Beruflich Dokumente
Kultur Dokumente
R EACTOR E NGINEERING
Volume 8
2010
Review R1
Cedric Briens
Franco Berruti
Thierry A. Gauthier
Abstract
Rapid, effective separation of phases is an essential step in many hydrocarbon
conversion gas-solid processes. In heavy oil upgrading, fluid catalytic cracking
(FCC) and biomass pyrolysis, for example, very fast separation of reacting gases
from catalytic or heat-bearing solids is required to inhibit further reaction. The
primary functions of a rapid separator in the context of a gas-solid flow are, first,
to terminate contact between the reacting gases and the catalytic or hot solid particles and, second, to recover the isolated product vapors. Uninterrupted vapor-solid
contact results in degraded, overcracked and generally less valuable products. Increasing interest in rapid separation has led in the past 25 years to a wide variety
of proposed separator designs in the literature.
Centrifugal separation is the fundamental principle underlying cyclone operation
and most rapid gas-solid separation processes. Traditional cyclones rely on centrifugal separation and are very widely applicable. They are used for the removal
of solid particulate matter from gas or liquid streams, for gas demisting, and in
hydrocyclones. Most FCC processes employ traditional cyclone configurations
for secondary and tertiary separation. The review briefly summarizes important
topics studied in reverse flow cyclone operation. For initial separation, however,
most fast gas-solid separators in the literature feature only partial cyclone layouts.
Separation typically occurs more quickly in these devices relative to traditional
cyclones, likely at the expense of separation efficiency.
The current review places emphasis on experimental results and novel separation techniques in the context of fast gas-solid separation. Several recent methods
of rapid gas-solid separation employ co-current or uniflow centrifugal arrangements. Hence, particular attention is directed to co-current centrifugal separators.
The authors express gratitude for the financial support of the Natural Sciences and Engineering
Research Council of Canada, of Agri-THERM Inc. and of the Ontario Centres of Excellence.
Please send correspondence to C. Briens, ICFAR, cbriens@eng.uwo.ca.
These devices have been studied much less intensively than their reverse flow
counterparts. Fast co-current separators, with their novel, partial cyclone layouts,
are interesting for their performance characteristics in relation to popular commercial designs.
KEYWORDS: gas-solid separation, cyclones, fluid catalytic cracking, pressure
drop
INTRODUCTION
Figure 1.1 - Illustrated Comparison of Reverse Flow Versus Co-Current Flow Cyclones, modified
from Gauthier (1990)
Table 1.1 - Summary of Recent Experimental Studies of Centrifugal and Inertial Separation Techniques
Inlet
hydraulic
diameter
[m]
Barrel
diam.
[m]
Inlet
gas
vel.
[m/s]
Gas
flowrate
[m3/h]
Particle
size
range or
mean
diam.
[m]
Particle
density
or bulk
density
(see
refs.)
[kg/m3]
Particle
loading
[wt/wt]
Particle
conc.
[kg/m3]
Particle
recovery
[%]
Particle
size
cutoff
[m]
Author(s)
Device
type
Phases
Separation
mechanism
Abrahamson
et al. (2002)
Reverse
flow
cyclone
Gas &
solid
Centrifugal
3.55
20
39 X
2000
1000
0.120
99.984
100
Andreussi et
al. (2007)
Axial
flow
cyclone
Gas &
solid or
gas &
liquid
Centrifugal
4000
8000
< 10
97 100
Gas &
solid
Centrifugal
& inertial
0.11
7.3
320
70
1400
0 16
2.9 18
66 95
Gas &
solid
Centrifugal
18
0.15
98
Gas &
liquid
Centrifugal
0.110
0.135
0110
0.135
5.4
8.2
< 280
2 25
1000
7.7
9.3
Gas &
solid
Centrifugal
0.048
0.168
9.9
22.7
100
230
< 39
2600
0.0014
0.0047
0.0017
0.0056
90 96
Gas &
solid
Centrifugal
0.0508
13
23
140
240
76
4.7 8.2
5.6 9.8
99.99
Gas &
solid
Centrifugal
0.152
17
28
111
179
800
0.58
4.1
0.70
4.9
Gas &
solid
Centrifugal
0.029
0.154
7 27
25 98
36
700
750
0 20
0 24
98 99.5
Andreux et
al. (2007)
Barnes and
Van
Bylandtlaan
(1992)
Brunazzi et
al. (2003)
Chmielniak
and
Bryczkowski
(2001)
Dewitz et al.
(1988)
Dewitz
(1989)
Fassani and
Goldstein
(2000)
Fast
primary
separato
r
Secondary
separato
r
Axial
flow
cyclone
Reverse
flow
cyclone
Reverse
flow
cyclone
Reverse
flow
cyclone
Reverse
flow
cyclone
Author(s)
Gartside and
Woebcke
(1984)
Gartside and
Norton
(2002)
Gauthier et
al. (1990)
Gauthier et
al. (2005)
Hoffmann et
al. (1991)
Hoffmann et
al. (1992)
Knowlton
and
Bachovkin
(1978)
Klujszo et al.
(1999)
Jiao et al.
(2008)
Ld et al.
(1989)
Device
type
Fast
primary
separato
r
Fast
primary
separato
r
Axial
flow
cyclone
Fast
primary
separato
r
Reverse
flow
cyclone
Reverse
flow
cyclone
Reverse
flow
cyclone
Axial
flow
cyclone
Reverse
flow
cyclone
Reverse
flow
cyclone
Barrel
diam.
[m]
Inlet
gas
vel.
[m/s]
Gas
flowrate
[m3/h]
Particle
size
range or
mean
diam.
[m]
Particle
density
or bulk
density
[kg/m3]
Particle
loading
[wt/wt]
Particle
conc.
[kg/m3]
Particle
recovery
[%]
Particle
size
cutoff
[m]
Phases
Separation
mechanism
Inlet
hydraulic
diameter
[m]
Gas &
solid
Centrifugal
& inertial
0.0508
0.152
12
27
72 145
100 &
50
1100 &
720
8.3 19
10 22
95.0
98.1
Gas &
solid
Centrifugal
& inertial
90
3.2 4.8
95 99
Gas &
solid
Centrifugal
0.0169
0.0508
9 31
3.5 29
29
2500
1.0 6.0
1.2 7.2
> 95
Gas &
solid
Centrifugal
& inertial
0.123
5 20
210
860
> 98
Gas &
solid
Centrifugal
0.057
0.2
10
20
140
280
0.3 60
2640
0 0.033
0
0.040
80 96
Gas &
solid
Centrifugal
0.45
15
2640
0 0.11
0 0.13
74 97
0.4 1
Gas &
solid
Centrifugal
0.029
0.102
6.6
26
23 88
1 420
870
0.003
0.985
0.21
4.7
92.9
99.7
Gas &
solid
Centrifugal
0.0508
0.0508
4.2
7.4
31 54
1 100
2500
< 0.004
< 0.005
71.0
88.0
5 10
Gas &
solid
Centrifugal
0.300
10
20
2 20
2750
26
Gas &
solid
Centrifugal
0.005
0.008
0.005
0.08
0.38
32
0.1
108
Author(s)
Nielsen et al.
(1997)
Nishida and
Fujiyama
(2000)
Ogawa and
Ugai (2000)
Ogawa et al.
(1994)
Peng et al.
(2004)
Ross Jr. and
Schaub
(1998)
Vaughan
(1988)
Tan (2008)
Tuzla and
Chen (1992)
Zhang et al.
(2001)
Device
type
Fast
primary
separator
Fast
primary
separator
Reverse
flow
cyclone
Axial
flow
cyclone
Reverse
flow
cyclone
Fast
primary
separator
Axial
flow
cyclone
Axial
flow
cyclone
Reverse
flow
cyclone
Axial
flow
cyclone
Particle
density
or bulk
density
(see
refs.)
