EPA-650/2-74-119-b
ENTRAINMENT SEPARATORS
FOR SCRUBBERS
- FINAL REPORT
by
Seymour Calvert, Shuichow Yung, and James Leung
A.P.T. , Inc.
4901 Morena Boulevard
Suite 402
San Diego, California 92117
Contract No. 68-02-0637
ROAP No. 21ACX-086
Program Element No. 1AB013
EPA Project Officer: L. E. Sparks
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, North Carolina 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
August 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These series are:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2 . ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SOCIOECONOMIC ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to
develop and demonstrate instrumentation, equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution. This work provides the new or improved
technology required for the control and treatment of pollution sources
to meet environmental quality standards.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
Publication No. EPA-650/2-74-119-b
11
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-650/2-74-119-b
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Entrainment Separators for Scrubbers-
Final Report
5. REPORT DATE
August 1975
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Seymour Calvert, Shuichow Yung, and James Leung
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORQANIZATION NAME AND ADDRESS
A. P.T. , Inc.
4901 Morena Blvd. , Suite 402
San Diego, CA 92117
10. PROGRAM ELEMENT NO.
1AB013: ROAP 21ACX-086
11. CONTRACT/GRANT NO.
68-02-0637
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final: 10/73 - 6/75
11. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
repOrf gives results of an analytical and experimental study of the use of
entrainment separation to remove the liquid mist carried out of a scrubber by the
effluent gas. It includes an evaluation of current technology, results of experimental
studies of entrainment separator characteristics, and theoretical analyses. Zigzag
baffle, knitted mesh, tube bank, packed bed, and cyclone devices were tested. Col-
lection efficiency and reentrainment were measured and related to drop size and
separator geometry. Pressure drop as a function of gas flow rate is also reported.
The effects of suspended solids on collection efficiency and the nature and extent of
solids deposition were also investigated. An auxiliary experiment was employed to
help determine solid deposition mechanisms. Mathematical models are given for
predicting primary collection efficiency and pressure drop.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Air Pollution
Scrubbers
Entrainment
Efficiency
Mathematical Models
Exhaust Gases
Air Pollution Control
Stationary Sources
Entrainment Separators
Collection Efficiency
Liquid Mist
Suspended Solids
13B
07A
07D
14A
12A
21B
8. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASo
Unclassified
.This Report)
20. SECURITY CL-'-El (This page)
Unclassified
21. NO. OF FA:
J^PRICF
EPA Form 2220-1 (9-73)
i i*a
-------
ACKNOWLEDGEMENT
A.P.T., Inc. wishes to express its appreciation for
excellent technical coordination and for very helpful
assistance in support of our technical effort to
Dr. Leslie Sparks, E.P.A. Project Officer.
111
-------
TABLE OF CONTENTS Pace
Acknowledgement iii
List of Figures v
List of Tables xiii
Nomenclature xiv
Abstract xix
Sections
Chapter 1 - Introduction 1
Chapter 2 - Summary and Conclusion 9
Chapter 3 - Basic Concepts 17
Chapter 4 - Experimental Pilot Plant 47
Chapter 5 - Mesh 65
Chapter 6 - Packed Bed 83
Chapter 7 - Tube Bank 97
Chapter 8 - Cyclone 109
Chapter 9 - Zigzag Baffles 123
Chapter 10 - Air-Water-Solid Experiments 149
Chapter 11 - Solids Deposition 155
Chapter 12 - Design Approach 179
Chapter 13 - Future Research and Development
Recommendations 195
References 201
IV
-------
LIST OF FIGURES
No. Page
3-1 Entrainment Correlation for Plates 20
3-2 Flooding Limits for Bubble Caps and
Perforated Plates 20
3-3 Sieve Plate Entrainment Size Distribution . . 21
3-4 Entrainment Flow Rate Versus Liquid to
Gas Ratio with Superficial Gas Velocity
as Parameter for Mobile Bed 26
3-5 Entrainment Drop Diameter Versus Liquid
to Gas Ratio with Superficial Gas Velocity
as Parameter for Mobile Bed 26
3-6 Theoretical Impaction Efficiency as a
Function of Inertial Parameter for
Different Targets 28
3-7 Terminal Settling Velocity and Reynolds
Number for Water Drops in Air at 20°
and 760 mm Hg 29
3-8 Extrapolation Method for Determination of
Point of Onset of Entrainment for Vertical
Downflow in 2.2 cm I.D. Tube 34
3-9 Breakdown of Disturbance Wave by
Undercutting 36
3-10 Breakdown of Disturbance Wave by Rolling. . . 56
3-11 Typical Impingement Separators 42
3-12 Typical Centrifugal Separators 44
4-1 System Flow Diagram for Vertical Test
Section 49
4-2 Nozzle Positions in the 30.5 cm jc 61 cm
Duct 52
v
-------
No. Page
4-7) Cyclone Assembly 55
4-4 Top View of Baffle Arrangement 55
4-5 Front View of Inclined Baffle Section -
45° Inclination 57
4-o Dimensions for a 30° Inclined Baffle 57
4-7 Dimensions for a 45° Inclined Baffle 57
4-8 The Effect of Gas Velocity on Drop
Diameter for M6 60
4-9 Drop Diameter Versus Volume Percentage
for Hollow Cone Nozzle Spraying Water
at 10.2 atm Gauge Pressure (Manufac-
turer's Data) 53
4-10 Drop Diameter Versus Volume Percentage
for Hollow Cone Nozzle Spraying Water
at 6.8 atm Gauge Pressure (Manufac-
turer's Data) 63
4-11 Drop Diameter Versus Volume Percentage
for Fulljet Nozzles Spraying Water at
2.7 atm Gauge Pressure (Manufacturer's
Data) 64
5-1 Friction Factor, f, Versus Reynolds
Number, N „ for Wire Mesh Entrainment
Ke , u
Separator with Entrainment Load 68
5-2 Pressure Drop Due to Presence of Liquid
in the Knitted Mesh with the Crimps in
the Same Direction 70
5-3 Pressure Drop Due to Presence of Liquid
in the Knitted Mesh with the Crimps in
the Alternate Direction 70
5-4 Effect of Liquid Entrainment Load on
Allowable Gas Velocity 72
VI
-------
No. Page
5-5 Experimental Collection Efficiency of
Wire Mesh for Horizontal Gas Flow 72
5-6 Experimental Penetration for Vertical
Gas Flow up Mesh 73
5-7 Pressure Drop in Wire Mesh Versus
Horizontal Gas Velocity with Liquid
Load as Parameter 75
5-8 Comparison Between Experimental and
Predicted Dry Pressure Drop for Mesh 75
5-9 Pressure Drop in Knitted Mesh Versus
Vertical Gas Velocity with Liquid
Load as Parameter 76
5-10 Outlet Drop Diameter for Mesh Separator
with Horizontal Gas Flow 76
5-11 Drop Diameter Versus Geometric Standard
Deviation for Mesh 77
5-12 Effect of Gas Velocity and Liquid Load
on Performance of Mesh 79
5-13 Effect of Entrainment Load on Reentrain-
ment Onset Velocity 79
5-14 Onset of Reentrainment Velocity Curves of
Mesh for Horizontal Gas Flow 80
6-1 Generalized Flooding and Pressure Drop
Correlation for Packed Beds (Perry, 1963) . . go
6-2 Experimental Collection Efficiency in
Packed Bed, Horizontal Gas Flow,
Pall Rings 91
6-3 Collection Efficiency in Packed Bed,
Vertical Gas Flow, Pall Rings 91
6-4 Dry Pressure Drop in Packed Bed,
Pall Rings 92
6-5 Wet Pressure Drop in Packed Bed,
Pall Rings 92
vii
-------
No. Page
6-6 Wet Pressure Drop in Packed Bed,
Pall Rings 93
6-7 Experimental Versus Predicted Pressure
Drop Across 30 cm of 2.5 Pall Rings 95
6-8 Correlation for Onset of Reentrainment
in Cross Flow Beds 95
7-1 Theoretical and Experimental Collection
Efficiencies of Rectangular Aerosol Jets. . . 98
7-2 Collection Efficiency Versus Gas Velo-
city in Tube Bank with n = 6, d
84 urn and a = 1.32 ?g 101
&
7-3 Collection Efficiency Versus Gas Velo-
city in Tube Bank with d = 380 ym and
*g - 1.5 P? 101
7-4 Collection Efficiency Versus Gas Velo-
city in Tube Bank 102
7-5 Collection Efficiency Versus Gas Velo-
city in Vertical Direction in Bank of
Tubes 102
7-6 Dry Pressure Drop in Tube Bank Versus
Gas Velocity 105
7-7 Wet Pressure Drop in Tube Bank Versus
Gas Velocity 105
7-8 Wet Pressure Drop in Tube Bank
infi
7-9 Experimental Results Showing the Effect
of Gas Velocity and Liquid Load on
Performance of Tube Bank in Cross-
Flow Pattern 108
7-10 Experimental Results Showing the Effect
of Gas Velocity and Liquid Load on
Reentrainment for Tube Banks with
Vertical. Gas Flow 108
Vlll
-------
No. Page
Cyclone with Tangential Gas Inlet 112
Theoretical Grade Efficiency Curve of
The Cyclone Used in the Present Study
with Inlet Gas Velocity as Parameter 113
8-3 Comparison of Entrainment Onset Veloc-
ity by Different Investigators 115
8-4 Experimental Penetration Versus Gas
Velocity in Cyclone Inlet with and
Without Vane 118
8-5 Experimental Dry Pressure Drop Versus
Gas Velocity in Cyclone Inlet 118
8-6 Experimental Dry Pressure Drop Versus
Volumetric Flow Rate in Cyclone 120
8-7 Comparison of Experimental Pressure
Drop Data and Predicted Pressure Drop
for Cyclone with Inlet Vane by Shepherd
S Lapple (1940) ' 120
9-1 Drag Coefficient Versus Reynolds Number
After Foust et al (1959) , with
Sphericity ip as the Parameter 126
9-2 Drag Coefficients for Flow Past
Inclined Flat Plates (Data from
A. Page P7 F. C. Johansen, (1927) . 126
P-3 Predicted Superficial Reentrainment
Velocity due to Tearing of Drops
with Vertical Flow 131
9-4 Predicted Superficial Reentrainment
Velocity due to Tearing of Drops
with Horizontal Flow 131
9-4a Predicted Superficial Reentrainment
Velocity and Maximum Reentrained Drop
Diameter for Horizontal Gas Flow 132
9-5 Experimental Collection Efficiency
for Zigzag Baffles 135
IX
-------
No. Page
9-6 Collection Efficiency for Vertical
Zigzag Baffle Device 135
9-7 Collection Efficiency for Vertical
Zigzag Baffles 136
9-8 Experimental Penetration Versus Gas
Velocity in Vertical Direction in
Zigzag Baffles 138
9-9 Overall Penetration Versus Vertical
Gas Velocity for Drops having Mass
Median Diameter of 1230 ymA for 45°
Inclined Baffles 138
9-10 Overall Penetraiton Versus Vertical
Gas Velocity for Drops Hav ing Mass
Median Drop Diameter of 400 ym for
45° Inclined Baffles 139
9-11 Overall Penetration Versus Vertical
Gas Velocity for Baffles Inclined
at 30° to Horizontal 139
9-12 Dry Pressure Drop in Zigzag Baffles 141
9-13 Wet Pressure Drop in Zigzag Baffles 141
9-14 Predicted Pressure Drop from Gen-
eralized Pressure Drop Correlations for
Packed Bed Versus Experimental Pressure
Drop in Zigzag Baffles 142
9-15 Effect of Gas Velocity and Liquid Load
on Performance of Vertical Baffles 144
9-16 Effect of Gas Velocity and Liquid Load
on Performance of Horizontal Baffles .... 144
9-17 Effect of Gas Velocity and Liquid Load
on Performance of 45° Inclined Baffles . . . 145
9-18 Effect of Gas Velocity and Liquid Load
on Performance of 30° Inclined Baffles . . . 145
9-19 Some Observed Phenomena in Entrainment
Separator 147
A
-------
No. Page
10 Air-water-solid System 151
11-1 Trapping of Particle by Thick Liquid
Film 158
11-2 Trapping of Particle by Thin Liquid
Film 158
11-3 Experimental Set-up for Solid Depo-
sition Test ,,n
16(J
11-4 Baffle Structure .,.
loz
11-5 Particle Size Distribution for CaCO,
Particles 3 ,-.
