream. They commonly are con-
structed of sheet metal, although other materials can be used. They have a low capital
cost, small space requirement, and no moving parts. Of course, an external device,
such as a blower or other source of pressure, is required to move the gas stream.
Cyclones are able to handle very heavy dust loading, and they can be used in high-
temperature gas streams. Sometimes they are lined with castable refractory material
to resist abrasion and to insulate the metal body from high-temperature gas.
A typical cyclone is illustrated in Figure 21.1. It has a tangential inlet to a
cylindrical body, causing the gas stream to be swirled around. Particles are thrown
toward the wall of the cyclone body. As the particles reach the stagnant boundary
layer at the wall, they leave the flowing gas stream and presumably slide down the
wall, although some particles may be re-entrained as they bounce off of the wall
back into the gas stream. As the gas loses energy in the swirling vortex, it starts
spinning inside the vortex and exits at the top.
The vortex finder tube does not create the vortex or the swirling flow. Its function
is to prevent short-circuiting from the inlet directly to the outlet. Cyclones will work
without a vortex finder, although the efficiency will be reduced.

21.1 COLLECTION EFFICIENCY

When a particle moves at a constant speed in a circular direction, the velocity vector
changes continuously in direction, although not in magnitude. This creates acceler-
ation resulting from a change in direction of the velocity, which is just as real and
just as much an acceleration as that arising from the change in the magnitude of
velocity. By definition, acceleration is the time rate of change of velocity, and
velocity, being a vector, can change in direction as well as magnitude. Force, of
course, is defined by Newton’s Second Law (F = ma). Centrifugal force is given by:

mV 2
F= (21.1)
r
where
F = centrifugal force
m = mass of particle
V = velocity of particle, assumed to equal inlet gas velocity
r = radius of cyclone body

© 2002 by CRC Press LLC

9588ch21 frame Page 306 Wednesday, September 5, 2001 10:07 PM

FIGURE 21.1 Schematic of standard cyclone.

Because the operating principle of a cyclone is based on using centrifugal force

to move particles to the cyclone wall, a simple mistake in the piping configuration,
shown in Figure 21.2a, reduces efficiency. Ensure that particles are given a head
start in the right direction by using the configuration shown in Figure 21.2b.

21.1.1 FACTORS AFFECTING COLLECTION EFFICIENCY

Several factors that affect collection efficiency can be predicted. Increasing the inlet
velocity increases the centrifugal force, and therefore the efficiency, but it also
increases the pressure drop. Decreasing the cyclone diameter also increases centrif-
ugal force, efficiency, and pressure drop. Increasing the gas flow rate through a given
cyclone has the effect of efficiency shown in Equation 21.2:

0.5
Pt 2  Q1 
= (21.2)
Pt 1  Q 2 

where
Pt = penetration (Pt = 1 – η)
η = particle removal efficiency
Q = volumetric gas flow

© 2002 by CRC Press LLC

9588ch21 frame Page 307 Wednesday, September 5, 2001 10:07 PM

FIGURE 21.2 Inlet piping configuration.

Interestingly, decreasing the gas viscosity improves efficiency, because drag

force is reduced. Centrifugal force drives the particle toward the wall of the cyclone,
while drag opposes the centrifugal force. The terminal velocity of the particle toward
the wall is the result of the force balance between the centrifugal and drag forces.
Increasing gas to particle density difference affects penetration as shown in
Equation 21.3:

0.5
Pt 2  µ 2 
= (21.3)
Pt1  µ1 

where: µ = gas viscosity. Note that decreasing the gas temperature increases the gas
density, but contrary to intuition, decreases the gas viscosity, which reduces drag
force and results in a small efficiency improvement. However, decreasing the gas
temperature also decreases the volumetric flow rate, which affects efficiency as
described above in Equation 21.2.
Finally, particle loading also affects efficiency. High dust loading causes particles
to bounce into each other as they move toward the wall, driving more particles
toward the wall and their removal.