[kg/m3]
Particle
loading
[wt/wt]
Particle
conc.
[kg/m3]
Particle
recovery
[%]
Particle
size
cutoff
[m]
Phases
Separation
mechanism
Inlet
hydraulic
diameter
[m]
Barrel
diam.
[m]
Inlet
gas
vel.
[m/s]
Gas
flowrate
[m3/h]
Particle
size
range or
mean
diameter
[m]
Gas &
solid
Centrifugal
& inertial
0.0349
7.7
11.0
30 38
373
1410
~ 2600
6.4 24
7.7 29
99.064
99.997
Gas &
solid
Centrifugal
& inertial
0.04
0.08
0.10
5 30
23 140
63
850
4 87
5 104
92 99>
Gas &
solid
Centrifugal
0.031
0.104
5 20
14 54
18 &
18.3
2120 &
2310
0.0008
0.042
0.001
0.050
60 98
Gas &
solid
Centrifugal
0.033
0.099
3 25
9.2 77
11
~ 2100
0.0017
0.05
0.002
0.06
Gas &
solid
Centrifugal
0.110
0.110
9.4
11
320
370
3.65
2730
0.0013
0.0016
91.5
91.9
0.9
Gas &
solid
Centrifugal
& inertial
0.102
6.1
14
180
330
2.7 19
3.2 22
95 99
Gas &
liquid
Centrifugal
0.010
0.010
13
51
3.6 14
0.1 15
760
1.35
Gas &
solid
Centrifugal
0.0542
0.241
7.6
16.8
160
350
2650 &
2070
0.001
0.002
0.001
0.002
Gas &
solid
Centrifugal
0.165
0.406
3.6
390
300
2400
1.4 5.6
> 99.99
Gas &
solid
Centrifugal
0.084
0.084
>3
60 490
0.5 20
1.0 2.9
X 10-6
1.2 3.5
X 10-6
4.5
2
2.1
Much attention has been directed to fast gas-solid separation devices for their
applicability in FCC, biomass pyrolysis and other related processes with some
very innovative designs having been proposed. To date, FCC has been the main
field of application of gas-solid reactor termination devices. Significant
experience has been developed with these devices at industrial scale. For most
other applications, the developments have not led to large scale operation. These
separators often exist as the initial means of separation in advance of higher
efficiency traditional cyclone trains.
The major goal of the initial (or primary) separator is to achieve fast,
almost complete separation of product gases from catalytic or heat-bearing solids.
In primary separation, there is great interest in recovering most of the product
vapors. Achieving a high recovery of solid particulates is of more importance in
downstream separation. Another key requirement of primary separators is the
ability to operate properly over a wide range of conditions during unit start-up. At
start-up, gas and solid phase velocities are low, which may lead to entrainment or
carry-over of particulates to the downstream fractionator if the termination device
is poorly designed. This problem is very costly as it results in unit shut-down.
Steam stripping of the collected solids after primary separation is also an
important, subsequent step in the vapor recovery, but it is not the focus here.
Secondary separation is not discussed in detail in this review: its objective is to
remove fine particulate matter from the product gases to prevent their
accumulation in the bottom liquids of downstream distillation columns. Also,
many of the topics discussed in this and a later section on reverse flow cyclones
are applicable to secondary separation.
Most of the primary separators presented in this section employ
centrifugal separation techniques but are not considered cyclones. Since short
contact times between the heat bearing solids and the hydrocarbon feedstock are
crucial in order to avoid product degradation, many of the separators employ
partial cyclone arrangements. This means that upon entering the device,
separation of the various components occurs along less than one full vortex turn,
as opposed to several turns in a traditional cyclonic configuration. Fig. 2.1
illustrates the difference between a traditional reverse flow cyclone (Fig. 2.1a)
and a typical primary separator layout (Fig. 2.1b).
Figure 2.1 - General Flow Patterns in Two Types of Centrifugal Separators, modified from Peng et
al. (2004)
Figure 2.2 - Comparison of Riser Termination Methods, modified from Gauthier et al. (2005)
10
example in Fig 2.3. In either case, spent catalyst or heat-bearing solids are
captured for regeneration or reuse, while the product gases continue past the
separator for further processing.
Figure 2.3 - Comparison of Two FCC Process Configurations: a. Riser Reactor (modified from
Van Den Akker et al., 1990); b. Downer Reactor (modified from Gartside and Woebcke, 1981)
2.2
2.2.1
Downer Separators
Stone & Webster U-Turn Inertial Separator
In the late 1970s and early 1980s the Stone & Webster Engineering Corp.
(SWEC) developed a novel U-turn gas-solid separator for their thermal
regenerative cracking downer reactor process. Several patents relating to the
device were awarded to the company, and numerous later patents cite SWECs
initial invention (Gartside and Woebcke, 1981; Gartside and Woebcke, 1984;
Gartside and Woebcke, 1985; Forde and Stangeland, 1995; Gartside and Norton,
2002). Fig. 2.4 illustrates one conception of Gartside and Woebckes original
design. In this arrangement, the separator is located at the bottom of the downer
reactor pipe, and a hot gas-solid mixture enters the separator from above, as
shown in Fig. 2.4.
Figure 2.4 - Stone & Webster Eng. Corp. U-Turn Inertial Separator, modified from Gartside and
Woebcke (1984)
Solid particles disengage from the gas phase by inertia upon entering the
separator and immediately collide with a solids collection bed at the bottom of the
separator. Gauthier (1991) found that in a horizontal cyclone with a downward,
tangential inlet, the entering solids cannot complete a full rotation and become
trapped along the bottom side of the cyclone. This phenomenon was exploited for
a separator design by Dewitz et al. (1988), as discussed in the following subsection. In the SWEC design, accumulated solids are constantly dislodged from
the bed and transported over the weir and out of the separator horizontally at the
end opposite of the entrance. Thus the solids make a 90 turn within the device.
Meanwhile the product gases reverse direction and make a 180 turn to exit the
separator: this provides centrifugal separation that acts on the particles that have
not been collected by inertia.
Gartside and Woebcke tested numerous realized conceptions of the
separator shown in Fig. 2.4 with varying success, depending on the particular
arrangement (1984). Table 2.1 compares the geometry (labeled in Fig. 2.5),
operating conditions, and performance of the smallest and largest tested
configurations. 98% efficiency was achieved in a large scale test by adjusting the
pressure differential between the gas and solids outlets, thereby changing the gas
underflow (i.e. the fraction of the total gas flow escaping with the solids).
Typically about 10% of the product gases were lost with the collected solids.
Also, the gas residence time was usually observed to be of the order of 10ms.
Forde and Stangeland (1995) proposed the use of baffle plates positioned above
the catalyst collection bed to inhibit re-entrainment of particles into the cleaned
11
12
gas flow, as illustrated in Fig. 2.6. Note that the weir present in Fig. 2.4 is not
shown in the more recent design. In fact Forde and Stangeland specified that the
weir was optional and not essential to proper fast separation.
Figure 2.5 Important Dimensions in the Stone & Webster U-Turn Inertial Separator, modified
from Gartside and Woebcke (1984)
Table 2.1 - U-Turn Separator Testing Conditions and Performance, after Gartside and Woebcke
(1984)
Small Scale
Large Scale
Test
Test
Geometry
CL
8.9
28.0
Dgi
5.08
15.2
Dgo
2.54
10.2
L
25.4
0
H
19.1
30.5
Hw
0
5.72
Dso
5.08
15.2
W
5.08
15.2
(all dim. in [cm])
Operating
Conditions
Air & FCC
Air & Silica
Material
Catalyst
Alumina
Gas Inlet Vel.
20 m/s
12 27 m/s
Solid Mass Flow
0.80 kg/s
5.0 kg/s
Solid Load
17 kg/kg
22 kg/kg
Solids
19 kg/m3
22 kg/m3
Concentration
Performance
Collection Eff.
