Professional Documents
Culture Documents
Contents
1. Introduction 67
2. Qualitative Description of Fluid and Particle Movements. 70
3. Calculation of Fluid Dynamic Properties. 75
4. Calculation of Separation Efficiency. . . 80
4.1 Mechanism of Particle-Gas Separation. . 80
4.2 Calculation of Separation Efficiency and Dust Emission Rate . 83
4.3 Adaptation of Fractional Separation Efficiency to Special Conditions 85
5. Application of Computer Programs to the Calculation and Design
of Cyclones. . . . . . . . . . . . . 88
5.1 Definition of the Aims of Calculations. . . . 89
5.2 Structure of Computer Programs . . . . . . 89
5.3 Range of Application of Computer Programs. 91
6. Discussion of the Results Obtained by Computer Programs. 92
6.1 Basic Assumptions . . . . . . 92
6.2 One Cyclone. . . . . . . . . . 93
6.3 Cyclones in Serial Arrangement. . 95
6.4 Cyclones in Parallel Arrangement. 98
6.5 Comparison of Separation Efficiency for Various Arrangements of Cyclones 101
7. Some Remarks on Cyclone Design 103
8. List of Symbols. 104
9. References . . . 106
1. Introduction
The centrifugal force increases with decreasing radius r. This equation indicates
that the separation efficiency of a small cyclone is better than that of a large
cyclone.
2. How to generate a rotational motion of gas and particles? By tangential
introduction of gas and particles into a cylindrical element.
3. How to withdraw gas and dust particles after separation? By separate
withdrawal at opposite ends of the cyclone, of the gas by pressure force and of
the dust by gravity force.
Fig. 1 schematically describes a cyclone of conventional design. It consists of a
cyclindrical entrance chamber, a conical dust separation or vortex chamber, an
inlet channel, an outlet tube with extension into the cylindrical entrance
chamber, a dust collection chamber, and a conical vortex spoiler which sep-
arates the separation chamber from the collection chamber, thereby preventing
dust re-entrainement into the gas.
Some of the important properties of the cyclone are as follows.
1. In the cyclone, direct dust-gas separation is possible without introducing any
separation agent such as liquid drops. The cyclone is a dry dust separator
that does not generate new pollution problems.
2. The cyclone is designed for a continuous separation process with continuous
withdrawal of gas and dust.
3. The technical attractiveness is at least partially due to its simplicity of design.
There is no technical sophistication.
4. The physical phenomena of gas-dust separation in a cyclone are relatively
well understood. This is a sound basis for a mathematical description of the
vat;tex
chamber
dust collection
chamber
process. Pressure drop and separation efficiency can be fairly well determined
in advance.
5. The particle cut diameter is in the range from 2 to Sllm, with a fair chance to
reduce it still further.
6. If the cut diameter does not meet the requirements, then the cyclone may be
applied as a pre-separator.
7. The cyclone may be used either as a single unit or with many units operated
in parallel.
8. Low cost for investment and for operation.
The photographs in Fig. 2 demonstrate design by various companies and
technical application of cyclones. Fig. 2a depicts a single cyclone. An arrange-
ment of three cyclones operated in parallel is shown in Fig. 2b. The clean gas
tubes from three cyclones are introduced into one collection unit. Typical
The main elements of a cyclone have already been described by Fig. 1. The
cyclone essentially consists of an entrance chamber, separation chamber, dust
collection chamber, and gas outled tube. There are various designs for the inlet
element available. The more important ones are presented in Fig. 3: tangential
inlet, spiral inlet, and axial inlet. The mean gas velocity in the entrance section is
Ve. The clean gas leaves the cyclone through the exit tube having the mean axial
velocity Vi. The cross sectional areas are for the inlet Fe and for the outlet Fi.
The characteristic velocities of the fluid within the cyclone, the trajectory of a
particle from the inlet to the outlet, and the streamlines of the axial flow
component are given in Fig. 4. Within the cyclone a three-dimensional flow is
established, which is characterized by a rotational movement, superposed in the
outer ring space by an axial movement directed toward the dust collection
chamber, and in the inner cylindrical space by an axial movement directed
toward the exit tube. The axial movements in the inner and outer space of the
2. Qualitative Description of Fluid and Particle Movements 71
tv,-
e I
a c
particle trajectory
secondary m- -.--++nm.- - m
flow streamlines
of axial flow
particle components
trajectory
inlet
channel
Fig. 4. Fluid and particle movements in a
cyclone
Neglecting the non-slip condition at the cyclone wall, the tangential velocity at r
=ra is given by ua. In the direction of the cyclone axis, the velocity u increases
from U a to its maximum value U i at r = r i . With further decreasing radius, the
tangential velocity rises too; at r = 0 the symmetry condition demands u = O. At
the surface of the exit tube with r = r i , the tangential velocity is of course zero.
At the surface of the inner cylindrical element i-i, the tangential velocity U i
changes scarcely. A slight decrease of U i is observed close to the dust collection
chamber and may safely be neglected. Furthermore, the radial fluid velocity Vr
may be assumed to be constant over the height of the cylindrical surface i-i.