164
11-6 Solid Deposition Rate Versus Slurry
Flow Rate for Vertical Baffle at an
Angle of 30° with the Direction of
Gas Flow 166
11-7 Solid Deposition Rate Versus Slurry
Flow Rate for Inclined Baffle with
Slurry Sprayed at the Upper Surface 166
11-8 Slurry Deposition Rate Versus Slurry
Flux for Inclined Baffle and with the
Slurry Sprayed at the Under Surface 167
11-9 Comparison of Figures 11-6, 11-7, and
11-8 167
11-10 Slurry Deposition Rates for Inclined
and Vertical Baffles 168
11-11 Drop Size Distribution Plot for Run
#10 168
11-12 Solids Deposition Rate vs. Slurry
Flux for Big Drops 171
11-13 Solids Deposition Rate vs. Slurry
Flux for Big Drops 171
11-14 Deposition Rate Versus Slurry Flux
for Big and Small Drops 172
-------
No. Page
11-15 Predicted Penetration Versus Drop
Diameter for Zigzag Baffles 172
11-16 Predicted Deposit Thickness Along
A Baffle Surface 30 cm from Top 178
11-17 Predicted Deposit Thickness Versus
Distance from Top Edge of Baffle
At 3 cm from Leading Edge 178
12-1 Entrainment Separator Design and
Selection Information Sheet 184
12-2 Entrainment Separator Approximate
Operating Range 185
12-3 Integrated (overall) Penetration as a
Function of Cut Diameter, Particle Para-
meters and Collector Characteristic
12-4 Overall Penetration as a Function of
Cut Diameter and Particle Parameters for
Common Scrubber Characteristic, B = 2. . . .
12-5 Performance Cut Diameter as a Function
of Pressure Drop for Several Entrainment
Separators 191
12-6 Ratio of Drop Diameter to cut Diameter
as a Function of Collection Efficiency . . . 191
xn
A
-------
LIST OF TABLES
No. Page
4-1 Nozzles Used in Spray Section 51
4-2 Drop Size Analysis 52
6-1 Bed Porosity, e, for Various Packing
Materials 86
6-2 Experimental Values of j , Channel Width
as Fraction of Packing Diameter 86
6-3 Packing Factors, "F", for Dumped
Pieces (m2/m3) 87
Packing Factors , "F" :
Stacked Pieces (m2/m )
6-4 Packing Factors, "F", for Grids and
88
7-1 Comparison of Tube Banks 104
9-1 Comparison of Baffle Type Entrainment
Separators 134
10-1 Experimental Results for Cyclone
(Air-Water-Solid System) 150
10-2 Experimental Results for Baffle
(Air-Water-Solid System) 150
12-1 Comparison of Various Types of
Entrainment 186
12-2 Summary of Design Information 193
JClll
-------
NOMENCLATURE
Latin
A = cyclone inlet area, cm2
A = total projected area of baffles per row in the
direction of inlet air flow, cm2
A. = duct cross-sectional area, cm2
a = cyclone inlet duct height, cm
= acceleration due to centrifugal force, cm/sec2
a' = cross-sectional area of all the tubes in
one row, cm2
aa = specific area of mesh; surface area of wires
per unit volume of mesh pad, cm2/cm3
as = constant
7 = constant
b = distance between baffles normal to gas flow, cm
= cyclone inlet duct width, cm
= jet orifice width, cm
= channel width, cm
C-. = drag coefficient
C' = Cunningham slip factor
Cl = constant defined by equation (4-1)
c2 = constant defined by equation (4-1)
d = duct width or channel width, cm
d = cyclone diameter, cm
\f
= collector diameter, cm
= packing diameter, cm
d, = drop diameter, cm
d = exit pipe diameter of the cyclone, cm
C
d = equivalent (hydraulic) diameter of liquid
eq
film, cm
xiv
A
-------
d, = hole diameter ,cm
d, = drop attachment length
d = mass median drop diameter, cm
d = inlet mass median drop diameter, cm
IT &
d = Sauter mean diameter, cm
E = primary collection efficiency, fraction
F = centrifugal force, dyne
= packing factor, cm2/cm3
F-, = foam density
F = column wall curvature correction factor
w
f = friction factor
f-rj =.drag coefficient
f, = fraction of the perforated open area in
the plate
fr = friction factor in the absence of liquid
b
phase
f• = interfacial friction factor
G = mass flow rate of gas, Kg/m2-sec
g = acceleration of gravity, cm/sec2
H, = fractional liquid hold-up in the bed
h, = dry plate head loss
h = head over weir
h = residual pressure drop
h = height of vertical cylinder of cyclone, cm
hw = weir height, cm
j = ratio of channel width to packing diameter
L = mass flow rate of liquid, Kg/m2-sec
= natural length of the cyclone
L/A = superficial liquid velocity, cm/min
i = length of baffle, cm
= distance between orifice and impingement plane
= length of settler
£2 = length of mesh pad in the direction of flow, cm
m = mass of drop, g
N = number of stages in the tube bank
xv
-------
NR ~ drop Reynolds number
NR „ = gas Reynolds number
N~ y = liquid Reynolds number
n = number of rows of baffles or tubes
= vortex exponent
n. = number of semicircular bends
n. = collection efficiency for a given particle
diameter in one stage of rectangular jet
impingement
P = pressure, dyne/cm2
Pt = fractional penetration
Ap = pressure drop, cm W.C.
Ap, = pressure drop in absence of liquid, cm W.C.
Apy = pressure drop due to presence of liquid,
cm W.C.
Qp = volumetric flow rate of gas, m3/sec
Qy = volumetric flow rate of liquid, m3/sec
R = universal gas constant
= radius at the water line along the particle
surface made by remaining water
= radius of the circle, cm
R = solid deposition rate, mg/cm2-sec
r = distance from vertical axis of the cyclone
r' = radius of curvature between the particle
surface
r = collection wall radius, cm
r = drop radius, cm
S = height of exit pipe inside cyclone, cm
T = absolute temperature, °K
t = mean residence time, sec
= drop travelling time, sec
U,- = flooding gas velocity, m/sec
Up = superficial gas velocity, based on empty
duct, cm/sec
xvi
A
-------
Up, = gas velocity through channel, cm/sec
u' = actual gas velocity, cm/sec
u = drop terminal velocity, cm/sec
u. = drop terminal centrifugal velocity, cm/sec
u = tangential velocity, cm/sec
v = geometric average of the gas velocity
3. v G
at the cyclone inlet and outlet, cm/sec
Ve = effective volume of the cyclone, m3
v, = velocity of gas through hole, cm/sec
V, = annular shaped volume above exit duct
inlet to mid-level of entrance duct, m3
V9 = volume of cyclone below exit duct inlet
to the natural length of the cyclone, m3
W = weight fraction of solid in slurry
w = baffle width, cm
w.. = weir length, m
Z = bed length, cm
Greek
a = angle made between suspension surface and
contact angle of the medium against the
particle
(3 = parameter defined by equation 7-1
6 = liquid film thickness, cm
4> = mole of entrained liquid per mole of gross
downflowing liquid
= ratio of water density to entrained liquid
density
Pi = drop density, g/cm3
•~-r = gas density, g/cm3
PT = liquid density, g/cm3
LJ
pwater = density o£ water, g/cm3
a = liquid surface tension, dyne/cm
a = geometric standard deviation
xvi i
-------
n = collection efficiency, fraction
yr = gas viscosity, poise
y, = liquid viscosity, poise
T. = interfacial shear stress, g/cm-sec2
e = porosity
VT = kinematic viscosity of liquid, cm2/sec
LI
8 = angle of inclination of the baffle to the
flow path, degree
= slurry flux, mg/cm2-sec
Subscripts
a = air
i = interfacial
G = gas
L = liquid
p = drop
w = water
xvi 11
A
-------
ABSTRACT
Entrainment separation, which is used to remove the
liquid mist carried out of a scrubber by the effluent gas,
has been studied in the analytical and experimental pro-
gram described in this report. Included in the report
are an evaluation of current technology, the results of
experimental studies of entrainment separator character-
istics, and theoretical analyses.
Zigzag baffle, knitted mesh, tube bank, packed bed,
and cyclone devices were tested. Collection efficiency and
reentrainment were measured and related to drop size and se-
parator geometry. Pressure drop as a function of pas flow
rate, the effects of suspended solids on collection effi-
ciency, and the nature and extent of solids deposition were
also investigated. An auxiliary experiment was employed to
help determine solid deposition mechanisms. Mathematical
models for predicting primary collection efficiency and
pressure drop were developed.
xix
-------
A
-------
CHAPTER 1
INTRODUCTION
A scrubber is designed to promote good contact between
the gas and liquid and a frequent consequence is that
small drops of liquid are formed and carried out with the
gas. To make matters worse, it is also common to find
that the gas flow rate is increased as much as possible
in order to attain more capacity with a given piece of
equipment. This will cause both a higher rate of drop
formation and a greater tendency for drops to be swept
out with the gas.
The liquid entrainment or mist, as it is commonly
referred to, will generally contain both suspended and
dissolved solids. The suspended solids can be due to
the particles collected by the scrubber, substances
introduced into the scrubbing liquid, or products of
chemical reaction occurring within the scrubber. Dis-
solved solids may similarly come from the impurities in
the gas, reagents introduced into the scrubber liquid,
or products of reaction.
Entrainment carryover can cause a variety of problems
both within the air pollution control system and in the
ambient atmosphere after the effluent has been emitted.
Drops can collect on the fan blades where they may either
dry out or deposit solids causing vibration and con-
sequent failure of the fan blades, housing, or sup-
porting structure. The entrainment also can cause
corrosion or erosion of the fan blades or housing. Liquid
or residual solid entrainment can also be deposited in
the ductwork and smoke stack, causing eventual plugging and
-------
possible corrosion, depending on the chemical nature of
the system and the materials of construction. In cases
where the scrubber effluent is reheated, entrainment can
collect on the heat exchange surfaces of the reheater where
it can cause eventual plugging and/or corrosion. Plugging
will cause an increase in resistance to gas flow and
therefore an increase in pressure drop through the system.
This will in turn cause increased power consumption and
possible overloading of the fan motor. Entrainment which
finally emerges from the stack can cause problems in the
area immediately surrounding the point of emission due to
"rain-out" of liquid drops. In cases where a reheater has
been used, the emission will include the solid residues
of the dried out entrainment drops and in some cases the
quantity of material can even exceed the quantity of parti-
culate matter which entered the scrubber. The composition
of the particulate matter can be quite different than that
of the particulate which entered the scrubber, especially
where reactive solutions or slurries are used for gas scrub-
bing. Thus, a bizarre consequence of excessive entrainment
from a scrubber system can be that more pollutant is
emitted either in total or within a certain size range than
entered the scrubber.
In many cases the occurrence of excessive entrainment
will impose a limitation upon the capacity of the scrubber.
That is, while the scrubber itself might be capable of
handling a larger gas flow rate, the generation of entrain-
ment would be considered excessive at some point and this
criterion will dictate a maximum gas flow rate which could
be handled with a given piece of equipment.