95.0%
98.1%
Gas Res. Time
~ 0.02s
Figure 2.6 - Baffle Plates in a Downer Separator to Inhibit Entrained Solids, modified from Forde
and Stangeland (1995)
13
14
Figure 2.7 - Double U-Turn Inertial Separator, modified from Gartside and Norton (2002)
2.2.2
15
16
Table 2.2 - Geometry and Operating Conditions in Slotted Cyclone Testing, from Dewitz et al.
(1988)
Geometry
H
W
DC
Ws
Operating
Conditions
Material
Gas Inlet
Velocity
Solid Mass
Flow
Solid Load
2.2.3
76.2 mm
38.1 mm
152 mm
25.4 mm
53
Air & FCC
Catalyst
24 m/s
0.38 kg/s
4.5 kg/kg
Another interesting separation device for use in a downer reactor was developed
and patented in conjunction by the Petroleum Energy Center and Nippon
Mitsubishi Oil Corp., both of Japan (Nishida and Fujiyama, 2000). Fig. 2.10
illustrates one conception of the separator design. The bottom of the downer pipe
is capped off, forcing the mixture radially outward. Guide vanes impart a
tangential component to the flow velocity. These vanes are less susceptible to
erosion than swirl vanes used in some reverse flow cyclones since the incoming
particles do not impinge directly on them. The solids drift to the wall of an
enclosing body and settle by gravity. The gases travel upward toward a gas outlet.
Particle separation efficiency was reported to be greater than 92% in all
configurations and under all operating conditions, and was usually better than
99%. The inventors tested five devices, all with an inner cylinder diameter of
40mm and with varying solid loads from 4 to 89 kg/kg. Unlike most separation
devices, the collection performance tended to decrease with increasing gas
velocity. Nishida and Fujiyama recommended that the gas velocity in the inner
cylindrical body be within the range 3 15 m/s for optimal separation.
Unfortunately no particular information is provided regarding separation time
within this device.
Figure 2.9 - Downer Separator with Internal Swirl Baffle, modified from Nishida and Fujiyama
(2000)
2.3
2.3.1
Riser Separators
Shell Oil Company Inverted U-turn Separator
Shell Oil Company developed a fast gas-solid riser separator in the late 1980s
(Van Den Akker et al., 1990) that was cited numerous times in later patent
applications on fast separation techniques (Ross Jr. et al., 1993; Letzsch and Earl,
1997; Ross Jr. and Schaub, 1998). The separator is illustrated in Fig. 2.11, and
forms part of the riser assembly shown in Figure 2.3a. It employs a double
chamber U-turn configuration. The patent document suggests that as many as four
separation chambers could be mounted atop the end of the riser reactor column,
but only two chambers are used in the preferred embodiment. Upon entering the
separator, the flow mixture branches into the two chambers. The solids complete a
180 half-turn within the device before settling out of the separator. The product
gases exit the device via two outlet pipes mounted concentrically to the curved
separator walls. Thus the gases make a 90 turn relative to the entrance to exit the
device. No information is given in the patent regarding separator performance.
Van Den Akker et al. (1990) recommended that the gas outlets be connected to a
secondary separator according to a related patent claim awarded to Barnes and
Van Bylandtlaan (1992). This secondary separator is globe-shaped, as shown in
Fig. 2.12. The globe separator achieved 98% separation efficiency of cracking
17
18
catalyst particles at a gas inlet velocity of 18m/s and solid load of 0.15 kg/kg in a
hot flow test.
SWEC modified the dual chamber U-turn arrangement of Van Den Akker
et al. by the insertion of horizontally disposed slots on the two gas outlet pipes, as
shown in Fig. 2.13. These gas openings were intended to decrease the average
kinetic residence time of the cracked hydrocarbon vapors. Ross et al. (1993) and
Ross and Schaub (1998) positioned the slots closer to the spent catalyst outlets as
illustrated in Fig. 2.12a. Letzsch and Earls (1997) preferred position for the slots
was closer to the separator inlet as shown in Fig. 2.12b. In both cases, the
preferred ranges for angles and were 0 30 and 30 90, respectively. Ross
et al. (1993) reported catalyst-gas separation efficiencies in the range of 95 99%
and separator residence times of 0.1 0.2 seconds.
Figure 2.10 - Double Chamber U-Turn Separator, modified from Van Den Akker et al. (1990)
Figure 2.11 - Shell Oil Company Globe-Shaped Secondary Separator, modified from Barnes and
Van Bylandtlaan (1992)
Figure 2.12 - Dual Chamber U-Turn Separators with Gas Slots: a. gas opening facing solids outlet;
b. gas opening facing riser tube, modified from Ross and Schaub (1993)
2.3.2
The first industrial operation of rapid gas-solid termination in FCC is probably the
closed cyclones concept developed by the Mobil Research and Development
Corporation (Avidan et al., 1990). In this arrangement the riser is connected
19
20
Figure 2.13 - Tracer Residence Time Distributions of Two Separation System Types, after Avidan
et al. (1990)
2.3.3
21
22
Figure 2.14 - Flow Schematic of Amoco Quench Separator, modified from Forgac et al. (1991)
2.3.4
Another fast internal riser separator was developed by UOP LLC (Niewiedzial,
1996). The separator uses arms disposed both radially and tangentially though
which the gas-solids mixture flows to induce a swirling flow of the incoming gassolid mixture, as shown in Fig. 2.16. This feature is similar to the downer
separator of Nishida and Fujiyama (2000). The end of the riser is capped off just
above the radial arms, forcing the mixture into the arms. Centrifugal forces drive
the solids to the wall of an enclosing body, where they settle by gravity.
Meanwhile the gases disengage and enter a small gap between the riser pipe and a
baffle positioned around the riser. No data is provided in the patent document as
to the separator performance.
Figure 2.15 - Riser Separator with Swirl Tubes and Baffle, modified from Niewiedzial (1996)
23
24
Figure 2.16 - UOP LLC Riser Termination Devices: a. VSS; b. VDS, after UOP LLC
(2003)
2.3.5
Figure 2.17 - Single Loop U-Turn Separator with Gas-Accelerating Venturi Ramp, modified from
Nielsen et al. (1997)
25
26
2.3.6
Gauthier et al. (2005) evaluated the performance of a recent riser separator and
stripper design termed the RS2. The overall system, as illustrated in Fig. 2.19a,
is globe-shaped and contains two separation and stripping chambers, respectively.
A gas-solid mixture enters the separation chambers from the riser. The mixture is
separated centrifugally and the catalyst particles are collected in the two diplegs
arranged concentrically about the riser. The cracked gases pass from the
separation chambers into the stripping chambers through the windows connecting
the adjacent sections. Product gases mix with stripping gases in the stripping
chambers and flow out through the gas outlet collector. Gauthier et al. also
stressed the indirect, but important effect of the separator pressure balance on
solids collection efficiency. Solids collection is influenced by the gas underflow,
which in turn is determined from the separator pressure balance. This pressure
balance depends in part on the pressure drop between the separator inlet and gas
outlet. For several reasons, Gauthier et al. insisted that the inlet-to-outlet pressure
drop must be kept to a minimum. Most importantly, the overall pressure balance
(of which the separator pressure drop forms a part) has a strong influence on gas
flow repartition. Gas repartition is the distribution of gas flow to each outlet. This
repartition affects the overall gas residence time distribution, which affects
product formation. In addition, a lower separator pressure drop allows for
flexibility in the dipleg immersion, which in part allows for flexible overall unit
operation. Another important feature of the RS2 design is the arrangement of the
diplegs around the riser. Since the cross-section of the diplegs is quite large, the
downward catalyst particle velocity is low. This configuration allowed for upward
bubbling flow of stripping gas, which promoted good stripping.