Small deviations from the constant value are observed only near the entrance
to the exit tube t - t and the dust collection chamber.
The particles which are transported by the gas streams into the cyclone will
be subjected to the action of the centrifugal, frictional, and other forces typical
of the three-dimensional flow field. The centrifugal force is due to the rotational
movement of the fluid while the frictional force is due to the movement of the
fluid in the radial direction toward the axis of the cyclone. The acceleration u 2/r
caused by the centrifugal force is in many cases 100 or even 10,000 times higher
than the acceleration caused by the attraction of the earth.
Under the influences of the combined action of centrifugal and frictional
forces, all particles move on a spiral-like path as shown in Fig. 5. The larger
particles move along the spiral outwards while the smaller particles move along
the spiral inwards. Outward moving large particles are separated from the gas;
they will collide with the cyclone wall and move toward the dust collection
chamber. Inward moving small particles are not be separated from the gas
stream but carried away by it, leaving the cyclone through the exit tube.
The rotational fluid movement establishes a pressure field that has its maxi-
mum near the cyclone wall and its minimum in the cyclone axis. The pressure
gradient in the radial direction, d pldr, is given by the equation:
dp u2
-=p - .
dr r
The pressure increases with radius r, tangential velocity u, and fluid density p; Cis
a constant of integration. One may safely assume that the pressure Pa at the
cyclone wall may be determined by the above equation, neglecting the zero
velocity condition at the wall. The pressure gradient will force small particles to
follow a secondary fluid motion indicated by arrows in Fig. 4. Above plane t - t
that is above the entrance to the exit tube. small particles will move upwards
along the cyclone wall, then follow the fluid in a radial direction toward the exit
tube where they fall downwards along the tube and reach plane t - t. Here. the
small particles may be picked up again by the fluid that enters the exit tube. The
secondary motion of the fluid in the upper part of the separation chamber will
decrease the collection efficiency of the cyclone. This can be prevented by the
installation of a collar around the gas outlet tube.
In the bottom part of the cyclone, that is the conical element in Fig. 4, the
secondary motion of the fluid supports the motion of the particles toward the
collection chamber.
Furthermore, the secondary motion supports the formation of particle strands
close to the entrance to the cyclone and their movement on spiral- and screw-
like paths down the cyclone wall. Strand formation will be observed especially
at high dust loading of the raw gas.
Important properties of fluid flow in the cyclone may be deduced from Fig. 6
in which the pressure at various cross sections in the cyclone is given on the
basis of experimental evidence available. According to Fig. 1 cross section e - e
is to be found at the entrance of the cyclone, cross section i - i below the exit
tube in the separation chamber, cross section t - t at the entrance of the exit
tube, and cross section m - m at the outlet of the exit tube. For a dimensionless
representation of the pressure in the cyclone giving a maximum of information
on the fluid flow, the overall pressure Llp* and the static pressure LlP~at have
been used. These two quantities are defined as follows:
p and Pstat are the local pressure values at the considered cross sections, Pm and
Pm, stat are the values at cross section m - m at the outlet of the exit tube, and p
- Pm and Pstat - Pm, stat local pressure differences related to the conditions at the
cyclone outlet. p vt /2 is related to the dynamic pressure of the fluid flow in the
exit tube; p is the fluid density and Vi the mean translational velocity in the exit
tube.
In order to simplify the considerations, the pressure differences Llp* and LlP~at
at the defined cross sections have been connected by straight lines. According to
74 Chapter 4: Design and Operation of Cyclones
e m
Fig. 6 the sharpest decrease in overall pressure L1 p* is observed in the exit tube
between cross sections t - m and the inner cylindrical element between cross sec-
tions i - t. The extremely great pressure drop in the exit tube is due to a very strong
rotational movement superimposed upon the axial fluid movement. This not only
causes the great friction losses at the wall of the exit tube but also the formation
of a low pressure region in the axis of the exit tube. Into this low pressure space,
fluid moves countercurrently to the primary axial movement. The axial flow
field in the exit tube is schematically illustrated in Fig. 7. Because of the very
great pressure drop and its importance for the collection efficiency, the exit tube
is considered to be the most important element of the cyclone. The important
dimensions of the exit tube are the radius r i and the length of protrusion into the
separation chamber, h - hi'
The curve given for the static pressure in Fig. 6 is quite different from that
characterizing the overall pressure, although it reflects the same physical conditions.
The strongest decrease in static pressure is observed between the cross sections i
- t where the most complicated flow field with rotational and countercurrent
axial movement exists. At a first glance, it will surprise that the static pressure
increases along the exit tube. This is due to the fact that the rotational
movement is reduced while the fluid is flowing through the exit tube.
The friction factors (e -i' ( i- m and (z given in Fig. 6 will be discussed in
Section 3.