A
-------
All scrubber systems include an entrainment separator,
either as an integral part of the scrubber configuration
or as a separate, clearly identifiable device. Some entrain-
ment separation will occur by gravitational settling or due
to centrifugal forces caused by a change in gas flow
direction within the exit region of the scrubber. For
example, plate type scrubbers are routinely designed with
a definite amount of clear space for disengaging entrain-
ment above the top plate. Scrubber geometry may or may not
be such that the entrainment, once it has been removed
from the gas, is permitted to drain back into the scrubber
rather than being swept along the walls of the scrubber
into the outlet gas.
There are a number of devices which are commonly used
as entrainment separators (or mist eliminators) which are
added either within the scrubber body or in another vessel.
Zigzag baffles, knitted mesh, packed beds, cyclone separators,
and guide vanes causing rotation of the gas stream are
frequently used for this purpose. While entrainment sep-
arators have been used for many years, their major application
had been in relatively clean systems, such as chemical
processing equipment. Consequently, the performance of
the entrainment separators was not too critical and the
duty not very severe. Where entrainment separators were
used in air pollution control systems, there was often a
lack of awareness of the importance of the entrainment
separator unless the problems encountered were especially
severe or the air pollution control requirements were un-
usually restrictive.
The situation at the time that the research program
being reported here was initiated could be characterized
as one in which increasing demands on air pollution control
-------
systems had forced the recognition of many of the short-
comings and problems associated with existing entrainment
separators. For one thing, the collection efficiency
of the entrainment separator for the incoming entrainment
was limited and very likely unknown. The nature of the
entrainment,in terms of drop size, was also unknown for
most situations. Once the drops are captured, there is
the problem of removing them from the entrainment separator
without their being reentrained. This liquid handling
capacity was another cause of a limitation in the capacity
of the entrainment separator. Where solids were present,
the entrainment separators were susceptible to plugging}
caused by solids deposition and this in turn would cause
increased pressure drop and possible corrosion of the
materials.
In general, the characteristics of the entrainment
separators were not known well enough to permit good designs
and specifications to be made. Consequently, the entrainment
separator might be sized either too large or too small
and its cost might be too high or not realistically high
enough. The materials of construction could be inappro-
priate for coping with the corrosive effects of moist
deposited solids. Maintenance might not be convenient or
even possible. And in many cases, the type of entrainment
separator might be totally inappropriate, causing a higher
pressure drop than would actually be required to perform
the necessary function.
The program which is reported here was undertaken in
order to develop better information on the characteristics
of existing entrainment separators and to point the way
to the development of improved entrainment separators.
The objectives and scope of the research are presented in
the following section.
A
-------
SCOPE OF WORK
The scope of work carried out in this program included
the following:
Evaluate Technology
The status of present technology relating to wet
scrubber entrainment separators was evaluated and included:
1. Review and assessment of the published literature
and available unpublished information, including,
where appropriate, information acquired through
private communication with manufacturers, de-
signers and users of entrainment separators.
2. Determination of the availability and adequacy of
operational and design data for entrainment sepa-
rators .
3. Determination and evaluation of the adequacy of
existing theoretical models and design methods
for predicting the performance of entrainment
separators .
4. Review and evaluation of the performance of all
major types of entrainment separators currently
available. Assessment of advantages, disadvan-
tages and limitations for each type of equipment.
5. Identification of specific operating and main-
tenance problems associated with entrainment
separators. Particular attention was paid to
the problems encountered in S07 scrubbing systems
Lt
under development in E.P.A. programs.
Experimental Study
An experimental study of gas-water systems was aimed
at simulating the performance of various types of entrain-
ment separators in the presence of soluble and insoluble
particulate matter. The experimental study investigated
such variables as efficiency, pressure drop, reentrainment
velocity, plugging and related problems.
-------
Selection and Design
Improved engineering equations and methods were
developed for entrainment separator selection and design.
Recommendations
Specific research and development recommendations
for improving wet scrubber entrainment separators were
developed.
GUIDE TO REPORT
The primary objective of this study is to review and
evaluate the performance of all major types of entrainment
separators currently available and to identify specific
operating and maintenance problems associated with entrain-
ment separators. This report is written in the hope that
it will be helpful to the process engineer in the selection
and design of entrainment separators for scrubbers.
Chapter 3 gives an overall view of all the entrainment
separators available. The mechanisms of drop collection
and drop formation are defined, and the performance of each
entrainment separator, as regards to inlet drop size, pri-
mary collection efficiency, reentrainment, and pressure
drop are compared. The last part of Chapter 3 also gives
an account of the operational problems frequently encoun-
tered in entrainment separators.
Chapters 5 through 9 give an account of the design
equations in predicting primary collection efficiency,
pressure drop and reentrainment of the five common types
of entrainment separators, namely, wire mesh, packed bed,
tube bank, cyclone and zigzag baffles. The experimental
results are compared to the mathematical models for each
type of entrainment separator tested.
Chapter 10 studies the effect of solids in entrained
drops on the performance of cyclone and zigzag baffles.
A
-------
In Chapter 11, the problems dealing with solids
deposition on an entrainment separator are investigated.
The mechanisms of solids deposition are defined and an
equation to predict the deposition trend on a baffle sur-
face is developed.
Chapter 12 summarizes the design methods and infor-
mation developed and identified in this study. It is
intended to guide the engineer in the design or selection
of an entrainment separator.
Chapter 13 defines the areas in which future research
and development are needed.
-------
Page Intentionally Blank
A
-------
CHAPTER 2
SUMMARY AND CONCLUSIONS
This program involves the experimental and theoretical
investigation of wet scrubber entrainment separation.
The objectives of this study are to (1) Evaluate
present technology, (2) Conduct an experimental study of
air-water systems aimed at simulating the performance of
various types of entrainment separators, (3) Develop im-
proved engineering equations and methods for entrainment
separator selection, (4) Develop and evaluate on a small
pilot basis new entrainment separator design, and (5) De-
velop specific research and development recommendations.
EVALUATE PRESENT TECHNOLOGY
A literature search was carried out to evaluate the
technology on wet scrubber entrainment separators. Manu-
facturers of entrainment separators were contacted by
mail and asked for information. Visits were made to E.P.A.
and T.V.A. facilities to identify the specific operating
and maintenance problems associated with entrainment sepa-
rators .
The study indicates that presently available entrain-
ment separators suffer from various shortcomings. Examples
are: overdesign, which necessitates large equipment size;
low operating velocities due to flooding or reentrainment;
unpredictable performance due to lack of reliable industrial
operating data; and plugging by solids.
The existing theoretical and empirical models which
predict the performance of the entrainment separators were
evaluated. The criteria for this evaluation were soundness
of derivation and closeness of comparison with actual per-
formance .
Preceding page blank
-------
EXPERIMENTAL STUDY
A pilot plant to study wet scrubber entrainment
separators was built. It has a gas flow capacity of
85 m3/roin (3,000 CFM) and it consists of prefilter, blower,
heater, spray section, observation sections, test section,
various supply and catch tanks, and auxiliary equipment.
Five types of entrainment separators, namely, mesh, tube
bank, packed bed, cyclone, and baffles were studied. The
experiments were done with air and water, with and with-
out suspended solids in the water. Observations included
collection efficiency, pressure drop, reentrainment, flood-
ing, drainage, drop size distribution, solid deposition,
and other variables.
SELECTION AND DESIGN
Mathematical models for determining the following were
developed in the present study:
1. Primary collection efficiency in zigzag baffle
type entrainment separators.
2. Pressure drop in zigzag baffle type separators.
3. Primary collection based on either complete
turbulent mixing or no mixing.
4. Reentrainment in vertical zigzag baffles.
5. Reentrainment in horizontal zigzag baffles.
6. Reentrainment in a cyclone.
7. Solid deposition in zigzag baffle.
CONCLUSIONS
The principal objectives of this study were achieved.
The following conclusions can be drawn, based on evaluation
of experimental results.
Primary Collection Efficiency
1. At low gas velocities (under industrially used
conditions), primary collection efficiency of
10
A
-------
knitted mesh, packed bed, tube bank, and cyclone
can be predicted reasonably well by means of
mathematical models presented in the literature.
2. Theoretical models were developed in the present
study for zigzag baffles. One model, based on
turbulent mixing, reaches 100% efficiency as an
assymtote with increasing gas velocity. On the
other hand the model, based on no mixing, reaches
1001 efficiency as a straight line on efficiency
versus gas velocity curve. The assumption of
turbulent mixing gives better agreement with
actual performance of entrainment separators.
3. The primary collection efficiency can be quickly
predicted by means of a graphical correlation of
cut diameter with pressure drop for some typical
zigzag baffles, packed bed, tube bank, and knitted
mesh. The same correlation can be used for other
separator types.
4. The efficiency is not affected by the presence
of solids in the entrainment as long as the solids
deposited do not change the separator geometry
significantly.
5. The orientation of separator mounting method has
no effect on primary collection efficiency despite
its effect on the liquid drainage capability and
onset of reentrainment.
Capacity
The capacity of an entrainment separator is limited
by reentrainment which is a function of gas velocity, entrain-
ment flow rate, and drainage. Thus, capacity can be defined
in terms of these variables.
1. Maximum gas velocity and liquid flow for negligible
11
-------
reentrainment has been determined experimentally
for knitted mesh tube, packed bed, and zigzag
baffles. A correlation given by Chien and Ibele
is recommended for determining the onset of
reentrainment in a cyclone.
2. Liquid drainage capability of an entrainment sep-
arator has great effect on reentrainment velocity.
Cross flow configuration with horizontal gas
flow has the highest drainage capability and
thus the highest reentrainment velocity.
3. Relationships between quantity of reentrainment
and flow rates of gas and liquid have been exper-
imentally determined for all five types of sep-
arators used in this program.
Nature of Reentrainment
1. At high gas velocities, reentrainment is a defi-
nite problem. Reentrainment may take place by
various mechanisms such as: a) Transition from
separated flow to separated-entrained flow, b)
Rupture of bubbles, c) Creeping of liquid on the
entrainment separator surface, and d) Shattering
of liquid drops resulting from splashing.
2. Transition from separated flow to separated-
entrained flow depends upon gas velocity, liquid
Reynolds number and liquid properties. The tran-
sition does not depend upon the duct dimensions.
The drop size distribution is independent of the
duct dimensions. The average drop diameter re-
sulting from this transition is about 250 ym. The
reentrainment velocity is considerably reduced if
jets of air stream strike the liquid film at an
angle. Therefore, sharp angles should be reduced
to avoid reentrainment.
12
-------
3. The mechanism of reentrainment is zigzag baffles
is tearing of the liquid sheets caused by high gas
velocities and shattering of liquid drops. Reen-
trainment in cross flow baffles with horizontal
gas flow should be less than in baffles with ver-
tical gas flow. Zigzag baffles inclined at 30°
from gas flow direction should have less reentrain-
ment than baffles inclined at 45° from horizontal
eas flow direction.
4. The reentrainment mechanisms in packed bed and mesh
pad are shattering of drops and rupture of bubbles.
Reentrainment resulting from small drops (less than
40 ym) due to rupture of bubbles is insignificant.
5. The mass median drop diameter due to reentrainment
was determined to vary between 80 ym and 750 ym.
Large drops (above 200 ym) are present due to shat-
tering of drops.
6. Sampling of liquid drops and entrainment needs care-
ful consideration. Due to large drop size in the
reentrainment, a sedimentation effect is present.
Pressure Drop
1. Zigzag baffles- The pressure drop in zigzag baffles
can be determined from drag coefficients for in-
clined plates held in the flow. The effect of liq-
uid load on pressure drop is small. Wet pressure
drop for vertical gas flow can also be predicted
from generalized pressure drop correlation for
packed beds.
2. Tube bank - Pressure drop is predictable by means
of correlations available from the published
literature relating to heat exchanger tube bundles.
3. Packed bed - Generalized pressure drop correlation
predicts a higher pressure drop across the bed
13
-------
than that measured in this study.