Figure 2.18 - RS2 System with Integrated Separation and Stripping, modified from Gauthier et al.
(2005)
2.3.7
27
28
Figure 2.19 - Fast Gas-Solid Single Half-Turn Separator used in the study by Andreux et al.
(2007)
2.4
Scale-Up Issues
29
30
Figure 2.20 - CFD Tracer Concentration Study within RS2, after Gauthier et al. (2005)
One drawback to some current CFD commercial packages that they are
less reliable for dense particle loadings in cyclone and fast separator simulations.
Figure 2.22 shows how the dilute and dense flow regimes are generally classified.
In general, the dilute flow regime has a practical upper limit of 0.1% solids by
volume in CFD simulations (Corts and Gil, 2007). In the dilute flow regime,
particles are usually tracked discretely using time-of-flight Lagrangian models.
For very dilute solids loading, the transport equations are one-way coupled,
meaning that the solid phase flow is solved using the results of a single-phase gas
flow calculation. Two-way coupling refers to simultaneous gas and solid phase
calculations such that the flow patterns of each phase are mutually affected, but
still neglecting particle-particle interactions. In the dense flow regime, both the
gas and solid phases are treated as continuous, inter-penetrable media. The flow
of both phases is tracked using Eulerian models. The mass and momentum
transport equations for each phase in the dense flow regime are four-way coupled
to account for mutual inter-phase and intra-phase interactions such as particleparticle collisions. Four-way coupling refers to the simultaneous solution of the
gas and solid phases, as with two-way coupling, as well as the solution of mutual
interactions between particles.
Current modeling techniques are not fully accurate for dense solids
loading conditions (0.1% solids by volume). This is due to a lack of accurate
representation of particle-particle and particle-wall interactions as well as gasparticle momentum effects. Hence CFD has typically been used to predict scaleup effects under dilute loading conditions. Numerical simulations of highly
loaded separators should, therefore, be accompanied by experiments to increase
confidence in CFD results.
The pressure balance existing within a laboratory scale separation device
is an extremely important consideration in separator scale-up. The pressure at the
inlet and at both gas and solids outlets strongly affects flow patterns and gas flow
repartition within the separator. Consequently the collection efficiency of both the
gas and solid phases are affected by this pressure balance. Hoffmann and Stein
(2007) compare several pressure drop models for reverse flow devices, which are
reviewed briefly in a later section. Most of these models (as most separator
models) are defined from inlet to gas outlet, without regard to the dipleg pressure.
This may be because it is common in laboratory scale tests to block gas underflow
by collecting solid particles in a closed hopper hence the pressure at the solids
outlet is of little interest. It is also important to note that the pressure balance
around the separator is different between positive and negative pressure
separators. Suction or negative pressure separators are often used in smaller scale
experiments with closed solids collection hoppers. Therefore no gas underflow
occurs in negative pressure devices. In FCC applications, in most cases, positive
pressure separators are used. In these separators, there is always some gas
underflow. The effect of gas underflow is further discussed in a later sub-section.
Several numerical simulations have also been performed to predict the
pressure drop in cyclone separators. Griffiths and Boysan (1996) obtained
31
32
33
34
Figure 2.23 - Comparison of Unloaded Gas Residence Time Distributions in a Cyclone Reactor at:
a. Laboratory Scale (2.8cm); and b. Industrial Scale (30cm), modified from Ld et al. (1989)
2.5
which eventually leads to a balance between the accumulation and loss of fine
particles. Reppenhagen and Werther (2000) and Klett et al. (2007) found that the
time-dependent rate of attrition in a cyclone eventually reached a steady-state
value as shown in Fig. 2.25. The rate of attrition was defined in both studies as the
ratio of the attrited particle mass flow to the total solids mass flow into the
cyclone. They observed a high initial attrition rate, which decreased rapidly with
subsequent passes through the cyclone, and which finally settled to a steady-state
value. Reppenhagen and Werther (2000) also showed that the solid particle
attrition rate in reverse flow cyclones was proportional to the square of the gas
inlet velocity. However, the proportion of attrited particles was also inversely
proportional to the square root of the solids loading. This was demonstrated
clearly from their results shown in Fig. 2.26. An implication for cyclone design
and scale-up with regard to attrition is the importance of the inlet geometry since
the inlet dimensions determine the gas inlet velocity. Reppenhagen and Werther
(2001) predicted that the catalyst loss rate in an industrial cyclone would actually
decrease as cyclone size increased, even though the separation performance of the
cyclone was worse as the size increased. This was due to a lower fines production
rate by attrition, which in turn was caused by a lower gas inlet velocity as the inlet
area increased, assuming a constant gas flowrate.
35
36
Figure 2.25 Effect of Gas Inlet Velocity on Steady-State Particle Attrition Rate in Cyclones,
modified from Reppenhagen and Werther (2000)
2.6
Inter-Particle Forces
Inter-particle forces, such as electrostatic and Van der Waals forces, are assumed
to not be present in some standard cyclone scaling models (Hoffmann and Stein,
2002). However, in some cases cyclone performance is greatly affected by interparticle forces. Electrostatic effects are especially problematic in cold flow tests,
as supported by Gauthier et al. (1990) and Gauthier (1991). Gauthier (1991) found
that inter-particle forces in a uniflow cyclone circulation system were modified by
changing the relative humidity of air over the range 20 80%. Figure 2.27a shows
particle charge measurements plotted versus relative humidity. Electrostatic
particle charges became negligible above 70% relative humidity, at which the
highest collection efficiency was observed. Unfortunately, at high relative
humidity, particle agglomeration due to surface condensation was significant.
37
38
When electrostatic charges were strong, on the other hand, decreased collection
efficiency was observed. Figure 2.27b shows collection efficiency measurements
plotted against air relative humidity. Gauthier used small traces of ammonia (30
60 ppm) to reduce electrostatic charging while avoiding particle surface
condensation.
Studies on particle entrainment from fluidized beds have demonstrated the
crucial role of particle-particle agglomeration (Bnoni et al., 1994). This
agglomeration has also been observed at high temperatures, under industrial
conditions. It is likely that the agglomeration of fine particles with larger particles
also has a great impact on separator efficiency. Since such agglomeration is
affected by the surface properties of the particles, and the adsorption of gaseous
impurities on the particle surface, it is nearly impossible to predict the effect of
agglomeration on cyclone performance. This severely limits the reliability of any
cyclone model.
Figure 2.26 - Air Relative Humidity Effects in a Uniflow Cyclone: a. Electrostatic Particle
Charge; b. Solid Collection Efficiency, after Gauthier (1991)
2.8
Predicting the gas residence time in fast separators is very important in catalyst
cracking processes, since degradation of the product gases will occur for extended
separation times. Since many of the fast separators discussed above often achieve
separation of the various components in less than 0.1s at laboratory scale,
measuring the time spent in the device by the gases is a difficult endeavor. Ld et
al. (1989) used helium and hydrogen tracer techniques to measure the gas
residence time distributions (RTD) in four different reverse flow cyclone reactors.
Gauthier (1991) used two constant temperature hot wire anemometers and helium
as a tracer gas to measure the gas residence time distribution in a uniflow cyclone.
Two anemometers were initially positioned upstream of the cyclone, one inserted
in the helium injection line and the other at the cyclone inlet. After measuring
tracer signals between the helium injection line and the cyclone inlet, the inlet
probe was moved to the cyclone gas outlet tube, where it was used to measure the
outlet gas signal. Van der Lans et al. (1997) measured the residence time
distribution of a cold swirling air flow in a laboratory scale coal burner also using
helium as a tracer gas and a hot wire constant temperature anemometer. The
measured residence time distribution was reproducible within 20% under all
tested conditions. Zhongxi et al. (2005) measured the gas residence time
distribution in a rough-cut cyclone using a tracer but did not indicate either the
type of tracer or the probe type used. Recently Stief et al. (2008) also used helium
tracer injections and two thermal conductivity detectors (TCDs) to measure the
residence time distribution of gas flow through a microreactor system. The TCDs
had a response time of 4ms in helium, and as such would likely be useful for the
measurement of gas residence times in fast separators.