3. Calculation of Fluid Dynamic Properties 75
v =-
v Mean velocity at the entrance of cyclone (5)
e Fe
v·=-
v Mean velocity in exit tube of cyclone (6)
, F,
Fig. 1 gives the mean velocities in the mentioned cross sections of the cyclone.
For particle separation in the vortex chamber, the tangential velocity u and the
radial velocity vr are of great importance. Equations for the inner and outer
tangential velocity U j and ua are listed in Table 2. The outer tangential velocity
u a is a function of the entrance coefficient IJ.. given for three different designs of
the entrance element (Fig. 3).
For the calculation of the inner tangential velocity U j , the wall friction coef-
ficient Amust be known. The equations used for the calculation of A are listed in
Table 3. Eqs. (20) and (21) may be employed for an approximate calculation of
A as a function of the dust loading ratio Jl. Eqs. (23) and (24) give the wall
friction coefficient in the case of a dust free gas, stream and for a technically
smooth cyclone wall, with k s /r.::;::;:2.6.1O- s .
76 Chapter 4: Design and Operation of Cyclones
Tangential inlet:
In Figs. 8 and 9 the wall friction coefficients for a dust free gas stream are plotted
against the Reynolds number. In the first case the gas stream is in a cyclone with a
cylindrical vortex chamber and in the second case in a cyclone containing a conical
vortex chamber. Curves a apply to technically smooth, curves band c to rough
cyclone walls. The roughness parameter ks/ra is the ratio of sand roughness ks to
cyclone radius ra' In Fig. 8 curve a is described by Eq. (23) and in Fig. 9 by Eq.
(24). Experimental data fit fairly well curves a.
3. Calculation of Fluid Dynamic Properties 77
(19)
Approximate solution:
/1<1
Al =";dl+2~) (20)
/1>1
Al=Adl+3~) (21)
Loading ratio /1:
Mp YM
/1 == --;- = .-=:J' (22)
M Vp
(24)
(27)
Froude number Fr i :
v·
Fr =--'- (28)
i - y'2r i g
(29)
78 Chapter 4: Design and Operation of Cyclones
2
I I II
I
. I . I
,~ cylmdrlcal vortex chamber
b'\. I
"" "\
~ :-- """c
k 5 /rg :6. /0, 3
'" :::-...
~
J-
1. /0, 3
,
a ~~ n ~2.6.1(j5
~ 0
technically smooth
cyclone wall
2
I
, 6 BId
Reynolds number Re z
Fig. 8. Wall friction factor A.G for a cyclone with a cylindrical vortex chamber
"-
1.. I I I
, ~ conical vortex chamber
k$ /ro=6. 10-1
....:'"
'-
2
~~ ..-:; !---
I
/./0,3
>-
0
<:; f-"'"
~~ e--
~
10' 2 I
~ B ---
........ ....... r--..!!
,
.~
or-- r:-.
6
....'- a 0,,2.6.10 5
0
ti technically smooth
~ cyclone wall
2
I
~ 6 B /0'
Reynolds number Rez
Fig. 9. Wall friction factor A.G for a cyclone with a conical vortex chamber
The overall pressure drop Ll Pz for cyclone flow may be considered as a sum of
two parts: LlPe_i and LlPi_m' The pressure drop LlPe_i occurs between cross
sections e - e and i - i while that of Ll Pi _ m occurs between cross sections i - i and
m - m. The cross sections are indicated in Fig. 1.
The pressure drop Ll Pz in the cyclone may be calculated by means of the
equations listed in Table 4.
For cyclones with spiral inlet an additional pressure drop Ll Ps has to be taken
into account. In Fig. 10 the second fraction of the cyclone friction factor (i-m is
plotted against the velocity ratio u/v i and compared with experimental data.
3. Calculation of Fluid Dynamic Properties 79
LIps
(s= pv;J2 friction factor of spiral inlet (36)
(z=(e-i+C-m (37)
Cyclones with spiral inlet :
(39)
(40)
1 0.886)2 ( 1)
(s= (F*+r*F* 1-()(2
(41)
e a e S
R• . =7.5.10 2
I
9
6 R~ =2 ./03
, R:. ~ 5 . /0' V
D
0
~
If
..,:,.-
~l>
2
,E
.,gCO
.!,! /0 '
;j
.::: 8 ///
...CO V~
c 6 Fig. 10. Cyclone friction
<:;
~ , .--; V
./
V V factor ( i-m as a function of
L--- the velocity ratio uJv i and
it'" V Rei as a parameter, \7: Re
------
0
2 r- =7.5.10 2, 0: Re=2 . 10 3,0:
/0-1 2 6 8 10 0 2
Re~5. 104
velocity ratio ui Iv,
80 Chapter 4': Design and Operation of Cyclones
entrance channel
entrance and
vortex
chamber
J •
~c
Kc 1 d~pp u
- ---
Kr 18 fl1
In these equations dp is the particle diameter, Pp the particle density, u the gas
velocity, r the radius of curvature, IJ the gas viscosity, and Wr the relative velocity
between gas and particle. For simplicity, it has been assumed that wr=u. This
assumption favors frictional force. However, the ratio Kc/Kr still indicates
predominance of the centrifugal force. The consequence is that almost all
particles will move along an almost straight trajectory and hit the cyclone wall.