0.18
Pt 2  L1 
= (21.4)
Pt1  L 2 

where L = inlet particle concentration (loading).

© 2002 by CRC Press LLC

9588ch21 frame Page 308 Wednesday, September 5, 2001 10:07 PM

FIGURE 21.3 Generalized efficiency relationships.

Figure 21.3 shows generalized efficiency relationships for high-efficiency con-

ventional and high-throughput cyclones. It simply demonstrates that the dimensions
of the cyclones can be tuned to the application. Figure 21.4 and Table 21.1 illustrate
typical cyclone dimensions. Relative dimensions are based upon the diameter of the
body of the cyclones. High-efficiency cyclones tend to have long, narrow bodies,
while high-throughput cyclones generate less pressure drop with fat bodies.

21.1.2 THEORETICAL COLLECTION EFFICIENCY

The force balance between centrifugal and drag forces determines the velocity of
the particles toward the wall. Resident time of particles in the cyclone, which allows
time for particles to move toward the wall, is determined by the number of effective
turns that the gas path makes within the cyclone body. An empirical relationship for
the number of effective turns is provided in Equation 21.5:

1 L 
Ne = L + c (21.5)
H b 2 

where
Ne = number of effective turns
H = height of the tangential inlet
Lb = length of cyclone body
Lc = length of cyclone lower cone

The theoretical efficiency of a cyclone can be calculated by balancing the

terminal velocity with the residence time resulting from a distance traveled in the
cyclone. This force and time balance results in Equation 21.6:

© 2002 by CRC Press LLC

9588ch21 frame Page 309 Wednesday, September 5, 2001 10:07 PM

FIGURE 21.4 Cyclone dimensions.

TABLE 21.1
Typical Cyclone Dimensions
High Efficiency Standard High Throughput

Inlet height H/D 0.44 0.5 0.8

Inlet width W/D 0.21 0.25 0.35
Gas exit diameter De /D 0.4 0.5 0.75
Body length Lb /D 1.4 1.75 1.7
Cone length Lc /D 2.5 2.0 2.0
Vortex finder S/D 0.5 0.6 0.85
Dust outlet diameter Dd /D 0.4 0.4 0.4

0.5
 x 9µW 
d px =  (21.6)
 (
100 π N e Vi ρp − ρg ) 


where
dpx = diameter of a particle with x% removal efficiency
µ = viscosity

© 2002 by CRC Press LLC

9588ch21 frame Page 310 Wednesday, September 5, 2001 10:07 PM

FIGURE 21.5 Lapple’s efficiency curve.

W = inlet width
Ne = number of effective turns
Vi = inlet velocity
ρp = density of particle
ρg = density of gas

21.1.3 LAPPLE’S EFFICIENCY CORRELATION

Unfortunately, the theoretical efficiency relationship derived above does not correlate
well with real data. The relationship works reasonably well for determining the 50%
cut diameter (the diameter of the particle that is collected with 50% efficiency). To
better match data with reasonable accuracy, the efficiency of other particle diameters
can be determined from Lapple’s empirical efficiency correlation,1 which is shown
in Figure 21.5. This correlation can be set up for automated calculations using the
algebraic fit given by Equation 21.7:

1
ηj = 2 (21.7)
d 
1 +  p 50 
 d pj 

© 2002 by CRC Press LLC

9588ch21 frame Page 311 Wednesday, September 5, 2001 10:07 PM

FIGURE 21.6 Effect of slope parameter, B.

where
ηj = collection efficiency of particle with diameter j
dp50 = diameter of particles with 50% collection efficiency
dpj = diameter of particle j

Lapple’s efficiency curve was developed from measured data for cyclones with
the “standard” dimensions shown in Table 21.1. The efficiency curve can be tailored
for different industrial cyclone dimensions by adding a slope parameter, B, to the
correlation:
1
ηj = B (21.8)
d 
1 +  p 50 
 d pj 

where B = slope parameter, typically ranging from 2 to 6.