3
3.1
The current review now shifts to co-current flow, axial flow or uniflow
centrifugal separation. Axial flow cyclones (AFCs) are considered here as an
alternate class to reverse flow cyclones. In reverse flow separation, the cleaned
gases make a 180 turn relative to the particulates to exit the device. However, in
axial flow cyclones, the gases do not reverse flow direction and are withdrawn
near the natural end of the outer vortex. Hence, co-current flow cyclones share
an important conceptual feature with many of the fast separators discussed in the
previous chapter. The term axial flow should not be confused with axial
entry. Axial flow refers to the mechanism of gas withdrawal, whereas axial entry
refers to the gas entry direction relative to the cyclone longitudinal axis. Fig. 3.1
shows a hierarchy of the various types of cyclonic separators reviewed in the
present work. AFCs are used industrially for gas dedusting and demisting, in
aerosol science for selective gas demisting, and in liquid-liquid hydrocyclones.
The first published work on these devices is attributed to Umney (1948). There
exist surprisingly few published results in the literature for AFCs compared to
reverse flow cyclones.
39
40
3.2
Gas dedusting is the primary application of axial flow cyclones. With regard to
general performance, Jackson (1963) suggested that axial flow dedusting cyclones
have greater capacities than similar reverse flow separators. Later, Novikov et al.
(1980) compared reverse flow and axial flow dedusters with identical cyclone
diameters. At a constant gas flowrate, the axial flow deduster had better solids
collection efficiency as well as a lower pressure drop.
3.2.1
Axial flow dedusters can be further subdivided into two groups based on the inlet
type: tangential entry versus axial entry, as shown in Fig. 3.1 above. Sumner et al.
(1987) and Gauthier et al. (1990) each evaluated the performance of an identical
tangential entry uniflow cyclone, as illustrated in Fig. 3.2. Their application was a
rapid biomass and heavy oil pyrolysis process. The experimental cyclone
geometry, operating conditions and performance for three uniflow, tangential
entry cyclone studies (Gauthier et al., 1990; Ogawa et al., 1994; Tan, 2008) are
given in Table 3.1.
Figure 3.2 Tangential Entry Uniflow Cyclone, modified from Gauthier et al. (1990)
41
42
Table 3.1 - Cyclone Geometry, Operating Conditions and Performance for Various Uniflow,
Tangential Entry Deduster Studies
Cyclone
Geometry
Cyclone Diameter,
DC
Cyclone Inlet
Height
Cyclone Inlet
Width
Cyclone Inlet
Diameter
Separation Length,
LS
Gas Outlet
Diameter
Operating
Conditions
Particle Mean
Diameter, dP
Gas Inlet Velocity,
Vi
Solids Mass
Flowrate, mS
Solids Loading,
wt.solids / wt.gas
Solids
Concentration
Cyclone
Performance
Solids Collection
Efficiency
Particle Size CutOff, dp50
Gas Residence
Time
Gauthier et al.
(1990)
Ogawa et al.
(1994)
Tan
(2008)
50.8 mm
99 mm
241 mm
25.4 mm
184.1 mm
12.7 mm
31.8 mm
33 mm
50.8 533 mm
101 953 mm
165 mm
22 mm
33 mm
29 m
11 m
9 31 m/s
3 25m/s
5 130 g/s
16
0.0017 0.05
0.001 0.002
1200 7200
g/m3
2.0 60 g/m3
1.06 2.02
g/m3
4 m
8 m
15ms 60ms
usually greater than 99.5%, depending on the particular geometry and operating
conditions. They found that the collection efficiency responded strongly to
changes in the separation length. Separation length to cyclone diameter ratios of
LS/DC = 2 to 3 resulted in the maximum observed collection efficiencies (>
99.95%) over a wide range of gas inlet velocities. However, the solids collection
efficiency was greatly hindered at short separation lengths by flow interference at
the inlet. This problem was overcome by the placement of a single-turn helical
roof (Fig. 3.3). The solids collection efficiency was greater than 99.98% for most
experiments using this roof. Gauthier et al. also suggested that there was an
important relation between solids collection efficiency, gas flow patterns near the
gas outlet, and the system backpressure downstream of the cyclone. The study
showed no clear trend between solids collection efficiency and solids loading. The
estimated separation times were 15 to 60 ms, depending on the gas inlet velocity.
Ogawa et al. (1994) created an analytical model for the solids collection
efficiency in a tangential entry AFC and verified the model experimentally. The
model accounted for turbulent effects including particle sedimentation velocity
and re-entrainment into the gas flow near the gas outlet. The experimental results
verify the models prediction of maximum collection efficiency for a separation
length to cyclone diameter ratio LS/DC = 2.2 which agreed well to the
experimental results of Gauthier et al. (1990). Ogawa et al. plotted grade
efficiency curves and found that the minimum particle cut size was 4 m. The
experimental geometry used by Ogawa is provided in Table 3.1.
Tan (2008) performed a set of experiments with two types of dust in a
240mm diameter cyclone to verify an analytical fractional collection efficiency
model. The device had a tangential entry and was arranged horizontally, as shown
in Fig. 3.4. Tan showed that the fractional collection efficiency was related
directly to the Stokes number and found that the efficiency was greater than
99.5% when the Stokes number was larger than 1. An efficiency of 50% was
43
44
Figure 3.4 - Horizontal, Tangential Entry Axial Flow Deduster, modified from Tan (2008)
3.2.2
Horizontal, axial flow dedusters were initially proposed by Umney (1948) and by
Daniels (1957). These devices usually have axial entry inlets where the flow
vortex is induced by a set of vanes mounted on a central body. They tend to be
more compact than tangential entry cyclones, but there is a risk of vane erosion
from impinging particles. Recently Klujszo et al. (1999) and Zhang et al. (2001)
independently developed horizontal flow swirlers for industrial gas dedusting.
Klujszos apparatus is illustrated in Fig. 3.5, while the cyclone used by Zhang et
al. is shown in Fig. 3.6. The two devices are very similar but differ in scale and by
the shape of the swirl vanes. Zhangs swirler also had a central core running the
length of the separator.
Figure 3.5 - Horizontal Axial Flow Swirler, modified from Klujszo et al. (1999)
45
Cyclone Geometry
Cyclone Diameter, DC
Vane Angle to Horizontal
Number of Vanes
Separation Length, LS
Gas Outlet Diameter
Operating Conditions
Particle Size Distribution
Gas Inlet Velocity, Vi
Solids Loading
Cyclone Performance
Dust Recovery
Particle Size Cut-Off, dp50
Cyclone Pressure Drop, PC
1 100 m
4.2 7.4 m/s
< 5000 mg/m3 air
0.5 20 m
71.0% - 88.0%
5 10 m
5 m
100 Pa
60
8
1070 mm
Zhang et al. used a swirl vane configuration in their study and tested the
performance of a horizontal deduster with two different dust types, as shown in
Fig. 3.6. They were interested in the economic feasibility of the deduster as an
industrial and agricultural air cleaner compared to other existing separation
techniques. The study showed that the cyclone particle size cut-off was about 5
m. The cyclone had a capacity of 188 L/s and a low pressure drop of 100 Pa.
Zhang plotted grade efficiency curves for both dusts and achieved greater than
90% recovery for particle diameters larger than 10 m. Their results are
summarized in Table 3.2.