Moving along a straight path the particles will be separated from the gas. This
process is called "primary separation".
It may be objected that turbulence will hinder or even prevent primary
separation. The ratio of centrifugal force Kc to turbulent momentum flux K t is
given by:
Particles 3 and 5 are very small. After collision with the cyclone wall at points
3 a and 5 a, they are reflected. At the points of collision the particles transfer
some of their energy to the wall. After collision, their velocity and the centrifugal
force are reduced. These particles will therefore get under the strong influence of
the now prevailing frictional forces of the radial gas stream and will be carried
into the exit tube at point 3 band 5 b as indicated in Fig. 11. The centrifugal
forces are too weak to be effective in this case. Secondary separation has not
been successful.
Particles 2 and 4 are assumed to be rather large. They hit the cyclone wall at
points 2a and 4a and are reflected. Because of its size, particle 2 remains under
the influence of the dominating centrifugal force and is separated a second time.
For particle 4 three different trajectories are depicted in Fig. 11. From the
point of collision 4a to point 4 b, the three trajectories are identical. At this
point, turbulent momentum is transferred so that the particle is pushed closer to
the outlet tube. In this region, the frictional forces, due to the high gas velocity
in the radial direction, dominate over the centrifugal forces and carry the
particle into the outlet tube. Because of the action of turbulent momentum, the
secondary separation has not been successful.
The second trajectory, which will be discussed, leads from point 4b to point
4c. In this case, the centrifugal force is dominated by the frictional force directed
in the tangential direction. The particle approaches the cyclone wall asymptoti-
cally, and will therefore not be reflected again. The secondary separation process
has been successful.
The third trajectory to be discussed leads from point 4b to point 4c', and from
there to point 4c". In the first section of the trajectory, from 4b to 4c', the
frictional force in the tangential direction was not strong enough to ensure an
asymptotic approach to the wall. Consequently, collision with the wall and
reflection had to take place. The final asymptotic approach leads to point 4c".
There is a fundamental difference in primary and secondary separation,
although in both cases the centrifugal force is the "separating force". In primary
separation, almost all particles hit the wall and will be separated, independent of
their size. This means that the diameter distribution of the particles separated in
the first step is identical with the distribution of the particles at the cyclone inlet.
Following the re-entrainment process, secondary separation takes place in the
vortex chamber. Separation in a vortex field is a selective process. Large
particles are more easily separated from the fluid than small ones. The sepa-
ration process in a vortex field is described by a fractional separation efficiency
curve. It is not yet possible to determine a fractional efficiency curve by theory.
The fractional efficiency curve therefore has to be determined by experiments.
Fig. 12 gives some examples for fractional separation efficiency curves which
have been established by Muschelknautz. Curve a is related to a cyclone with
axial inlet, and curves band c to cyclones with tangential inlet. Table 5 lists the
data of the cyclones tested. For practical applications, Muschelknautz suggests
curve c when the cyclone has a tangential inlet.
Calculations based on curves band c reveal that the shape of the cyclone has
a relatively small effect on the separation efficiency in the range of medium and
high values of the loading ratio /1. This is the reason why Muschelknautz
4. Calculation of Separation Efficiency 83
v- ~ ~
1.0 I-
c:
.~ ch
"0... 0.8
§~
1I)s; 0.6 / 0 V
/'
a
-:0...
01.)
§c: 0.'
If V
~ 1/
.~:~
.......
0 ....
0.2
V
.... OJ
/
o ~
Fig. 12. Fractional separation efficiency ({JF versus particle diameter ratio dJ dp • j for three
types of cyclones (for details see Table 5)
Curve a b c
1
Shape of
W \17
cyclone
suggests the application of curve c for cyclones with tangential inlet and of curve
a for cyclones with axial inlet.
The steeper the gradient of the fractional separation efficiency at If>F=O.5, the
more effective is the cyclone. According to Fig. 12 the separation efficiency of a
cyclone with tangential inlet is much better than that of a cyclone with axial
inlet.
When the loading ratio J1 is greater than the re-entrainment load J1" primary
and secondary separation are effective. After non-selective primary separation in
the entrance chamber, the particles are subjected to the selective secondary
separation process in the vortex field.
(44)
Curve a: qJF, a =
9.90
[ 1.04 + e1.60d p /d p ,i
]-1 (48)
Curve b: (49)
2.55(dp /d p .l· 2
Curve c: (50)
1 + 2.55(dp /d p .l· 2
d.=
P.'
V 18
u;(pp-p)
vr r·"
," (51)
in Fig. 13; V is the volumetric flow rate of the fluid and Mpe and Mpm are the
masses of particles per unit volume of the fluid at cyclone entrance and cyclone
outlet. The separation efficiency of a cyclone is the sum of the efficiencies of
primary separation, ipp, and of secondary separation, ips' Separation in the
vortex field is determined by the fractional separation efficiency curve. The
procedure for the calculation of ips is explained in Fig. 14.