4. Cyclone - The experimental data can be correlated
by an equation which has the same form as that
given by Shepherd and Lapple. The only difference
between these two equations is that the constant
in the equation of present study is 2.7 times
smaller than that in Shepherd and LappleTs equation.
5. Mesh - Pressure drop depends on liquid velocity
and gas velocity. It varies according to u^1*55.
(j
6. The orientation of the separator has little effect
on pressure drop and except for knitted mesh, the
presence of liquid entrainment only increases the
pressure drop slightly.
Solid Deposition
Based on the results of solids deposition experiments,
it appears that:
1. The solids deposition rate depends largely on drop
size and entrainment flow rate. Small drops cause
a higher deposition rate than large drops. In-
creasing the liquid flow rate will increase the
liquid film thickness and thus increase the scou-
ring action of the liquid collected on the sur-
face.
2. Deposition rate is higher on an inclined surface
due to increased settling rate of the suspended
solids.
3. The empirical correlation on solid deposition
rate, derived from small scale experiments, agrees
fairly well with observations made on baffles.
Future Research
Entrainment separator design or specifications by
means of rational methods is possible to a useful degree.
14
A
-------
Several important areas require further study before the
state of knowledge will be adequate for the reasonably
through and accurate design of an entrainment separator.
Some of these are:
1. Reentrainment mechanism and loading for separa-
tors under various operating conditions.
2. Entrainment loading and drop size distribution
from various scrubbers under different
operating conditions.
3. Solid depositions and factors affect the depo-
sition rate.
4. Effective separator washing method and flow
rate of washing liquid.
15
-------
Page Intentionally Blank
A
-------
CHAPTER 3
BASIC CONCEPTS
Liquid entrainment can be defined as the carrying
over of liquid particles by a carrier gas or vapor which
moves at too high a velocity to permit the quick settling
out of the droplets by gravity. Liquid entrainment can
result in serious loss of liquid or contamination of the
atmosphere. For this reason, entrained drops of liquid
must be separated from the gas. Thus, entrainment sepa-
rators are frequently employed to separate the liquid
from gas.
The design and operation of most entrainment sepa-
rators are governed by three factors:
1. Pressure drop
2. Collection efficiency
3. Reentrainment velocity and reentrainment rate
Kno\vledge of the pressure drop through a separation
system is important in calculating the energy loss incurred
and in selecting the proper pumps and other auxiliary
equipment to overcome that energy loss.
Collection efficiency or overall collection efficiency
is defined as the fractional collection of the droplets
by the separator, i.e.
i effluent concentration \ r ^ "M
influent concentration / *• •*
When the gas velocity in the entrainment separator
is high, some separated droplets in the separator will be
reentrained in the gas stream. Because of this reentrain-
ment, the observed collection efficiency of the separator
Preceding page blank
17
-------
is less than the primary collection efficiency which is
defined as the efficiency an entrainment separator would
have if reentrainraent were not present.
Reentrainment velocity is the gas velocity at which
drops are first observed to become reentrained in the
gas. The onset of reentrainment will vary for different
kinds of entrainment separators and different operating
conditions. Reentrainment velocity determines the maximum
allowable gas velocity in the separator. Reentrainment
rate and drop size distribution are needed for the predic-
tion of emissions from the system.
Once design equations predicting the primary effi-
ciency, pressure drop, and reentrainment are available,
operating characteristics of the entrainment separator can
be established.
ENTRAINED LIQUID INFORMATION
In order to design a proper entrainment separator,
or to predict the collection efficiency of an entrainment
separator, certain entrainment liquid information is
needed. This includes:
1. Entrainment drop and size distribution.
2. Quantity or inlet loading.
An extremely important factor in chosing a.nd
designing an entrainment separator is drop size distribu-
tion. Different entrainment separators are limited to
certain drop diameters, below which their efficiency falls
off sharply. The size of the drops depends upon the way
they were formed. Basic mechanisms of drop formation
are described later in the section on reentrainment.
Little information is available on the drop size
distribution of entrainment from scrubbers. More attention
18
A
-------
seems to have been given to the quantity of entrainraent,
although the published data on this are also very limited.
The data found in this study are presented below and are
organized according to the scrubber type.
Plates
Figure 3-1 shows a correlation of the available data
for entrainment in bubble-cap and sieve plate gas liquid
contacting columns, (Perry, 1973). The entrainment is
expressed in "i^", moles of entrained liquid per mole of
gross downflowing liquid (net flow plus return of entrain-
ment) . For gas-water contacting the mole ratio is the
same as the mass ratio so "^" is the mass rate of entrain-
ment per unit of water mass flow rate. The parameter
"percent of flood" is the actual vapor velocity divided
by the flooding vapor velocity at the same L/G. Entrain-
ment increases with decreasing tray spacing and this effect
is accounted for in Figure 3-1 because the flooding velocity
is a function of tray spacing.
Figure 3-2 represents a correlation of flooding
velocities for sieve and bubble cap plates with several
fluid flow rate and property parameters. As shown, the
flooding velocity increases with plate spacing; therefore,
the entrainment ratio decreases with plate spacing.
Because this correlation was developed to describe the
entrainment from plate-to-plate, the rate given by Figure
3-1 is that which would be measured at a distance of one
plate spacing above the top plate. Scrubbers usually have
more clear space above the top plate so the entrainment
rate leaving the scrubber would be less than predicted
by Figure 3-1.
Other studies, such as by Hunt, et.al. (1955) ,
Atteridge et.al. '(1956), Brooks et.al. (1955) and Jones
and Pyle (1955) indicate lower entrainment ratios than
19
-------
o.ooi
0.005 0.01 0.02 0.05 0.1 0.2
0.5 1.0
Figure 3-1 - Entrainment Correlation for Plates
Tray spacing, cm
94 ' I I :
0.04
0.03
0.01 0.02 0.03 0.050.070.1 0.2 0.3 0.5 0.7 1.0
0.5
Figure 3-2 - Flooding limits for bubble caps and
perforated plates.
KV = 3.28 Uf /20.0
0.2
PL-PG
0.5
20
A
-------
, 0 C 0
501
I i • • 1 I
50 LJL
0.01
Measured at 13 cm
above froth
l i
xC
^> —
Calculated
for settlim
i i i
i i I / i i i i i 11 i i i l
0.1 0.51 2 5 10 20 30 40 50
CUMULATIVE MASS UNDERSIZE, %
Figure 5-5. Sieve plate entrainment size distribution
-------
given by Figure 3-1. Therefore designs based on Figure
3-1 will be conservatively large.
For illustration, we may note that for a water to
gas ratio of 1.34 £/m3 (10 gal/MCF) the mass ratio,
L/G = 1.1 Kg/Kg. If the plate spacing is 46 cm (18")
the flooding velocity evaluated for standard air and
water properties from Figure 5-2 is about 2 m/sec. At
50% of flooding (i.e., 1 m/sec superficial gas velocity),
the entrainment ratios from bubble cap and sieve plates
are given by Figure 3-1 as 0.024 and 0.018 mol/mol (or
Kg/Kg), respectively. This means that the predicted
liquid entrainment measured 46 cm above the top plate
would be 0.03 £/m3 and 0.024 &/m3 for cap and sieve
plates, respectively. At 91 cm above the top plate the
entrainment would correspond to that for 91 cm plate
spacing which for a sieve plate would be 0.0053 Kg/Kg.
Drop size distribution data for entrainment measured
13 cm above a sieve plate are reported in Perry (1973) as
shown in Figure 3-3, a log-probability plot. The facts
that the superficial air velocity at which these data were
taken was 61 cm/sec and the terminal settling velocity of
a 180 ym dia water drop is about 61 cm/sec (see Figure 3-7)
enable us to see the influence of sampling point elevation.
Figure 3-3 shows that 99.6% of the liquid volume was larger
than 180 \im and would settle out of the air stream if the
height above the plate were sufficient. If the drops larger
than 180 ym were removed, the remaining size distribution,
as shown by the dashed curve, would have a mass median dia-
meter of roughly 150 ym and a a of 1.8 (based on the small
&
diameter end of the curve).
22
A
-------
Gas Atomized Sprays
Entrainment rate and size distribution data for gas
atomized spray scrubbers such as Venturis have not been
reported. Estimates can be made, as discussed below,
but they are very rough because of uncertainties in pre-
dicting the characteristics of the initial atomization
and the drop separation occurring within the venturi
diffuser and similar flow elements.
Drop diameter can be predicted by means of the cor-
relation by Nukiyama and Tanasawa (1938-40). For air
and water at standard conditions the N+T correlation
for Sauter mean diameter is:
°
uUWsec)
&
where :
d = Sauter (volume-surface) mean diameter
of drops, cm
U = air velocity relative to drops, cm/sec
&
QT = water flow rate, m3/sec
Qr = air flow rate, m3/sec
According to Steinmeyer in Perry (1973) , the Sauter
mean diameter is typically 70% to 90% of the mass median
diameter. This implies that the geometric standard de-
viation, a , runs about 1.6 for 901 and 2.3 for 70%.
&
To illustrate the application of the above to the
prediction of entrainment characteristics for a venturi
scrubber, we can consider the case of a throat air velocity
of 100 m/sec and water to air ratio of 1 £/m3(10'3m3/m3) .
23
-------
The gas pressure drop would be about 100 cm W.C. and the
Sauter mean diameter computed from eq. (3-2) is 79 ym.
From the typical ratios of mass median to Sauter diameter,
we would expect the mass median drop diameter to range
from 88 to 113 ym, with a from 1.6 to 2.3, respectively.
One would therefore predict that the cumulative entrain-
ment concentration would be related to drnr> SIZP within
the range of high and low values tabulated below.
Drop diameter, ym 4 5 10 15 20
High concentration,
cm3/™3 0.035 0.11 2 8 20
Low concentration,
cm3/m3 - - 0.0025 0.06 0.6
If the entrainment contained 101 solids by weight,
the residual particle concentrations after evaporation
would be such that if one wanted to limit the particle
loading due to entrainment to 0.01 g/m3(0.0044 gr/ft3)
it would require the separation of all entrainment larger
than 5 ym diameter for the high estimate and 16 ym diameter
for the low. Since particle loadings of this magnitude can
be significant for plume opacity, the example shows the
efficiency with which entrainment must be controlled and
the necessity for good data on entrainment size distribu-
tion and concentration.
24
A
-------
Mobile Bed
Calvert et. al. (1975) measured the entrainment
flow rate and size distribution from a mobile bed (T.C.A.
type) scrubber. Data were taken at a location about 76 cm
above the top grid of the mobile bed. Figure 3-4 is a plot
of entrainment flow rate versus liquid to gas ratio with
superficial gas velocity as parameter. Figure 3-5 is a
plot of mass median drop diameter of the entrainment as
a function of liquid to gas ratio with superficial gas
velocity as parameter. The geometric standard deviation,
a , for all operating conditions is approximately equal
to 1.8.
COLLECTION MECHANISMS
Knowledge of the basic mechanisms of drop collection
is fundamental to an understanding of entrainment separators
The separation mechanisms which have been used for entrain-
ment are:
1. Inertial impaction
2. Sedimentation
3. Centrifugation
4. Interception
5. Diffusion
6. Electrostatic precipitation
Sub-micron drops are present in very small quantity
in the entrainment generated by scrubbers so diffusional
collection is not important. Cost considerations generally
weigh against the use of electrostatic precipitators for
entrainment separation. The design and operating conditions
of separators thus favor inertial impaction, sedimentation,
and centrifugation as the principal mechanisms of collection,
25
-------
LIQUID TO GAS RATIO, i/m1
Figure 3-4. Entrainment flow rate versus liquid to
gas ratio with superficial gas velocity
as parameter for mobile bed.
1,000
800
a eoo
400
10
15
LIQUID TO CAS RATIO, Jl/m2
Figure 3-5. Entrainment drop diameter versus
liquid to gas ratio with super-
ficial gas velocity as parameter
for mobile bed.