46
Figure 3.6 - Horizontal Uniflow Deduster modified from Zhang et al. (2001)
A very recent axial entry, axial flow cyclone was developed by Andreussi
et al. (2007) for TEA Sisterni S.p.A. The device was designed to separate either
solid or liquid particulates from a gas stream. The vane angle (measured from the
cyclones longitudinal axis) can be changed within the device without replacing
the vanes, as illustrated in Fig. 3.7. This allows for varying tangential flow
velocities and cyclone pressure drops. The separator is arranged vertically such
that the flow mixture enters the device from below and passes vertically upward
through the separator. The patent document suggests that the separators particle
size cut-off is 5 m, and that the separation efficiency is 100% for particle
diameters greater than 5 m when operating at a gas flowrate of 8000 m3/h. The
efficiency drops to 97% when the gas flowrate decreases to 4000 m3/h.
Figure 3.7 - Axial Entry, Axial Flow Cyclone modified from Andreussi et al. (2007)
3.3
47
48
Figure 3.8 - Axial Entry, Uniflow Demister modified from Vaughan (1988)
49
Table 3.3 - Comparison of Cyclone Geometry, Operating Conditions and Performance in Three
Demister Studies
Cyclone Geometry
Vaughan (1988)
Cyclone Diameter, DC
10 mm
76
Helix Pitch, P
5 mm
Separation Length, LS
12.5 22.5 mm
Brunazzi et al.
(2003)
(see Fig. 2.8)
110 mm (A)
135 mm (B, C)
30 (A)
70 (B, C)
1.8 (A)
6 (B, C)
98 mm (B, C)
500 mm (A)
670 mm (B, C)
54 mm (A)
210mm (B, C)
0.5 15 m
13 51
1 4 L/min
5 15 m/s
1.35 m
3 m
50
time within the device, as well as the possibility for particle deposition on the
helix blades. Cyclone configuration A performed worse than the other two
arrangements in the study for the collection of droplets smaller than 15 m but
the performance was more successfully characterized by the analytical model.
The experimental geometry, operating conditions and performance are shown in
Table 3.3.
Figure 3.9 - Three Demister Arrangements used in the study by Brunazzi et al. (2003)
3.4
Hydrocyclones
51
52
4
4.1
Reverse flow cyclones (RFCs) have garnered the most attention both in the
literature and in industrial usage in the field of centrifugal separation. They are
well known in the art for their usually high efficiencies, typically on the order of
90% or greater, depending on the application. RFCs are used to clean gases laden
with particulate dust or liquid droplets. A standard RFC arrangement is illustrated
in Fig. 4.1a. The characteristic feature of a RFC is the method by which the gases
disengage from the dense phase(s) and reverse direction by 180 relative to said
dense phase(s) before flowing out of the cyclone. RFCs are broadly categorized
into two groups depending on the type of separator inlet they employ: tangential
entry versus swirl tube (axial entry) (Fig. 4.1b). The current chapter provides a
very brief sampling of the reverse flow cyclone literature. Hoffmann and Stein
(2007) provide a much more detailed review of these devices.
Figure 4.1 - Comparison of Two Reverse Flow Cyclone Types: a. Tangential Entry RFC; b. Swirl
Tube (after Peng et al. 2004)
4.2
The two main performance characteristics of RFCs are the particle collection
efficiency and the separator pressure drop. Accurate prediction of the pressure
drop is important in any type of separator because:
The particle collection efficiency and pressure balance around the
separator depend in part on the cyclone pressure drop;
In many fluidized bed reactors, proper design and operation of
standpipes or diplegs depend on the separator pressure balance;
and
The energy requirement of the cyclone must often be known.
Many models have been proposed to predict the pressure drop for various
RFC designs. To date, none of these models has been adopted universally, and
they should be used with caution. There is also great variety in sophistication
among the pressure drop models. The simplest models (Casal and BenetMartinez, 1983; Coker, 1993; Shepherd and Lapple, 1939) only consider the
effect of cyclone geometry more specifically, the ratio of the inlet area to the
outlet area. These models only consider single-phase gas flow for the sake of
simplicity, although most authors acknowledge a significant decrease in the
pressure drop when solid particles are introduced. Gimbun et al. (2005) showed
that the models from Casal and Benet-Martinez and from Coker produced
significant underestimates over a wide range of values of gas inlet velocity and
operating temperature.
Dirgo (1988) performed a statistical analysis of experimental data
collected from 98 different cyclone designs, and from this created an empirical
model based on various geometric parameters. The model does not account for
particle loading effects since much of the data was taken from sources where the
solids loading conditions were not specified. However, due to its simplicity and
decent accuracy for single-phase gas flow, Dirgos pressure drop model remains
widely used. Zenz (1982) proposed a more sophisticated model that divides the
total pressure drop into additive components: a gas contraction loss just upstream
of the cyclone inlet, a frictional pressure drop of the gas-solid strand along the
barrel, a loss associated to the gas direction reversal at the end of the outer vortex,
and an exit contraction loss. This model does not fully account for the presence of
particles and thus only gives a rough estimate of the total pressure drop.
Bohnet and Lorenz (1993), among many others, found that the operating
temperature of a cyclone has a profound effect on its collection efficiency and
pressure drop. As the temperature of the incoming gas increases, the pressure drop
decreases. They improved on a model proposed by Meissner (1978) which
attempted to quantify the effect of operating temperature and particle loading on
the pressure drop. The model successfully predicted the pressure drop within the
experimental measurement error over the range of temperature 293 973K for the
conditions tested. However, it could be argued that any model that accurately
accounted for the effect of gas viscosity and density would make an additional
term for the effect of temperature unnecessary.
Likely the most sophisticated pressure drop model presently available in
the literature was developed recently by Chen and Shi (2007). The model
accounts for either single-phase or dust-laden gas flow and for either low or high
operating temperature. The authors claim their model to be the most accurate. In
this model the total pressure drop is composed of additive components, as was the
earlier model from Zenz (1982): an expansion loss at the inlet, a frictional loss
from the swirling flow along several surfaces, a contraction loss at the outlet, and
53
54
a dissipative loss within the gas outlet pipe. The contraction loss was negligible
(i.e. < 1% of the total pressure drop), while the most important contributions were
the frictional loss and from the flow in the outlet pipe. The solids loading was
assumed to affect only the cyclone barrel friction factor and the tangential
velocity along the barrel. The models from Zenz and from Chen and Shi are
advantageous over other models because they break the total pressure drop into
components. This allows the pressure drop between the inlet and the solids outlet
to be calculated, which is an important consideration for the design and
implementation of cyclones in continuous systems, especially in terms of the
cyclone pressure balance and dipleg vertical position.
The accuracy of cyclone pressure drop models varies greatly. Figure 4.2
compares predicted cyclone pressure drop values from 11 different models to
experimental data for single-phase gas flow at room conditions for various gas
inlet velocities and cyclone geometries. Many of the results shown here are
presented in Chen and Shi, with several pressure drop models added. The
experimental data were obtained for single-phase air flow at room conditions. The
results show that Chen and Shis model was the most accurate for the particular
set of cyclone operating conditions considered here.
55
104
103
102
102
103
104
Figure 4.2 Comparison of Various Reverse Flow Cyclone Pressure Drop Models,
modified from Chen and Shi, 2007
56
4.3
Due to its relevance in FCC processes, many studies have investigated the effect
of solids loading on solid recovery and separator pressure drop in RFCs (Derksen
et al., 2006, Fassani and Goldstein, 2000, Hoffmann et al., 1991, Hoffmann et al.,
1992; Knowlton and Bachovkin, 1978; Tuzla and Chen, 1992). Muschelknautz
(1970; 1980) proposed that the gas stream entering a cyclone could transport only
a certain particle concentration. Above this maximum saturated concentration, the
extra particle load was not suspended in the gas and entered the cyclone as a
dense stream, which was immediately collected along the wall. The result was
increased efficiency as the solids loading increased. Mothes and Lffler (1985)
also observed increasing efficiency as the solids loading increased, but their
results did not agree with Muschelknautzs hypothesis. They proposed instead the
concept of particle sweeping, where fine particles that would otherwise escape
were swept to the wall by larger particles and were collected. However, these
results apply only to very low solids loading (< 0.001 wt./wt.)