In Table 6 all equations necessary for the calculation of collection efficiency,
particle emission rate, and particle size distribution curve are listed.
According to Eq. (47) the re-entrainment load flr is a function of the mean
settling velocity wsk of a particle collective, which again depends on the settling
velocity Ws of a single particle. Equations for the calculation of ws, k are listed in
Table 7.
Eq. (60) is described by the curve given in Fig. 15, and Eq. (61) by curve a given
in Fig. 16.
Table 7. Settling velocities of single particles and mean settling velocities of particle
collectives
(56)
Stokes range
Res:O:;: 1:
24
(=- Stokes friction factor (57)
Res
(pp - p)gd; settling velocity of single
(58)
Ws 18'1 particles in the Stokes region
(62)
(63)
,...
"-" 6
....
o
t 2
.,(? /
.,~
c:: 10
.~
t
6
-I "
:s
l
2 ["I.1a
100
" :-. a
b c
6
2
I •
··f .,-
10' /
-
/0 :;:; ' 6 10 -/2 l 6 /0 0:; ' 6 10 /:; ' 6 10 2 2 l 6 10 3 :; l 6 10':; l 6 /0 5:; '6 10 6
10
P
/
~
. . "'c
10'
/'
IOJ V
.
,"
Ki
~
V
~
"
c:
Id
~... /CI /
1Il'
by I
W
,(/
V ~ ~ d ~ d d III IrJO
Archimedes number Ar. t9dJ)IV1feple)-1
Fig. 16. Settling velocity number for solid spherical particles
1.0
<::
:30
...0
~
"' S} 0.5 0.5
o t;-
<::<::
..g.~
~~
.......
.... '"
0
0 0.5 1.0 1.5 2.0
p article diameter ratio dp/dpJ
fractional collection efficiency curve may be of advantage when, for example, the
cyclone collects more particles in the small size range as expected.
An equation for <PF that is of rather general importance has been developed
by Spilger [13]:
[ 1 d JS
<PF
4s~ (65)
1+ [ - 1 - dp JS'
X O. 5 d p . i
In this equation XO. 5 has been introduced for the sake of abbreviation:
x o. 5 =(d p /d p)o.5' The gradient at X O. 5 is given by:
(X O. S)5
d<pF I (66)
d(dp/d p ) Xo 5
1+(_1 )5
X O. 5
With s 21.5 the curves will assume the shape of an s. The curves presented in
Fig. 17 have been calculated by means of Eq. (65) for various values of s,
assuming xo.s = 1.
wall friction
factor A
Reynolds'
numbers Rei ,Rez
cyclone friction
factor ~z
re-entrainment
load J1.,
mean settling
velocity wsk
Fig. 18. Modular computer program for cyclones, when dimensions of cyclones are given
waif friction
factor A
cyclone friction
factor ~z
re-entrainment
load J1.r
mean settling
velocity wsk
Fig. 19. Modular computer program for cyclones at given pressure drop
wall friction factor Ie must be known. A special module for the calculation of ), is
available. Furthermore, a special module for the cyclone friction factor (z has
been prepared in order to calculate the pressure drop Ll Pz in the cyclone.
When, as in the second case, the pressure drop Ll pz is given, the fluid dynamic
calculations may again be carried out in three steps:
b(i) Calculation of inlet coefficient,
b(ii) calculation of velocities in cyclone,
b(iii) calculation of dimensions of cyclone.
For each of these steps a special module is available as shown in Fig. 19. Sub-
moduli have been prepared for the calculation of wall friction factor A, cyclone
friction factor (v and Reynolds' number Rei for fluid flow in the outlet tube of
the cyclone.
5. Application of Computer Programs to the Calculation and Design of Cyclones 91
For both cases considered, the calculation of the separation efficiency follows
the same procedure. The structure of this part of the computer program is also
indicated in Figs. 18 and 19.
A thorough analysis of the computer program revealed that the range of its
safe application is primarily described by Eq. (47) which relates to the re-
entrainment load J1 r • This is the reason why the range of application of this
equation for J1 r will be explained.
The re-entrainment load J1 r depends on 10 dimensionless numbers; three of
these are of primary and 7 of secondary importance. It has been found that only
those of primary importance need to be taken into account when the range of
application is investigated. The numbers of secondary importance will be given
fixed values:
h* ==h/rj = 11.0
r: == r,!r j = 2.0
F:==Fe/Fj =1.1
b/ra =0.57
qJ == 1. - ~pm/Mpe = depending on the three primary numbers
J1 == Mp/M = 100
p/pps 0.0002.
Fig. 20 gives the space of the safe application of Eq. (47) for the re-entrainment
load Pr. This equation may be used when point P, given by coordinates Rei' Ga,
and vJ wsk , lies within the space exhibited in Fig. 20. The location of point P and
its determination is explained in detail in Fig. 21. Point pi is the limiting value of
P; pi is located within the surface of the space of application.