Figure 21.6 illustrates the effect of the slope parameter, B. Note that the larger
value for B results in a sharper cut. Since more mass is associated with larger
particles, the sharper cut results in higher overall mass removal efficiency.

21.1.4 LEITH AND LICHT EFFICIENCY MODEL

Other models have been developed to predict cyclone performance. One is the Leith
and Licht model2 shown in Equation 21.9:

© 2002 by CRC Press LLC

9588ch21 frame Page 312 Wednesday, September 5, 2001 10:07 PM

(
η = 1 − exp − Ψ d M
p ) (21.9)

1
M= (21.9a)
m +1

 T  
0.3
m = 1 −  1 − 0.67D0c .14 
( )
 283  
(21.9b)


M
 K Qρp C′ (m + 1)  2
Ψ = 2  (21.9c)
 18 µ D3C 

where
dp = particle diameter in meters
DC = cyclone body diameter in meters
T = gas temperature, °K
K = dimensional geometric configuration parameter
Q = volumetric gas flow
ρp = particle density
C′ = cunningham slip correction factor
µ = gas viscosity

The geometric configuration parameter is estimated based on the cyclone con-

figuration. Table 21.2 shows relative dimensions for three types of cyclones: the
standard cyclone, the Stairmand design,3 and the Swift design.4 Note that the Stair-
mand and the Swift cyclones have smaller inlet openings than the standard design,
which means a higher inlet velocity for the same size body. This results in more
centrifugal force and increased efficiency. In the Leith and Licht model, a larger
geometric configuration parameter results in a higher predicted efficiency.

TABLE 21.2
Geometric Configuration Parameter
Standard Stairmand Swift

Inlet height H/D 0.5 0.5 0.44

Inlet width W/D 0.25 0.2 0.21
Gas exit diameter De /D 0.5 0.5 0.4
Body length Lb /D 2.0 1.5 1.4
Cone length Lc /D 2.0 2.5 2.5
Vortex finder S/D 0.625 0.5 0.5
Dust outlet diameter Dd /D 0.25 0.375 0.4
Geometric configuration paramater K 402.9 551.3 699.2

© 2002 by CRC Press LLC

9588ch21 frame Page 313 Wednesday, September 5, 2001 10:07 PM

FIGURE 21.7 Cyclone efficiency curves.

21.1.5 COMPARISON OF EFFICIENCY MODEL RESULTS

Efficiency models are adequate for getting a fair idea of performance, but there can
be a rather wide variation in model predictions. Part, but not all, of the variation
can be explained by empirical factors for the cyclone configuration. Figure 21.7
shows cyclone efficiency curves as a function of particle diameter based on several
sources. Each curve is based upon the same gas flow and gas and particle conditions.
The lowest efficiency is predicted by Lapple’s curve for a standard cyclone. Inter-
estingly, the Leith and Licht model for the same standard cyclone predicts a signif-
icantly higher efficiency. The Leith and Licht model for the higher efficiency Stair-
mand and Swift cyclone designs shows incremental improvement over the standard
design. Vendor data also were collected for the same set of gas and particle condi-
tions, with significant predicted performance improvement. Perhaps the vendors
were being overoptimistic about their designs, or perhaps there have been significant
improvements in cyclone design over the years. It does point out that performance
guarantees for cyclones must be written with specific information about the gas and
particle properties, including the particle size distribution, to ensure that vendor
guarantees can be measured and substantiated after installation.