Hoffmann et al. (1991) plotted solids collection efficiency and separator
pressure drop curves as a function of the solids loading at intermediate solids
loading. The efficiency increased dramatically with increasing particle loading
usually from around 80 85% at very dilute loadings to 90 95% at about 0.04
wt/wt. In addition, the cyclone pressure drop always decreased as solids loading
increased, with the greatest improvement observed at the highest gas inlet
velocity. The relevant dimensions of a typical RFC are shown in Fig. 4.2. Table
4.1 lists the dimensions of the cyclone used by Hoffmann et al., as well as the
operating conditions and cyclone performance.
57
Table 4.1 - Summary of Experimental Geometry, Operating Conditions and Performance in the
studies by Hoffmann et al. (1991) and Fassani and Goldstein (2000)
Cyclone
Geometry
DC
a
b
LS
Dgo
Dso
H
All dims. in
[mm]
Operating
Conditions
Particle Size
Distribution
Mean Particle
Size, dP
Gas Inlet
Velocity, Vi
Solids loading,
kg/kg
Cyclone
Performance
Solids
Collection
Efficiency
Cyclone
Pressure Drop,
PC
Fassani and
Goldstein
(2000)
154
48
21
52
35
76
550
200
100
40
100
75 100
75
800
Knowlton and
Bachovkin
(1978)
50.8 102
46
21
64
51
25
406
Tuzla and
Chen
(1992)
406
248
124
300
251
129
1347
1 200 m
0.3 60 m
1 420 m
~ 50 450
m
36 m
3.5 m
14 m
300 m
7 27 m/s
10 20 m/s
6.6 26 m/s
3.6 6
0 20
0 0.04
0.003 0.985
1.4 5.6
98 99.5%
80 95%
92.9 99.7%
> 99.99%
0 1000 Pa
250 1800 Pa
500 83000 Pa
Hoffmann et
al. (1991)
58
4.5
Axial entry RFCs, referred to as swirl tubes in the literature, are an alternate subclass to standard tangential entry RFCs, as discussed earlier and illustrated in Fig.
3.1. Swirl tubes use a set of vanes mounted on the gas outlet pipe to induce a
swirling vortex in the incoming flow. Peng et al. (2004) tested a laboratory scale
swirl tube cyclone to separate chalk dust (median particle size of 3.65m) from
air, and plotted the grade efficiency curves against those from a traditional RFC of
similar dimensions. They found that the grade efficiency curve from a swirl tube
59
60
was steeper than that from a tangential entry RFC; in other words, the cut size was
sharper. Recall that Vaughan (1988) observed the same behavior in an axial entry
uniflow cyclone. The particle cut size in the study by Peng et al. was actually less
than 1 m. Generally the collection efficiency increased with the separation
length, to a maximum value of about 92% for a separation length to cyclone
diameter ratio LS / DC = 5. The minimum cyclone pressure drop also occurred for
LS / DC = 5. Peng et al. suggested that swirl tubes are more compact than
tangential entry RFCs operating at the same pressure drop and capacity. The
dimensions of the swirl tube studied by Peng et al. are shown in Fig. 4.4.
Figure 4.5 - Swirl Tube Dimensions used in the study by Peng et al. (2004)
4.6
drop decreased by about 36% from 1100 to 700 Pa. Jiao et al. (2008) used a
rotating blade impeller in the gas outlet of a reverse flow cyclone as a secondary
means of separation to block entrained talcum powder particles from escaping
with the exiting gas stream. They observed that the grade efficiency curves shifted
upward by about 25% as the impeller speed increased from zero to 3600rpm.
Figure 4.6 Reverse flow cyclone with rotating slotted baffle plates mounted in the gas outlet,
modified from Chmielniak and Bryczkowski (2001)
4.7
Vortex Breakers
61
62
significantly, by as much as 13%, with increasing gap size over the same range of
Lgap.
Redding (2005) used an unconstrained, freely moving rod or tube in a gasliquid hydrocyclone to select (i.e. separate) a lighter gas-liquid emulsion from
heavier emulsions, as shown in Figure 4.7b. The rod or tube was made to be
slightly denser than the liquid phase in the cyclone to prevent it from floating. In
operation, the rod was allowed to swirl in the cyclone barrel with the gas-liquid
flow. The rod interrupted the swirling gas-liquid outer vortex, which apparently
assisted the formation of the inner vortex composed of a light gas-liquid
emulsion. Redding also suggested that the unconstrained rod vortex breaker
allowed a huge cyclone scale-up for the production of desirable gas-liquid
emulsions, from a barrel diameter of DC = 19 mm to DC = 200 mm.
5
CONCLUSIONS
RECOMMENDATIONS
63
64
losses as attrition and re-entrainment in the cyclone cone section increase with gas
velocity, which in turn may lead to decreased separation efficiency. To lessen
cyclone erosion damage due to particle-wall collisions, the following methods are
recommended:
use of more regular (i.e. spherical) particulates, if the process
allows it;
steeper cone angle; and
use of special refractory materials to cover high-erosion surfaces.
CFD has emerged as a powerful tool to simulate gas-solid flow patterns in
separation devices and to assist in separator scale-up. However, numerical
simulations should be coupled with experimental studies whenever possible
especially in the case of novel fast separator designs. Particle-particle and
particle-wall interactions have not been modeled with sufficient accuracy to allow
CFD gas-solid flow simulations to be used with full confidence. These
simulations do, however, give good qualitative insight into general flow patterns
and are recommended to identify potential problems and to help improve
separator design.
REFERENCES
Abrahamson, J., Jones, R., Lau, A. and Reveley, S., Influence of entry duct
bends on the performance on return-flow cyclone dust collectors, Powder
Technology, Vol. 123, 126 137 (2002).
Alexander, R. McK, Fundamentals of cyclone design and operation,
Proceedings of the Australian Institute of Mining Metals, Nos. 152 153,
203 228 (1949).
Andreussi, P., Ciandri, P. and Ansiati, A., Axial Flow Cyclone with a Variable
Internal Arrangement for Abating Liquid or Solid Particulate from a Gas
Stream, World Intellectual Property Organization Patent: WO
2007/129276 A2.
Andreux, R., Ferschneider, G., Hmati, M. and Simonin, O., Experimental study
of a fast gas-solid separator, Chemical Engineering Research and Design,
Vol. 85 (6A), 808 814 (2007).
Avidan, A.A., Krambeck, F.J., Owen, H. and Schipper, P.H., "FCC Closed
Cyclone System Eliminates Post-Riser Cracking", Natural Petroleum
Refiners Association Annual Meeting, AM-90-33 (1990).
Barnes, P.H. and van Bylandtlaan, C., Apparatus and process for solids-fluid
separation, European Patent: EP0206399 (1992).
Barth, W., Leineweber, L., Beurteilung und Auslegung von Zyklonabscheidern,
Staub, Vol. 24, Iss. 2, 41 55 (1964).
Bnoni, D., Briens, C.L., Baron, T., Duchesne, E. and Knowlton, T.M., A
procedure to determine particle agglomeration in a fluidized bed and its
effect on entrainment, Powder Technology, Vol. 78, 33 42 (1994).
Bohnet, M. and Lorenz, T., Separation efficiency and pressure drop of cyclones
at high temperatures, in Gas Cleaning at High Temperatures, 17 31,
Clift, R., Seville, J.P.K., eds., London: Blackie Academic and Professional
(1993).