The results discussed in this chapter have been worked out by Spilger [13].
~
I
I I
I
,, I
I
I
'I
type A B
fA 3 4
hlry 23.5 11
h;lr; 20 7.5
~/F; 0.9 0.44
bl& 0.27 0.17
aR 0.76 0.89 Fig. 22. Characteristic dimensions for two types
of cyclones
1.0
Il::
~
~
.~ 0.8
;; \
£ 0.6
~
'" 0.'
\
.~ ~
~_ O. 2 I'--.
o l -I--
r-- t-- Fig. 23. Particle distribution
00 2 6 8 10 12 l' 16 18
curve used as a basis for cyclone
particle diameter dp [J.Lm] computations
9-
'"
~ a6 ~~~~~~
.~
.
.!:!
::::
J.le ~
Q lOa ''""
b 10-1 '"
600 ..
~
c 10-2
d 10-3
8 10 12 I~ 16 18
mean velocity in cyclone outlet tube vi [m/51
Fig. 24. Separation efficiency <p and pressure drop Ll pz for cyclone type A as a function of
mean velocity Vi in cyclone outlet
With increasing mean velocity Vi in the outlet tube, <p and L1 Pz rise. The
calculations could not be carried out in the very low velocity region because this
region is outside of the tested range of application of the equations. The
conventional range of the velocity Vi is from about 6 m/s to 14 m/ s. For a mean
value of Vi = 10 m/s the following results have been obtained:
With type B cyclone, the higher separation efficiency is obtained, but the
pressure drop is higher by a factor of at least 4.
Curves a and b represent cases in which primary and secondary separation
are effective. In the case of curve c, there are two branches c 1 and c 2 , with c 1
being due to secondary and with branch c 2 being due to primary and secondary
separation. The inclusion of primary separation improves the separation ef-
ficiency quite remarkably. According to Eq. (47), the re-entrainment load flr falls
with increasing gas velocity. This means that with rising gas velocity there will
always be a value of the velocity beyond which primary separation will become
6. Discussion of the Results Obtained by Computer Programs 95
1.0 3000
d
0.8 2400
So r-,
~ 'P
'"~
.~ 0.6 180cF.
.!:! ~
::::., ';j
Q.
.~ 0.4 1200~
'0'-
0
curve
a
Jle
10°
.
~
Q. LI)
d 10-3
0
0 2 4 6 8 10 12 14 16 18 20 0
mean velocity in cyclone oullet tube vi [m/s1
Fig. 25. Separation efficiency q; and pressure drop Ll Pz for cyclone type B as a function of
mean velocity Vi in cyclone outlet
Curve J.l c
a 10 0
b 10 - 1
C 10 - 2
d 10 - 3
effective. This situation is described by curve c. For curve d, however, the re-
entrained dust load f.1 r cannot be reduced such that primary separation becomes
effective within the range of gas velocity investigated.
The superiority of cyclone type B over type A with respect to separation is
impressively demonstrated by the plot of particle mass flow rate hlp at the
cyclone outlet versus particle diameter d p (Fig. 26). The total mass flow rate of
dust emitted from cyclone A is about 4.5 times larger than that emitted from
cyclone B. However, the pressure drop in cyclone B is with Ll pz > 3,OOON/m2
much higher than that in cyclone A with Llpz=930 N/m2. These calculations
prove that both types of cyclone may be effectively used for the separation of
dust with a diameter well below 5 11m.
14
curve 1: inlet of cyclones A
and B
12
curve2: outlet of cyclone A
?
"
~10
curve 3: outlet of cyclone B
r j =0.075 m
.::t:.o. V; = 10 m/s
8 p.. = 10-1
~
0
'-
~ 6
!!
....
''0"" 4
E
2
~
0
0 2 4 6 8 10 U " 16 18 20
particle diameter dp {p.m]
Fig. 26. Particle mass flow rate hlp as a function of particle diameter d p
rj = 0.075 m
Vj =10 mls
Jl. e = 10 - 1
TO-I
8
6 2
~
/' 1/
E
e-
j--> V
k"
.,<:
II -~
~
2
r-
10-z
V V
V
!!
v 8 i--"
~ 6
~
I--'
i;; /
!! 2 V
....0
/O-J
-;; 8
., 6
V
/ 3 cyclones arranged in series
cyclone type A
V
~
<;
curve 1: LIp,! : LlPtZ : Llp,J =1 : I : I
.~ 2
2: =I : 2 :3
~ /
10-' 3: =3 : 2 : 1
8
'"0
% 6
.: V
10101 pressure drop : IL1pz =2.74 N/nf
!!
2 / gas flow rale : V=2.5 · /OJ ro1lh
/0-5
10-' 2 4 6 8 10-J 2 .: 6 8 1([2 2 4 6 8 /0-1 2 4 6 8 100 2 .: 6 8 10! 2 4 6 8 10 2
loading ratio at inlet of first cyclone f.l.