NOT REPRODUCIBLE
26
A
-------
Inertial Impaction
Inertial impaction is the major collection mechanism
in scrubber entrainment separators. When a fluid approaches
an obstacle the fluid streamlines spread around it. At the
same time inertial forces carry drops across the streamlines
so that the drops hit and stick to the obstacle. It is as-
sumed that all drops colliding with the obstacle adhere to
it.
Two factors determine impaction collection efficiency.
The first is the velocity distribution of the gas flowing
by the collector, which varies with the Reynolds number of
the gas with respect to the collector. The second factor
is the drop trajectory, which depends on the mass of the
drop, its air resistance, the size and shape of the collec-
tor, and the rate of flow of the gas stream.
Collection efficiency can be predicted from the equa-
tions of motion of a drop for a given gas flow pattern
and a collection parameter. The "target" efficiency ex-
presses the fraction of the particles in the entraining
fluid, moving past an object in fluid, which impinge on
the object. Figure 3-6 from Golovin and Putman (1962),
gives theoretical "target" efficiency as a function of
the inertial parameter for different targets.
Sedimentation
The second collection mechanism important in entrain-
ment separators is sedimentation. Figure 3-7, from Fuchs
(1964) , is a plot of drop terminal settling velocity versus
drop radius. Drop diameters encountered in wet scrubber
entrainment may vary from 50 to 500 ym, and the terminal
settling velocity for these drops will range from 0.1 to
2.0 m/sec. The gas velocities used in entrainment sepa-
rators vary from 1.0 to 12.0 m/sec; however, except for
27
-------
u
i—i
v->
w
1.0
0.8
0.6
0.4
0.2
0
• — - Rectangular half body
. _ _ (ribbon with vtaks)
^^w-^ Ellipsoid
i^^^p of revolu-
NACA 65*004 -3 t
ze TO .in'le 06
attack 4t thick
low-drag s^rrjne:-
r i c 21 airfoil
Joukouski 15% thick syn-
ne cricai ai rfo i 1 .it ze re
angle of atta^k
0.1 1 10
INERTIA! PARAMETER, K =
' P
100
C'Pd,.dd_UG
Figure 3-6 - Theoretical impaction efficiency as
a function of inertial parameter
for different targets.
28
A
-------
0.1
-100
w
CQ
LO
Q
nJ
O
W
Di
O
C£
O
0)
a;
100
1,000 3,000
Figure 3 - 7 .
DROP RADIUS, ym
Terminal settling velocity and
Reynolds number for water drops
in air at 20°C and 760 mm Hg.
29
-------
cyclone-type separators, which operate at very high velo-
cities, most operate below 4.0 to 5.0 m/sec. Therefore,
sedimentation can be expected to affect the separation
of drops .
Centrifugation
When the entrainment laden gas is put into spinning
motion, centrifugal force affects the droplets. The centri
fugal force is much greater than gravity, therefore, drop-
lets are thrown to the wall and collected.
If a gas stream moves round the arc of a circle, and
it is assumed that the droplet has the same tangential
velocity as the gas stream, then the centrifugal force
on the droplet is given by:
u
F = m -£ (3-3)
R ^ J
where F = centrifugal force
m = mass of the drop
u = tangential component of the gas velocity
R = radius of the circle
If the droplets are sufficiently large and have
high enough initial velocity, they are thrown to the
wall close to the inlet. On the other hand, when liquid
drops are small, they are carried by the gas flow part
of the way before being thrown out to the wall by centri-
fugal force.
The time required for the drop- travel from the initial
position "R" to the wall is
R
re- C3-3a)
30
A
-------
where t = time elapsed, sec
p, = drop density, g/cm3
VP = gas viscosity, poise
d, = drop diameter, cm
r = collector wall radius, cm
G
n = vortex component
= 0.5 - 0.7 for cyclones
If the time required is less than the residence time
of the gas, the drop will be collected.
REENTRAINMENT
The overall collection efficiency of an entrainment
separator is often found to be less than the primary
efficiency because of reentrainment. Increasing overall
efficiency means reducing reentrainment, the achievement
of which requires a knowledge of the parameters important
in determining the extent of reentrainment. Thus, engi-
neering equations describing this process are vital to im-
proved efficiency.
One cause of reentrainment is high gas velocity. To
avoid this hazard, entrainment separators have been oper-
ated at lower gas velocities than necessary, resulting in
the use of equipment which is larger and more expensive
than needed.
Reentrainment from an entrainment separator may take
place by any one of more of the following mechanisms:
1. Transition from separated to separated-entrained
flow caused by high gas velocity.
2. Rupture of bubbles at the gas liquid interface
and subsequent drop formation.
3. Creeping of the liquid along the solid surface
and movement into the gas exit in the entrain-
ment separator.
31
-------
4. Shattering of liquid drops due to impaction.
The last three mechanisms of reentrainment depend
upon the design of the entrainment separators. The first
mechanism represents the upper limit of the operation of
entrainment separators.
Transition from Separated to Separated-Entrained Flow
Reentrainment may occur at high gas velocities due
to transition from separated to separated-entrained
flow. In simple geometries such as straight tubes, the
transition takes place at much higher velocities than
those at which entrainment separators are operated.
Yet reentrainment is observed in separators at the
lower velocity. This is caused by such phenomena as
the impingement of the gas stream onto the liquid at
an angle and the presence of gas jets. Also the flow
pattern in the entrainment separator is not so uniform
as in circular tubes.
In the operation of entrainment separators, flows
may be horizontal, vertically upward or downward, or
inclined. The onset of reentrainment depends upon the
flow direction, flow geometry and the fluid properties.
The reentrainment models for simple geometries has been
examined and given in the "Initial Report" (1974) .
Effect of Impingement of Gas Jets - As mentioned earlier, the
gas and liquid phases do not flow parallel in the entrainment
separator. Jets of gas are present, which may impinge on the
liquid film at various angles. The presence of gas jets,
their impingement on the liquid film at various angles, etc.
depend upon the entrainment separator design.
32
A
-------
Wallis (1962) studied entrainment in ducts with various
inlet designs. The reentrainment velocity varies with inlet
design, from 18 m/sec to 24 m/sec. The data are shown in
Figure 3-8.
Interfacial Waves - The study of interfacial wave behavior
is important in the determination of transition from sepa-
rated flow to separated-entrained flow. Experimental and
theoretical studies of wave behavior and its influence on
other phenomena are still at a very early stage of develop-
ment. The most advanced theoretical studies have been conr
cerned with the problem of the initial formation of waves,
rather than their development and influence. However, the
instability of the waves represents the physical phenomenon
responsible for transition from separated to separated-
entrained flow. Thus, to understand the physical phenomena
responsible for reentrainment in entrainment separators,
one should look at the interfacial waves, breaking of the
waves, drag friction on the film due to gas flow, etc.
For vertical flow the only forces opposing these normal
stresses are those due to surface tension. For a stable
interface condition, the surface tension stresses exactly
balance the effects of the normal stress. On the other hand,
the wave will grow in amplitude when the sum of the local
liquid and gas normal stresses exceeds the surface tension
stress. It can be further deduced that the thinner the
liquid film the greater the gas velocity needed to cause
an increase in amplitude of a given size wave.
It may be expected that the effect of waves will be to
increase the friction factor. To determine the liquid flow
rate, it is necessary to have interfacial friction factor.
Roberts and Hartley (1961) found, on plotting friction
factor as a function of liquid film thickness for a given
gas velocity, that the friction factor did not begin to in-
crease with film thickness until a certain value had been
33
-------
60
50
H
S 40
f^,
2
i—i
H 3°
PJ
Ox= 20
10
1.28 2,/min H20 rate
20
40
60
AIR VELOCITY, m/sec
Figure 3-8 - Extrapolation method for determination of
point of onset of entrainment for vertical
downflow in 2.2 cm I.D. tube.
34
A
-------
exceeded. They were able to correlate the difference be-
tween the interfacial friction factor and that for the same
gas flow rate in the absence of the liquid phase, by the
equation:
r / \u'5
•F - -P 4. i c ^ 5 / 2 \ r •? A-\
±i - ±G + 1.5 a r, { f- 1 C3-4J
-
6
3 —
eq
5
Re , G
/ \ °-5 1
/ 2 \
N .
where
£i =
f . = interfacial friction factor
fp = friction factor in the absence of liquid
phase
6 = liquid film thickness, cm
= equivalent (hydraulic) diameter, cm
= interfacial shear stress, g/cm-sec2
eq
T.
N
R
= gas Reynolds number
Thus, for very thin liquid films there would be no
significant waves on the interface and no effective rough-
ness. For thicker films there would be a minimum instant-
aneous film thickness corresponding to the troughs of the
waves on the surface.
A number of possible mechanisms have been suggested
by which transfer of droplets can be effected by the waves,
but at present there is no definite evidence to favor any
particular one. Lane (1957) described the mechanism illus-
trated in Figure 3-9. The gas starts to "undercut" the
wave and a round, open ended bubble begins to form. The
bubble grows, leaving a thick-ringed filament around its
base and eventually breaks up into droplets. Once the
breakup occurs, the excess (dynamic) pressure inside the
bubble gives rise to a rapid radial transport of the
droplets.
35
-------
Gas
• Tims
Figure 3-9
Breakdown of Disturbance Wave by Undercutting
Time-
Figure 3-10
Breakdown of Disturbance Wave by Rolling
36
A
-------
An alternative form of breakup is illustrated in
Figure 3-10. A large amplitude wave on a shallow liquid
layer tends to steepen at the front and then to form a
breaking wave. If the gas velocity is very high, it
might be expected that the tips of the waves would be
drawn out into thin liquid sheets with subsequent breakup.
Reentrainment Due to Rupture of Bubbles
The second mechanism which leads to reentrainment
is rupture cf bubbles. This mechanism is the main cause
of the reentrainment of liquid drops into the gas phase in
devices such as sieve plate, bubble cap plate, packed bed,
and mesh type separators. The collapse of a bubble when
exiting from the liquid phase is associated with thinning
of the liquid film starting at the top part of the bubble.
The upper surface thins to the extent of becoming weak
enough to rupture. Rupture of the upper part of the bubble
film takes place when the film thickness is of the order of
0.1 ym, provided there are no external disturbance forces
leading to the rupture of films (Kitchener, 1964; Jashnani,
1971). The collapse of the bubble at the interface leads
to the release of surface tension energy which is converted
into kinetic energy. The kinetic energy is sufficient to
impart high velocities to liquid drops formed during this
process.
Drop formation due to bubble burst occurs in three
steps. The first step, the lifetime of the bubble at the
interface, lasts on the order of l/100th sec.or longer; the
actual bubble burst, the second step, takes a few micro-
seconds; and events subsequent to the bubble burst extend
over a few milliseconds.
37
-------
Creeping of Fluids
The presence of drag forces due to gas flow leads
to creeping of liquid in the entrainment separator.
Creeping may be prevented by providing a proper drainage
system. If creeping is not prevented, reentrainment
may occur.
Consider liquid and gas flowing in a vertical tube.
The gas is flowing vertically upward and liquid is flow-
ing as a film and therefore forming an annulus. The
liquid film is subject to various forces: drag force due
to gas flow in the vertically upward direction, gravity
force in the downward direction and frictional force
due to tube wall.
For gas velocity lower than the critical velocity
the liquid near the wall flows downward due to gravity.
As the gas velocity is increased the liquid at the inter-
face reverses its flow direction and moves with the gas ;
as a result the liquid film begins to thicken. At a
critical gas flow rate the liquid does not flow down
any more, and the liquid film thickens rapidly.
Shattering of Drops
Reentrainment may take place due to shattering of
drops in two ways :
1. Due to splashing of drops on the solid surface
2. Due to high relative velocity between gas and
liquid drops.