21.2 PRESSURE DROP

Pressure drop provides the driving force that generates gas velocity and centrifugal
force within a cyclone. Several attempts have been made to calculate pressure drop
from fundamentals, but none of them has been very satisfying. Most correlations
are based on the number of inlet velocity heads as shown in Equation 21.10:

© 2002 by CRC Press LLC

9588ch21 frame Page 314 Wednesday, September 5, 2001 10:07 PM

1
∆P = ρ V2 N (21.10)
2gc g i H

where
∆P = pressure drop
ρg = gas density
Vi = inlet gas velocity
NH = pressure drop expressed as number of the inlet velocity heads

One of the correlations for number of inlet velocity heads is by Miller and
Lissman:5

2
 D
N H = K ∆P1   (21.11)
 De 

where
K∆P1 = constant based on the cyclone configuration and operating conditions
D = diameter of the cyclone body
De = diameter of the exit tube

A typical value for K∆P in the Miller and Lissman correlation is 3.2. For the
standard cyclone configuration described above, the Miller and Lissman correlation
results in 12.8 inlet velocity heads.
Another correlation for number of inlet velocity heads is by Shepherd and
Lapple:6

HW
N H = K ∆P 2 (21.12)
De2

where
K∆P2 = constant for cyclone configuration and operating conditions
H = height of the inlet opening
W = width of the inlet opening
De = diameter of the exit tube

The value for K∆P in the Shepherd and Lapple correlation is different, typically
ranging from 12 to 18. The Shepherd and Lapple correlation results in 8 inlet velocity
heads for the standard cyclone dimensions, 6.4 inlet velocity heads for the Stairmand
cyclone design, and 9.24 inlet velocity heads for the Swift cyclone design. As can
be seen, there is a substantial difference among the correlations. Again, it is best to
rely upon vendors’ experience when your own experience is lacking; however, to
enforce a performance guarantee, ensure that the specification is well-written and
can be documented for the expected conditions.

© 2002 by CRC Press LLC

9588ch21 frame Page 315 Wednesday, September 5, 2001 10:07 PM

21.3 SALTATION
The previous discussion of efficiency and pressure drop leaves the impression that
continually increasing the inlet gas velocity can give incrementally increasing effi-
ciency. However, the concept of “saltation” by Kalen and Zenz7 indicates that, more
than just diminishing return with increased velocity, collection efficiency actually
decreases with excess velocity. At velocities greater than the saltation velocity,
particles are not removed when they reach the cyclone wall, but are kept in suspen-
sion as the high velocity causes the fluid boundary layer to be very thin. A correlation
for the saltation velocity was given by Koch and Licht:8

  W  0.4 
( )
0.333
 ρp − ρg   
  D 
Vs = 2.055D0.067Vi0.667 4gµ   0.333  (21.13)
 3ρ2g   1 −  W   
 
   D   

where
Vs = saltation velocity, ft/s
D = cyclone diameter, ft
Vi = inlet Velocity, ft/s
g = acceleration of gravity, 32.2 ft/s2
µ = gas viscosity, lbm/ft-sec
ρp = particle density, lbm/ft3
ρg = gas density, lbm/ft3
W = width of inlet opening, ft

The maximum collection efficiency occurs at Vi = 1.25Vs, which typically is between

50 and 100 ft/s.

REFERENCES
1. Lapple, C. E., Processes use many collector types, Chem. Eng., 58, 5, May 1951.
2. Leith, D. and Licht, W., The collection efficiency of cyclone type particle collectors —
A new theoretical approach, AIChE Symp. Series, 126 (68), 1972.
3. Stairmand, C. J., The design and performance of cyclone separators, Trans. Ind.
Chem. Eng., 29, 1951.
4. Swift, P., Dust control in industry, Steam Heating Eng., 38, 1969.
5. Miller and Lissman, Calculation of cyclone pressure drop, presented at meeting of
American Soc. of Mech. Eng., New York, December 1940.
6. Shepherd, C. B. and Lapple, C. E., Flow pattern and pressure drop in cyclone dust
collectors, Ind. Eng. Chem., 32(9), 1940.
7. Kalen, B., and Zenz, F., Theoretical empirical approach to saltation velocity in cyclone
design, AIChE Symp. Series, 70(137), 1974.
8. Koch, W. H. and Licht, W., New design approach boosts cyclone efficiency, Chem.
Eng., 84(24), 1977.

© 2002 by CRC Press LLC