Bohnet, M., Influence of the gas temperature on the separation efficiency of
aerocyclones, Chem. Eng. Process., Vol. 34, 151 156 (1995).
Brunazzi, E., Paglianti, A. and Talamelli, A., Simplified Design of Axial-Flow
Cyclone Mist Eliminators, AIChE Journal, Vol. 49, 41 51 (2003).
Bugajski, M., Horn, M., Nix, D.G. and Antram, R.I., New multifunctional
refractory concretes for fluid catalytic cracking units in refineries,
UNITECR '05 - Proceedings of the Unified International Technical
Conference on Refractories: 9th Biennial Worldwide Congress on
Refractories, 921 925 (2006).
Casal, J. and Martinez, J.M., A better way to calculate cyclone pressure drop,
Chem. Eng., Vol. 90, 99 (1983).
Chen, J. and Shi, M., A universal model to calculate cyclone pressure drop,
Powder Technology, Vol. 171, 184 191 (2007).
Chen, Y-M., Dewitz, T.S., Dirkse, H.A., Wilhelmus, H., Dries, A. and Sanborn,
R.A., Cyclone Separator, United States Patent: 7,160,518 B2 (2007).
Chmielniak, T. and Bryczkowski, A., Method of calculation of new cyclone-type
separator with swirling baffle and bottom take off of clean gas part II:
experimental verification, Chemical Engineering and Processing, Vol.
40, 245 254 (2001).
65
66
Forgac, J.M., Hauschildt, F.W., Quinn, G.P., Rundell, D.N., Schwartz, J.G. and
Camp, M.S., Quenching downstream of an external vapor catalyst
separator, United States Patent: 5,043,058 (1991).
Garcia-Mallol, J.A., Horizontal cyclone separator for a fluidized bed reactor,
United States Patent: 6,245,300 (2001).
Gartside, R.J. and Woebcke, H.N., Low residence time solid-gas separation
device and system, United States Patent: 4,288,235 (1981).
Gartside, R.J. and Woebcke, H.N., Low residence time solid-gas separation
device and system, United States Patent: 4,433,984 (1984).
Gartside, R.J. and Woebcke, H.N., Low residence time solid-gas separation
device and system, United States Patent: 4,556,541 (1985).
Gartside, R.J. and Norton, R.C., Particulate solids cracking apparatus and
process, United States Patent: 6,482,312 (2002).
Gauthier T., "Etude de la sparation rapide dans un cyclone co-courant", Thse
de Doctorat de l'Universit Paris VI, Paris, France (1991).
Gauthier, T.A., Briens, C.L., Bergougnou, M.A. and Galtier, P., Uniflow
Cyclone Efficiency Study, Powder Technology, Vol. 62, Iss. 3, 217-225
(1990).
Gauthier, T., Andreux, R., Verstraete, J., Roux, R. and Ross, J., Industrial
Development and Operation of an Efficient Riser Separation System for
FCC Units, International Journal of Chemical Reactor Engineering, Vol.
3, Article A47 (2005).
Gimbun, J., Chuah, T.G., Fakhrul-Razi, A. and Choong, T.S.Y., The influence
of temperature and inlet velocity on cyclone pressure drop: a CFD study,
Chemical Engineering and Processing, Vol. 44, 7 12 (2005).
Griffiths, W.D. and Boysan, F., Computational Fluid Dynamics (CFD) and
Empirical Modelling of the Performance of a Number of Cyclone
Samplers, J. Aerosol Sci., Vol. 27, No. 2, 281 304 (1996).
67
68
Ld, J., Li, H.Z. and Villermaux, J., Le Cyclone Racteur Partie I: Mesure
Directe de la Distribution des Temps de Sjour de la Phase Gazeuse Lois
dExtrapolation, The Chemical Engineering Journal, Vol. 42, 37 55
(1989).
Letzsch, W.S. and Earl, G., Short residence time cracking apparatus and
process, United States Patent: 5,662,868 (1997).
Licht, W., Air Pollution Control Engineering, Dekker, New York (1980).
Lim, K.S., Kim, H.S. and Lee, K.W., Characteristics of the collection efficiency
for a cyclone with different vortex finder shapes, Aerosol Sci., Vol. 35,
743 754 (2004).
Liu, B.Y.H. and Rubow, K.L., A new axial flow cascade cyclone for size
characterisation of airborne particulate matter, In: Aerosols (Edited by
Liu, B. Y. H., Pui, D. Y. and Fissan, H. J.), Elsevier, Amsterdam, 115
118 (1984).
Maynard, A., A Simple Model of Axial Flow Cyclone Performance under
Laminar Flow Conditions, J. Aerosol Sci., Vol. 31, 151 167 (2000).
McLean, J.B., FCC catalyst properties can affect cyclone erosion, Oil and Gas
Journal, Vol. 98, 33 36 (2000).
Mothes, H. and Lffler, F., Motion and Deposition of Particles in Cyclones,
German Chemical Engineering, Vol. 8, Iss. 4, 223 233 (1985).
Muschelknautz, E., Die Berechnung von Zyklonabscheidern fr Gase, ChemieIng.-Techn., Vol. 44, 1 12 (1970).
Muschelknautz, E., Theorie der Fliehkraftabscheider mit besonderer
Bercksichtigung hoher Temperaturen und Drcke, VDI-Berichte, Vol.
363, 49 60 (1980).
Nielsen, B.B., Berruti, F. and Behie, L.A., Riser terminator for internally
circulating fluid bed reactor, United States Patent: 5,665,130 (1997).
Nieuwstadt, F.T.M. and Dirkzwager, M., A Fluid Mechanics Model for an Axial
Cyclone Separator, Industrial & Engineering Chemistry Research, Vol.
34, 3399 3404 (1995).
69
70
71
72
Tuzla, K. and Chen, J.C., Performance of a cyclone under high solids loading,
AIChE Symposium Series, Vol. 289, 130136 (1992).
Umney, L.E.R., On the Theory and Design of an Improved Centrifugal Air
Cleaner, Report R33, National Gas. Turbine Estab. (1948).
UOP LLC, Refining: FCC Vortex Separation Technology: VDS and
VSS, <www.uop.com> (2003)
Van Den Akker, H.E.A., Everts, R., Woudstra, J.J., Barnes, P.H., Verheul, C.M.
and Dirkse, H.A., Process for the Separation of Solids from a Mixture of
Solids and Fluid, United States Patent: 4,961,863 (1990).
Van der Lans, R.P., Glarborg, P., Dam-Johansen, K. and Larsen, P.S., Residence
time distributions in a cold, confined swirl flow, Chemical Engineering
Science, Vol. 52, Iss. 16, 2743 2756 (1997).
Vaughan, N.P., Construction and testing of an axial flow cyclone preseparator,
J. Aerosol Science, Vol. 19, Iss. 3, 295-305 (1988).
Wei, F., Jin, Y., Qian, Z., Yang, Y.H., Wang, Z.W., Wall-attached cutting type
fast gassolid separator, (Chinese) Chinese Patent: CN 1267564 (2000).
Zenz, F.A., Cyclones, Encyclopedia of Chemical Processing and Design, Vol.
15, J.J. McKetta ed., M. Dekker Inc. (1982).
Zenz, F.A., Cyclone-Design Tips, Chemical Engineering, Vol. 108, 60 64
(2001).
Zhang, Y., Wang, X., Riskowski, G.L., Christianson, L.L., and Ford, S.E.,
Particle separation efficiency of a uniflow Deduster with different types
of dusts, ASHRAE Transactions, Vol. 107 Part 2, 93 98 (2001).
Zhongxi, C., Guogang, S., Jiao, J., Zheng, Y., Bing, G. and Mingxian, S., Gas
flow behavior and residence time distribution in a rough-cut cyclone,
Chemical Engineering Journal, Vol. 106, 43 52 (2005).