Shattering of the drops due to high relative velocity
between gas and liquid drops does not increase entrain-
ment in the gas phase. However, small drops are more
38
A
-------
liable to be carried away in the gas phase than large
drops and therefore shattering of drops should be avoided.
ENTRAINMENT REMOVAL EQUIPMENT
We have already seen the principal mechanisms of
entrainment separation. In many cases, actual equipment
combines two or more of those mechanisms. The following
section discusses each of the main equipment types.
Gravity Settlers
The gravity settler is one of the earliest and simplest
types of equipment for separating particles from gases.
The function of a gravity settler is to reduce the gas
velocity, from one which permits entrainment down to a
velocity that will permit gravity to remove the entrained
droplets. There are two basic types, tranquil and stirred.
The only effect of stirring is to maintain an even concen-
tration throughout the separator. In most cases, it neither
helps nor hinders the settling.
Primary Collection Efficiency - If the gas passes vertically
upward through the settler, all particles having terminal
velocities equal to or greater than the velocity of the
gas stream will be removed. A 100 urn water particle has
a terminal velocity in air of about 30 cm/sec. Thus, very
low velocities and consequently large equipment sizes are
required to remove particles which are 100 ym or less in
diameter.
For complete removal to take place if the gas passes
-------
horizontally through the settler, the drop terminal velo-
city multiplied by the residence time must equal the maxi-
mum settling height:
ut = bT (l-V
where u = drop settling velocity, cm/sec
Qr = volumetric gas flow rate, m3/sec
b = width of settler, cm
£ = length of settler, cm.
When u is not equal to Q^/b£, the removal efficiency
becomes:
1 n
U1
~%T
E = -
For droplets greater than - 0.15 cm (500< NR ), Newton's
law applies and,
rr
o.s
Sd,P
Ut '
where g = gravitational acceleration
d, = droplet diameter, cm
p , = droplet density, g/cm3
PG = gas density, g/cm3
When dj < 100 pm, Stokes law applies:
ut =
-LUHQ
where UG = gas viscosity, poise.
40
A
-------
For drops larger than 100 ym in diameter and smaller
than 0.15 cm, drop settling velocity can be taken from
Figure 3- 7 .
Pressure Drop - The pressure drop across gravity settlers
can be estimated reasonably well by the standard methods
of calculating pressure drop in a conduit. It is usually
very low (less than 1 cm W.C.), consisting primarily of
entrance and exit losses.
Impingement Separators
For its removal qualities, the impingement type sep-
arator depends on particles colliding with a surface. Some
typical impingement separators are shown in Figure 3-11.
The most extensively used impingement type separators are:
1. Wire mesh
2. Packed bed
3. Vanes or baffles
In addition, tube bank (staggered rods) type separators
appear to have useful characteristics even thoi:gh they
are not commonly used. These four types of impingement
separators will be discussed in detail in Chapters 5,6,7
and 9 respectively.
Centrifugal Separators
The centrifugal 'separator is a device utilizing
radial acceleration for separating the entrained particles
from the carrier. Because of the liquid's greater density
and momentum, the circular motion imparted to the fluid
causes the entrained particles to separate from the carrier
and impinge on the walls, then move downward by the vertical
41
-------
Gas flow.
Inlet
Gas flow
Figure 3-11 Typical Impingement Separators
(a) Jet impactor
(b) Wave plate
(c) Staggered channels
(d) Zigzag baffle
(e) Peerless line separator
(f) Strong separator
(g) Karbate line separator (staggered streamline
rods)
(h) Type E horizontal separator
(i) PL separator
(j) Wire mesh
42
A
-------
component of the force, as well as by gravitation.
Figure3-12 shows some typical centrifugal separators.
The cyclone is undoubtedly the most commonly used
type of centrifugal separator. This is due primarily to
its simplicity of construction and low maintenance costs.
Its efficiency is not as high as those of some other types
of separators. Often, if higher removal efficiency is
needed, they may be preceded or followed by supplementary
separators. A droplet size of 5 to 10 ym is generally
considered the lower size limit for particle removal.
The spinning motion can be applied to the gas stream
in several ways and cyclone types can be classified ac-
cordingly. The gases can be drawn through curved vanes
in a duct, in a unit called the "straight through cyclone"
or "vortex air cleaner", or they can be spun in a special
turbine. In the conventional or "reverse flow cyclone"
the gases are admitted tangentially to a cylindrical upper
section ; it contains a centrally placed exhaust pipe
penetrating below the tangential inlet, while a conical
lower section is connected to the dust hopper. The gases,
in this case, spiral down towards the apex of the cone
and then are reversed up again through the exit.
The primary collection efficiency, pressure drop,
and reentrainment for cyclone separators are discussed
in Chapter 8.
Other Types of Entrainment Separators
In general, any device that can be used to remove solid
particulates can also be employed to remove entrainments.
For example, scrubbers and electrostatic precipitators are
commonly used to separate liquid mists.
43
-------
'Outlet
Inlet,
Gas out
Gas out
Gas in
Gas out
•Skimmer
By-pass
channel
Gas inlet
•Inlet
Receiver.
(a)
Skimmer
edge-
Primary
discharge -^
baffle ^
Secondary
drain
(e)
Figure 3-12 Typical Centrifugal Separators
(a) Multiclone
(b) Thermix ceramic tube
(c) Van Tongeren cyclone
(d) Sirocco type D collector
(e) Horizontal steam separator
44
A
-------
Tray Towers - Tray towers are vertical channels in which
the liquid and gas are contacted in stepwise fashion on
trays or plates. The liquid enters at the top and flows
downward by gravity. On the way, it flows across each
tray and through a downspout to the tray below. The gas
passes through openings in the tray, then bubbles through
the liquid to form a froth, disengages from the froth, and
passes onto the next tray above. There are various tray
geometries. The sieve tray and bubble cap are the two most
common types:
Sieve plates -
Primary efficiency - Taheri and Calvert (1968)
derived an equation for sieve plate primary col-
lection efficiency:
E = 1 - exp (-40 Fj Kp) (3-10)
where 0.30 < F£ < 0.65,
pXVh
KD = -ftra^ (3-11)
p yyGah
where F£ = foam density, ratio of clear liquid
height to total foam height
v, = velocity of gas through hole, cm/sec
d, = hole diameter, cm
h
Pressure Drop - Perry (1963) has suggested that
the pressure drop in sieve plates can be calcu-
lated according to:
AP = hw + how + hdp + hr (3-12)
45
-------
where,
h = weir height = 4-9 cm, assume 5 cm,
if unknown
n
XT
h = head over the weir = 0.143 F —
ow w w,
h, = dry plate head loss = -—
d? c2 PL 2§
\/J r\ "1" O T*
h = residual pressure drop = 0.013
r PL
1 i
-i- = 1 14 [0 4 (1 25 - f ) + fl - f ) ]
-2 ' ' ' h h (3-15)
where,
FW = column wall curvature correction factor =1.1
QT = liquid flow rate, here in m3/hr
w, = weir length, m
f^ = fraction of the perforated open area in the
plate
Bubble-cap Trays - Equations used to predict primary col-
lection efficiency and pressure drop of sieve plates can
also be applied to bubble-cap trays.
46
A
-------
CHAPTER 4
EXPERIMENTAL PILOT PLANT
An experimental pilot plant for the study of entrain-
ment separators was designed and built. The purpose of
the pilot plant was to do the following:
1. To obtain reliable data over a wide range of oper-
ating variables in order to provide a basis for the
improvement or development of new separators.
2. To check presently available design equations
for entrainment separators
A. Efficiency of separation
B. Pressure drop
3. To determine the effect of higher gas velocity
on reentrainment, bouncing of drops and
impaction mechanism
4. To study liquid drainage and flooding
5. To study problems associated with entrainment
separators
6. To study the effect of separator mounting
methods on its performance.
DESCRIPTION OF THE PILOT PLANT
The maximum capacity of the wet scrubber entrainment
separator is 85 m3/min (3,000 CFM). The capacity was
selected based on the following consideration. The en-
trainment separator cross-section was selected to be
30.5 cm x 61 cm. This section is sufficiently large to
have minimal wall effects for separators and provides a
fairly long (61 cm) collection element when cross-flow
effects are important. Normally, the maximum air velocity
in industrial separators is around 3.0 m/sec. If velocities
2.5 times higher are studied, the maximum air velocity will
be 7.5 m/sec. This will give the maximum capacity of
85 m3/nun.
47
-------
The sampling method and equipment used in studying
the horizontal test sections have already been described
in detail in the "Initial Report" and will not be repeated
here. The equipment used in studying the vertical test
section is given below.
EXPERIMENTAL SYSTEM '
Figure 4-1 is a flow diagram of the experimental system,
Liquid collected from--the drain was recirculated. The
amount of liquid recirculated into the system was recorded
by water meter #1. Barrel #2 acted as a reservoir. The
amount of liquid fed into the spray section was measured
by water meter #2.
Air Inlet
The air flow to the test section was supplied by a
Western Blower size 122 Bl and Class III. It has a capacity
rating of 88 m3/min at 30.5 cm W.C. (static pressure), a
7.5 kW (10 HP) motor, rotatable housing and an opposed
blade discharge damper. The blower was supported on a
hard rubber base to prevent vibrations and it was insulated
with accoustical fiberglass and concrete blocks to reduce
the noise level.
Spray Section
The spray section served to generate entrainment for
the test section. It was equipped with various nozzles
48
A
-------
as
. r, 1 e t
n
Observation sectior.
Test section
Pressure gauge
Spray section
Inlet and
drain
Automatic
level
control
n
Bypass
valve
i
Water Barrel 2
meter #1
Barrel 1
Figure 4-1. System flow diagram for vertical test section.
Water
meter
49
-------
from Spraying Systems Co. The nozzle specifications are
given in Table 4-1. In any section, the nozzles were
equispaced as shown in Figure A-2 to generate uniform flow.
The spray section was also equipped with a plexiglass
door, so that the spray drop size could be measured. Also
the spray nozzles could be changed without taking the whole
test assembly apart.
Observation Section
The observation section had dimensions of 50.5 cm x
61 cm, cress-section and 50 cm length. Two plexiglass
windows 30 cm x 30 cm, were installed on opposite siles
on each observation section. A door was provided for
sampling of entrainment drop diameters.
Liquid Catch and Liquid Supply Tanks
One 100 liter (30 gal) drum was used as the licmid
catch tank. The tank was connected to a water meter and
a pump with a liquid level controller for the recircula-
tion of liquid.
The liquid supply tank was a 200 liter (55 gal) drum,
The recirculated water from the liquid catch, tank was
fed back into the system through the liquid supply tank.
On the outlet side were located water meters and rota-
meters for flow measurements. The flowrate and pressure
into the system was controlled by the bypass valve.
Control Panel for Equipment
The control panel was equipped with the following:
1. Electrical connections
A. Magnetic starter for blower
B. Switches for pumps, heater, sampling pump,
50
A
-------
Table 4-1. NOZZLES USED IN SPRAY SECTION
Type of Nozzle
Hollow cone
Fogjet Nozzle
Full Cone
Hollow Cone
Model #
(Spray Systems)
1/4 M6SS
1-11 1/2 F18
1 1/2-11 1/2
F35
1/8 GG3
1/4 M26
Pressure
atm
13.6
2.7
2.7
2.7
2.7
# of
Nozzles
12
1
1
12
12
Flow rate
cm /sec
nozzle
14.2
1140
2200
63.0
27.2
51
-------
r
V,
c
\.
— 7.6^
•\ /-
J
•\
j
c
\.
c
L
-\
J
S
J
f
\,
•s
J
i i
5.1—
5.1
.•^n . q
— 5.1
— 7.6 —
K
0<
Of
/—
a
r—
V
3
-l
• . — 1
1
0
-1
00
-I-
I-O
00
Figure 4-2 - Nozzle positions in the 30 . 5cmx61cn\
duct. All dimensions in cm.