Fig. 27. Loading ratio at cyclone outlet Jl m versus loading ratio at cyclone inlet Jle for
cyclones in serial arrangement for various pressure ratios
6. Discussion of the Results Obtained by Computer Programs 97
Curve 1,' All three cyclones have the same pressure drop. Consequently the size
of all three cyclones is the same.
Curve 2,' The pressure drop for each cyclone increases. Consequently the size of
the cyclones decreases.
Curve 3,' The pressure drop for each cyclone decreases. Consequently the size
of the cyclones increases.
The curves presented in Fig. 27 show two distinct branches. In the first
branch, the loading ratio 11m at the outlet is a linear function of the loading ratio
l1e at the inlet. The pressure drop subdivision has no influence on the separation
efficiency in the first branch where collection is due to the action of centrifugal
forces in the vortex only, i.e. when only secondary separation occurs.
In the second branch of the curves, collection of particles is supported by
primary separation. The loading ratio 11m at the outlet of the third cyclone
becomes a rather weak function of the loading ratio l1e' This is of course a great
advantage for the operation of cyclones. The most effective particle separation is
achieved in case 3 for which the pressure drop decreases and the cyclone size
increases.
This was a very surprising result since it had been anticipated, that the most
effective separation would be obtained for case 2 for which the pressure drop
increases with rising number of cyclones. It had been assumed that cyclones
might be considered as sieves that should be arranged such that a sieve with
higher mesh number follows a sieve with smaller mesh number. This assumption
has been proved now to be incorrect when primary and secondary separation
occur.
Fig. 28 describes the loading ratio 11m at the outlet of the last cyclone as a
function of the loading ratio l1e at the inlet of the first cyclone. Five cases have
been considered:
98 Chapter 4 : Design and Operation of Cyclones
1. one cyclone
2. two cyclones in series 2:1
3. three cyclones in series 3:2: 1
4. four cyclones in series 4:3:2 : I
5. five cyclones in series 5:4:3 :2:1
Total pressure drop : ILl Pz= 2,474 N/m2
Gas flow rate: \1=2.5.10 3 m/h
/0"
8
6
~ V
J
"2 ~v
2
~ 10' z L /'~
~ 8
u
~
6
" ;...-" '1"'" 5
2
r-
..... /0·3 ~
o 8 cyclones arranged in ser ies
"
A
6
.~ kf' cyclone Iype A
Fig. 28. Loading ratio at cyclone outlet J.l m versus loading ratio at cyclone inlet J.l e for
cyclones in serial arrangement for varying number of cyclones
The chosen pressure drop subdivision is the most favorable one, as has been
established in the preceding example.
In the first branch of the curves, one cyclone is just as effective as a series of
five cyclones. In the second branch of the curves, the increased effectiveness of
separation is due to the influence of primary separation. When more than three
cyclones are arranged in series, the separation result becomes independent of the
loading ratio Pe .
, >lin
5·10
r-- r-;r.l. V=2.vr
n cyclones in parallel arrangement
?
. I. 5"V
• ,/
~ /0,3 ld ~v o ~
~"·~5·\.9/ /
:; 8
o 6
;;
//. '//.
0 '/
--- ~~\
o
n'~
,0 ./
".5'
-
~
·3 2 ~ ~ /' ~~
~
10" ~ ~v ~
!~ h '/ /
'/j '/jV;
~ ~
2
o~ ~~
10,5 ld ~ / ~
10,5 2 ~ 681cr' 2 ~ 6810-3 2 ~ 6810,2 2 ~ 6810-1 2 '6810° 2 '681d
loading ratio at inlet of cyclone p.
Fig. 29. Loading ratio at cyclone outlet f.1 m over loading ratio at cyclone inlet f.1e for
cyclones in parallel arrangement
if necessary, arranged in parallel because of large gas flow rates. Type B cyclone
has been selected for the calculation of dust separation.
The separation efficiency of a parallel arrangement of cyclones is clearly
demonstrated by the curves in Fig. 29 in which the loading ratio at the cyclone
exit f.1 m is plotted against J1 e at the cyclone entrance. In the first branch of the
curves, which demonstrates a linear relationship between J1 m and J1 e • only
secondary separation in the vortex takes place. In the second branch of the
curves, secondary separation is effectively supported by primary separation. For
J1 e = 10- 1, the following results have been obtained:
n (f.1m)n (f.1m)n/(f.1m)n ~ I
1 5.0.10- 2 1.0
5 7.5.10- 3 0.15
10 4.0 . 10 - 3 0.08
50 1.1 . 10- 3 0.022
100 6.0.10 - 4 0.012
500 2.2.10 - 4 0.0044
'" "
"
"" '"
"'
1.0
0.9
J
n=500
I n=5(j
V!5~1
V =5UG
'i ~./ V /
/ V/
.-
,/
/
V
/
-
/
V
V
.......
1--'-
Q.
0.6 I-n =2
I 1 II I / /
/
V = 1.25'10'
~ II
c:: I.
.!! 0.5 n = I J= 2.J.JO' {mJ/h]
.~
.....
.,
....
0.':
c:: n cyclones in paralle l arrangement
~
e 0.3 cyclone type B
o
/0-' 2 .: 6810-3 2 .: 6810"' 2 .: 6810-1 2 .: 6810° 2 .: 6810'
loading ratio at inlet of cyclone fJ..
./ ./
V V
v::: V-- type B
~!
./ . /r'\ h
d
1/
./
/"
V
./
~ 10"
/
.~ 6
~ ~
J
~ 2
Apl= 2~7~ Nl m 2
Pe =0.1
10-2
10' 2 .: 6 10 2 2 ~ 6 10 3 2 .: 6 /O ~ 2 ~ 6 10 5
volumetric flow rate per cyclone Ii {m 3/ h}
Fig. 32. Dimensions of cyclones versus volumetric flow rate per cyclone
loading ratio at the entrance of the cyclone, Ile = 0.1 . The dimensions are
proportional to the square root of flow rate V. For a volumetric flow rate of V
= 2,500 m 3 /h, the chief dimensions of a cyclone are: d a = 1.3 em] and h = 1.7 em].
10'
8
6
d'
"
2 1/
2 V ~
-
~ 1fT
~ 8 2\
./
u 6
~
.....
" /
")
I----'"
v
./
o 2
I---
~ 10-3 1/ ~r-r
:; 8 3
o 6
/ / cyclone type 8
<; "
~ 2 ~ curve 1 : one cyclone
"... 10', Vl/ 2 : 3 cyclones in series
.S' 86
b
~
3 : 3 cyclones in par allel
o I. ~ V pressure drop L/pz = 21.71. Nlm2
~
2 ~ gas flow ra te Ii = 2500 m3fh
1fT5
1(J' 2 I. 68 10.3 2 I. 6 810-22 I. 68 10-1 2 I. 6 810° 2 I. 6810'
loading ra t io at inlet of cy clone P.
Fig. 33. Comparison of separation efficiency of one cyclone with cyclones in serial and in
parallel arrangement
u r-----------------------------.-------,
cyclone type A
pressure drop L}pz = 21.71. Nlm 2
gas flow rate Ii =2.5"0 3 m 31h
dust loading ratio Po = /0-'
2 I. 6 8 10 12 U 16 20
particle diameter dp [p m]
clean gas
ou tiel iiil 1 ·1 ·1 " ~ ~ 1
'\' \ ' 1/ \ I II \ 11/\ \ LLI
II
aw gas
\ j\ I /\ I /\ ! ~ ~ . ~ 1\ ' l\ ilJ.~ Ldrnlet
. I 17
I
I I I ,
I
I I I
I I I,
[
I
I
I I I I
I
,
II I i
I, [
~.
I I I [
'------I d usl
d isposa/
Fig. 36 gives a view on a poly cyclone which consists of 40 small cyclones with
axial inlet. The long tubes that connect the cyclones with the belt, which carries
away the dust, prevent dust redispersion and gas by-pass. An absolutely uniform
distribution of the gas on all cyclones is not possible. Maldistribution of the gas
will diminish dust separation. This must be taken into account when the number
of cyclones is determined.
8. List of Symbols
9. References
[lJ E. Muschelknautz: Chem.-Ing.-Techn. 44. 63 (1972). - [2J E. Muschelknautz,
N. Rink: Zyklonabscheider. Handb. Fortbildungslehrgang "Mehrphasenstromungen".
VDI-Bildungswerk (1975). - [3J E. Muschelknautz: Hochschulkurs II. Mechanische
Verfahrenstechnik, Verfahrenstechnik 6 (1972) 6 (1972) 3, pp. I/IV ; 5, pp. l/V ; 9, pp. I/IV ; 10, pp.
I/IV; 12, pp. I/IV and 8 (1974) 1. pp. I/IV. - [4J E. Muschelknautz. K. Brun-
ner: Chem.-Ing.-Techn. 39. 531 (1967). - [5J W. Barth: Brennstoff-Wiirme-Kraft 8.
1 (1956). - [6J E. Muschelknautz, W. Krambrock: Chem.-Ing.-Techn. 42, 247 (1970). -
[7J W. Barth, L. Leineweber: Staub-Reinhaltung Luft 24, 41 (1964). - [8J A. Ogawa: 1.
College Eng., Nikon Univ., Series A, 19 157 (1978). - [9] T. Hikichi, A. Ogawa: ibid. 19,
167 (1978). - [1OJ Y. Fuzita, A. Ogawa: ibid. 19, 185 (1978). - [l1J Y. Fujita, A.
Ogawa: ibid. 20, 79 (1979). - [12J T. Hikichi, A. Ogawa: ibid. 20, 155 (1979). - [13J R.
Spilger: Methodik zur Uberflihrung wissenschaftlicher lnformationen aus dem Gebiet
der Verfahrenstechnik in problemorientierte Rechenprogramme, Dissertation, Techn.
Univers. Berlin, 1978.