52
A
-------
observation lights, etc.
C. Temperature recorder
2. Non-electrical connections
A. Rotameters and water meters
B. Dry gas meter
C. Pressure gauges
D. Needle valve, diaphram valves, globe valves,
and gate valves
E. Manometers to measure pressure drop
Electrical Supply Panel
A 110 V, 3 phase, 90 amp/phase electrical supply
panel was installed near the equipment site.
Water Supply
Water supply to the spray nozzles:
Centrifugal pump - model 165U (Barnes Pump)
Maximum pressure - 3.4 atm (50 Psi)
Motor rpm - 3,450
Motor output - 1.1 kW (1.5 HP)
Flow rate at maximum pressure - 120 £/min (.31 GPM)
Test Section
Five different types of entrainment separators were
tested:
1. Mesh
2. Packed bed
3. Tube bank
4. Cyclone
5. Zigzag baffles
The test sections for the mesh, packed bed, tube bank,
53
-------
and zigzag baffles were the same as those used for the
experiments with horizontal air flow. They were described
in detail in the "Initial Report". The inclined baffle sec-
tion was built for experiments with vertical air flow only.
A brief description of each section is given as follows:
1. Mesh - ACS model 4CA mesh was used. The thickness
of the mesh was 10 cm, with 0.028 cm diameter wires
arranged in layers crimped in alternate directions.
Voids occupied 98.2% of the total volume and the mesh
surface area was 2.8 cm2/cm3. The mesh was located
in the first 30 cm of the test section.
2. Packed bed - Packing - 2.5 cm pall rings. Specific
surface = 1.9 cm2/cm3. Density = 0.088 g/cm3.
Material of construction = Polypropylene plastic.
Bed length = 30 cm.
3- Tube Bank - Number of rows = 6. External diameter
=1.9 cm. Length = 61 cm. Tubes per row = 8. Tube
spacing within row = 3.8 cm center - center spacing
between rows = 2.13 cm c-c.
4- Cyclone - The cyclone is a cylinder 61 cm diameter x
243 cm overall height. The cyclone inlet is 30.5 cm
high and 15 cm wide, giving a maximum inlet velocity
of 3,000 cm/sec. Higher velocities were studied by
using a vane in the inlet. The design is described
by Stearman and Williamson (1972) and is a straight
cylinder with flat bottom. Figure 4-3 shows the
cyclone used in the present study.
5. Zigzag Baffles - Baffle dimension = 7.5 cm width x
61 cm height x 0.16 thickness. Number of rows = 6.
Spacing between rows = 2.5 cm. Angle between baffle
surface and air flow direction = 30°. Spacing be-
tween baffles in a row =7.3 cm. Figure 4-4 shows
the baffle arrangement.
54
A
-------
Inlet
30x15-
cm2
K
\/
61
70
I-'igure 4-3. Cyclone assembly. All
dimensions are in cm.
Figure 4-4. Top view of baffle
arrangement.
55
-------
6. Inclined Baffles - Two inclined baffle sections were
built. One section with the baffle inclined 45° to
the horizontal and the other section inclined 30°.
Figure 4-5 shows the front view of the inclined baffle
section and Figures 4-6 and 4-7 are dimensions of each
baffle for 30° inclination and 45° inclination respect-
fully. The mounting method is the same as that shown
in Figure 4-4.
Flow Measurements
Air flow rate was measured by a standard pilot tube
located at the inlet air duct. Liquid flow to the spray
section was metered with a calibrated water meter.
Alternatively, the total amount of liquid flow to the
spray section can be determined by the sum of the
amount of liquid recycled and the amount of liquid
lost after the experiment. The amount of liquid lost
was given by the difference in liquid level in the
two tanks (which were calibrated) before and after
the experiment.
EXPERIMENTAL PROCEDURE
The major points of the experimental procedure are
described below. The procedure was modified as required
for individual runs.
1. All the tanks were emptied before starting the
experiment in order to avoid rust in the water.
2. All the wet bulb thermometers were checked for
56
A
-------
Figure 4 - 5 . Front X'lew of inclined
baffle secticn - 4S°
inc 1 inat i 01;.
"5° sharp bend
10F.° sharp ben.'i
\\
Figure 4-6. Dimensions for a 30C inclined baffle.
i
\m°
M'° sharp bend £»
' * rlll V
X
sharp
bend
6P en,
Figure 4-7. Dimensions for a 45° inclined baffle.
57
-------
water.
3. All the valves were checked so that the required
valves were kept opened and the rest closed.
4. All the recycle pumps were kept on to maintain
the liquid level in the tanks between the upper
and lower limits.
5. All the catch tanks were filled with liquid until
the level was between the upper and lower controlled
limits.
6. The feed supply tank was filled to the overflow
line.
7. The zero position of the inclined manometer was
adjusted.
8. Readings were noted for all the water meters and
the liquid levels in the catch tanks.
9. The desired air flow was started.
10. Pressure drop across the test section was
measured.
11. The desired water flow rate was started.
12. About 1-5 minutes were needed to reach steady
state. The experiment was continued for 2 hours.
13. Air flow rate, water flow rate, etc., were
checked every few minutes.
14. Visual observations of penetration, flooding,
liquid drainage, bouncing of drops, liquid flow
on elements of the entrainment separator, etc.,
were made for the duration of the experiment.
15. Readings were taken of temperature (each hour),
entrainment drop size, pressure drop, entrainment
loading, etc. (once during each run).
58
A
-------
16. At the end of the experiment liquid and air flow
were shut down. Readings were noted for water
levels in the tanks and water meter readings.
INLET ENTRAINMENT DROP SIZE
Various nozzles were used in the experiments, although
only one type of nozzle was used in any given experiment.
A description of the nozzles is given in Table 4-1. A complex
relationship among the characteristics of the individual
spray nozzles, the interaction of multiple nozzles, the con-
figuration of the experimental duct and the air velocity
determines the inlet entrainment size distribution.
The spray generated from the M6 nozzles was analyzed
under experimental conditions by filter papers coated with
1% potassium ferricyanide and ferrous ammonium sulfate as
described in the initial report. The drop diameter gen-
erated from the other nozzles was greater than 100 ym.
For these, the manufacturer's data were used to determine
drop diameters.
The effect of gas velocity on mass median drop diameter
generated from M6 nozzles is shown in Figure 4-8. There
is no definite trend. The mass median diameter varies
from 76 to 102 ym and averages 84 ym, with an average
geometric standard deviation of 1.32. The minumum drop
diameter found in the inlet was 30 ym.
Inlet entrainment measurements were made at a point
75 cm downstream of the spray nozzles and 30 cm upstream of
the test section. The average water supply pressure was
13.6 atm. gauge (200 psig). The nozzles were oriented toward
the downstream side and gave the drops an initial velocity
59
-------
100
e
t
w
w
1— 1
Q
CL,
O
o 75
i— i
Q
W
LO
to
50
0
1 1 1 1 1 1 1 1 1
O
0 0
o o
0
°o
0 o $>
o
1 1 1 1 1 1 1 i 1
5 10
Figure 4-8.
GAS VELOCITY, m/sec
The effect of gas velocity on
drop diameter for M6.
60
-------
of 51 m/sec.
An analysis of the drop diameters created by each nozzle
is given in Table 4-2 , and more detailed information concerning
the size distribution curves, as provided by the manufacturers,
is presented in Figures 4-9 through 4-1] These distributions
were measured 30 cm from the nozzles.
In these experiments the M26 nozzles were operated at
2.7 atm pressure, but the drop size data provided by the
manufacturer are for 6.8 atm and 10.2 atm. The mass median
drop diameter produced by M26 nozzles was obtained from
fitting the following relation for the effect of operating
pressure on drop diameter :
dpg = ClA (4-1)
where d = mass median drop diameter, cm
pg
AP = pressure drop at nozzle, atm
c , c~ = constant
The mass median drop diameter for an operating pressure
of 2.7 atm was 380 ym . The geometric standard deviation was
1.5 and did not significantly vary with operating pressure.
The nozzles often plugged, due to formation of rust in
the water tanks. This resulted in a decreased water flow
rate and also may have caused some variation in the drop
diameter and standard deviation.
It was observed that the entrainment flow rate reaching
the entrainment separator decreased with decreasing air
velocity. This is due to an increase in collection by the
walls of the spray section.
61
-------
Table 4-2. DROP SIZE ANALYSIS
Source of
Data
Manufacturer
Manufacturer
This Study
Manufacturer
Manufacturer
Predicted From
Equation (7-1)
Manufacturer
Type of
Nozzle
M6
M6
M6
M26
M26
M26
GG3
Operating
Pressure
atm gauge
6.8
10.2
13.6
6.8
10.2
2.7
2.7
Mass Median
Diameter ,
ym
127
110
84
295
265
380
1,230
Minimum Drop
Diameter,
ym
45*
45*
30
110*
102*
_
450*
Geometric
Standard
Deviation
1.5
1 .5
1.3
1.5
1.5
1.5
1.8
ON
tx)
*2% of the drops are smaller than this diameter.
-------
1,000 i
500
300
200
a.
x.
3 100
30
20
10
III 111 I I I I II I
5 10
80 90 do 98
ACCUMULATED VOLUME, t
Figure 4-9. Drop diameter versus volume percentage for
hollow cone norzle spraying water at 10.2 atm
gauge pressure (Manufacturer's data)
500
300
200
100
50
30
20
10
I 1 till
510 20 50 SO 90 95 98
ACCUMULATED VOLUME, i
Figure 4-10. Lrop diameter versus volume percentage for
hollow cone nozzle spraying water at 6.8 atm
gauge pressure. (Manufacturer's data;
63
-------
3,000.
2,000
r 1,000
500
400
500
Figure 4-11.
10
20 30
50
70 80
90 95
ACCUMULATED VOLUME, -3
Drop diameter versus volume percentage for full jet
nozzles spraying water at 2.7 atm gauge pressure.
(Manufacturer's data)
64
A
-------
CHAPTER 5
MESH
Knitted mesh of varying density and voidage is widely
used for entrainment separators. There are basically three
different kinds of mesh: (1) Layers with crimp in the
same direction - each layer is actually a nested double
layer. (2) Layers with crimp in alternate directions -
this results in an increase in voidage, reduced sheltering,
a decrease in pressure drop per unit length and an increase
in target efficiency per layer (3) Spirally wound layers -
the pressure drop is lower by about 2/3 than in layers with
crimp in the same direction, but the creeping of fluids, which
contributes to reentrainment, is expected to be higher.
Standard mesh 10-15 cm thick having a density of about
0.15 g/cm3 is used to remove drops larger than 5 ym in dia-
meter. Gas velocities range from 0.3 to 5 m/sec and liquid
flow rate is limited by the drainage capacity of the mesh
to 2.5 x 10"3 g/sec cm2 of mesh. A lower density mesh made
of standard wires is used when 10-201 higher flow rates are
desired.
Often two mesh type separators in series are used to
remove drops in the 1-5 ym diameter range. The first mesh,
normally made of fine wires, coalesces the small drops, and
the second mesh, made of standard wires, removes them. The
first mesh is operated beyond the flooding velocity and the
second under flooding velocity. A major disadvantage with
this arrangement is a pressure drop which may reach 25 cm
W.C.
65
-------
Some manufacturers use two or three stages of mesh,
the first being coarser and the final being finer, to re-
move large and small drops successively.
A mesh type separator has the advantage that it can
be made to fit vessels of any shape. Any material which
can be drawn into the shape of a wire can be used for fab-
rication. However, mesh separators are limited in appli-
cation because they plug easily. This can be avoided by
upstream washing, which will decrease removal efficiency
and increase pressure drop.
MATHEMATICAL MODELS
Primary Efficiency
Bradie and Dickson (1969) present the following
expression for primary efficiency in mesh separators:
E = 1 - exp (-I TT a2£2 n) (5-1)
where a~ _ specific area of mesh, surface area of wires
per unit volume of mesh pad, cm2/cm3
£? = thickness of mesh pad in the direction of gas flow,
cm
n = collection efficiency of cylindrical wire
The collection efficiency of cylindrical wire "n"
can be obtained from Figure 3-6. The factor of 2/3 in the
exponential was introduced by Carpenter and Othmer (1955)
to correct for the fact that all the wires in the knitted
mesh are not perpendicular to the flow. That factor is
the ratio of the projected area of wires perpendicular to
the flow to the cross-sectional area of wires along the
wire length.
66
A
-------
If the specific area, "a?",is not specified, it can be
determined from the mesh porosity, "e", and the knitted
mesh wire diameter, "d "
(5-2)
Pressure Drop
York and Poppele (1963) have suggested that the total
pressure drop in the knitted mesh is the sum of the pres-
sure drop in the dry knitted mesh and the pressure drop
due to the presence of liquid:
AP = APdry + APL (5-3)
where AP, = pressure drop in absence of liquid, cm W.C.
APT = pressure drop due to presence of liquid,
cm W.C.
York and Poppele considered the mesh to be equivalent
to numerous small circular channels and used The D'Arcy
formula for pressure drop in a pipe to correlate the dry
pressure drop through the mesh. York and Poppele 's data
for knitted mesh with crimps in alternated and in same
direction are plotted in Figure 5-1. Their data are close
to those obtained by Stasangee (1948) and Shuring (1946) .
Similar curves obtained by Bradie and Dickson (1969) for
spiral-wound and layered mesh are also plotted in Figure
5-1. Figure 5-1 should be used in determining dry pressure
drop, which is calculated from the expression
p u?,
(5-4)
67
-------
1.0
0.5
0.1
0.05
0.01
Satsangee data (1948)
and Shuring data (1946)
•***^ —
jZrimps in alternated direction
Crimps in same direction
- Layered mesh
Spiral-wound mesh
10
100
1,000
10,000
NRe,G = PG Va2 yG
Figure 5-1. Friction Factor, £, versus Reynolds
number, Np ,, for wire mesh entrainment
separator ' with entrainment load.
68
A
-------
The unit of "APj " is in dynes/cm2. It can be converted
to cm W.C. by dividing it by 981.
Pressure drop data due to presence of liquid are not
available for all operating conditions or for mesh of
different styles. Values of "AP," obtained by York and
Poppele are presented in Figures 5-2 and 5-3, with liquid
velocity as the parameter. Liquid velocity is defined as
tIT tl
— T— where ' L ' is the volumetric flow rate of liquid and
'A' is the cross -sectional of the mesh in liquid flow
direction. The specifications of the knitted mesh used
are shown in the two figures .
Maximum Allowable Gas Velocity
Several factors govern the allowable gas velocity
through wire mesh for a given set of conditions:
1. PL and PG
2. liquid viscosity
3. specific surface
4. liquid entrainment loading
5. suspended solid content
Application of the Souders -Brown equation for the
calculation of allowable vapor velocity for wire mesh
mist eliminator based on gas and liquid densities has
been suggested by York (1954) .
UG max = 30-5 a3 -r <5-4
\ b /
where "a," varies with operating conditions and mesh de-
sign. For most cases, a_, = 0.35. For air-water system,
u0 max = 3.1 m/sec .
(j
When liquid viscosity and entrainment loading are
high, or the liquid very dirty, a reduced value of "a.,"
69
-------
tin
J I
to
r-t
(NI
CO
'3'M uio
o
o
SH
(11
C .*"!
4-> -H cfl
•HO) C
o.'-i gco
O -H .H
}H <-H 13 4-1
T) O 0) l/l O
4-1 Cu 0)
CD 0) 4-i E }-,
fn D -H -H .H
in a) -^ u
to t/l 0)
!-, d X-G rt
a, &4-i 4-1 c
Ofi
J I
11 i I
CTl
CO
u
0)
g
•J
(U
c xi e
O -H 4-i 03
•l-l •!-(!/)
(D -H
3 3
13 CT
(U
g C
Q.
(U r* O
to to
(D 1) O O
p t £1, 4-> 4-1
S O
o
*
o
70
-------
must be used. The influence of liquid entrainment load-
ing upon "a^" has been investigated by Poppelle (1958)
for an air-water system. The data for the incipient
flooding are shown in Figure 5-4 together with the flood-
ing velocity correlation by Sherwood (1938) for dumped
rings. Also shown is a recommended design curve.
EXPERIMENTAL RESULTS AND DISCUSSION
Overall Efficiency
The overall collection efficiency data for horizontal
flow through wire mesh are plotted in Figure 5-5. No pene-
tration was observed in the experiments at low gas velocity,
less than 3.0 m/sec. At higher velocities, penetration
due to reentrainment was observed. The dotted line, pre-
dicting 1001 efficiency, represents the theoretical curve
based on equation (5-1).
The overall collection efficiency data for vertical
flow through wire mesh is plotted in Figure 5-6. Water
flow rate is used as a parameter. M6 nozzles were used
in the experiments. The effect of higher water flow rate
is to increase the penetration and decrease the onset of
reentrainment velocity.
If the performance of entrainment separators with
vertical air flow and horizontal air flow is compared,
the experimental data lead to the following conclusions:
1. Reentrainment velocities are lower in the
system with vertical gas flow than with
horizontal gas flow. This is because
vertically installed mesh provides better
drainage.
2. The amount of reentrainment is higher in
the system with vertical gas flow than with
horizontal gas flow.
71
-------
1.0
0.1
0.03
0.0001
.^incipient flooding
wire mesh
Dumped
Wire mesh
design curve
0.001
0.01
0 .1
Figure 5-4.
Effect of liquid entrainment load on
allowable gas velocity.
100
u
CJ
£ 80
H
u
o
CJ
60
50
Inlet Drop Diameter, ym
V 84
<> 380
D 1.230
O >1,230
J_ 1 1 I I
6 7
10
GAS VELOCITY, m/sec
Figure 5-5. Experimental collection efficiency of wire
mesh for horizontal gas flow.
72
A
-------
50
40
30
20
10
WATER FLOW RATE
O 3.9 m£/cm2-min
rj 1.9 ma/cm2-min
l
I I
J I
23456
GAS VELOCITY, m/sec
Figure 5-6. Experimental penetration for vrertical
gas flow up mesh.
73
-------
Pressure Drop
The pressure drop in wire mesh is highly affected by
the liquid load, as seen in Figure 5-7. The slope of the
straight lines on the log-log plot is 1.65; thus "p" can
be represented as a function of u~ . In Figure 5-6,
L/A = 0, represents the dry pressure drop, "AP^ ", through
the mesh. For 0
-------
3.0
1.0
0.5
a 0.3
E
L)
0.1
0.05
0.03
A / -
A
12 5 10
GAS VELOCITY, m/sec
Figure 5-7. Pressure drop in wire mesh
versus horizontal gas velocity
with liquid load as parameter.
•£ 3
s
O
CL,
O
a:
n 2
UJ
OS
3
CO
to
a:
a.
o
j
f-
S 0
1 1 /\
/
.'
/
/-
/
'
/ 0 0°G° -
a / 8 ° 8
So0®*.
SO 1 2 3 4
UJ
2 PREDICTED DRY PRESSURE DROP , cm K.C.
Figure 5-8. Comparison between experimental and
predicted dry pressure drop for mesh.
75
-------
.01
I
1 5 10
SUPERFICIAL GAS VELOCITY, m/sec
Figure 5-9. Pressure drop in knitted mesh versus
vertical gas velocity with liquid load
as parameter.
L
T- = Superficial liquid velocity, cm/min
GAS VELOCITY, m/sec
Figure 5-10. Outlet drop diameter for mesh separator
with horizontal gas flow.
76
A
-------
OH
a
Q
W
oo
10
I
Inlet Drop Diameter, ym
A 84
O 380
D 1,225
I I
1.4 1.8 2.2
GEOMETRIC STANDARD DEVIATION
2.6
Figure 5-11. Drop diameter versus geometric standard
deviation for mesh.
77
-------
information presented in Figures 5-10 and 5-11 can be
employed to make a proper selection of the device.
Figure 5-12 shows the effect of liquid to gas ratio
on gas velocity for onset of reentrainment. The shaded
area is the region where reentrainment was observed. Thus,
the boundary line relates entrainment loading to maximum
permissible gas velocity through mesh without causing re-
entrainment .
Figure 5-13 compares the reentrainment onset velocity
obtained in the present study with Poppele's data. The pre-
sent study observed a higher reentrainment than the flooding
velocity observed by Poppelle.
Buerkholz (1970) collected reentrainment data for sul-
furic acid mist 150 cm downstream of a mesh separator.
He found that reentrainment increased from 1.6 to 4.0% of
collected liquid (0.3 to 1.3 mg/m3) as the gas velocity
was increased from 4.7 to 8.2 m/sec. The outlet mass
median drop diameter also increased from 150 to 750 ym.
Buerkholz' data, plotted in Figure 5-13, were collected
on a 15 cm x 15 cm mesh with sedimentation present between
the mesh and the sampling point. The solid line in Figure
5-14 is the onset of reentrainment curve obtained in the
present study. The data show good agreement in deter-
mining the reentrainment velocity of 5 m/sec at very
small liquid loads .
The reentrainment curve obtained from the manufac-
turer also appears in Figure 5-14. The manufacturer
predicts higher reentrainment velocity than the present
results.
Visual Observation of Reentrainment
Reentrainment in the mesh section was observed to
take place in the following ways:
A
-------
1x10
1x10
3x10
I I I I I
A Some reentrainment
Reentrainment in part of
'-' duct only
_ O No reentrainment
No reentrainment
O
O
012345678
GAS VELOCITY, m/sec
Figure 5-12. Effect of gas velocity and liquid
load on performance of mesh.
1.0
- 0.5
1.1
0.05
0 .001
- incipient ^~
- Flooding,
Poppelle (1958)
I i I i i i i i i
II I I 1 I T
Present study Stacked
.onset of rings
reentrainment
Dumped
rings,
Sherwood (193f
i i i i i i i i
0.001
0.01
0.1
G PL
Figure 5-13. Effect of entrainment load on reentrainment onset velocity
79
-------
10
s
m
S
•t
O
I-H
H
Pi
CO
CJ
o
cr
i—i
j
10'
10
—— Experimental data
— —Manufacturer's
catalog
O Reentrainment data1
observed by
Buerkholz(19701
GAS VELOCITY, m/sec
Figure 5-14. Onset of reentrainment velocity
curves of mesh for horizontal
gas flow.
80
A
-------
At low liquid loads the mesh operated without
flooding. Apparently, the drops that are collected
on the mesh wires grew to 3-5 mm diameter before
they drained down. If the air velocity is high,
the path of the drop is not vertically downward.
Some of these drops were airborne and struck the
wires of the grid supporting the mesh.
Normally, the drop shattered into one large drop
slightly smaller than the original size and 2-4
satellite drops, which were reentrained.
Some of the drops collected on the grid wire
drained at once, whereas the rest drained after
growing to a larger size. There were other drops
which missed striking any wires and emerged from
the mesh. These drops were collected at the down-
stream side of the mesh. The drops passing through
the mesh without striking any wires were carried
farther downstream of the mesh than others. All
these reentrained drops were 4-5 mm in diameter
and upon reaching the bottom, they shattered into
a few (3-4) satellite drops. The rest of the li-
quid in the original drop was mixed with the liquid
film at the bottom. These satellite drops flew
into the air due to kinetic energy, and their
initial trajectory formed a cone along a vertic.al
axis. The angle of the cone was dependent upon
initial drop velocity and was observed to range
from 0° to 90°. Some of these satellite drops
were reentrained while others fell down.
Some drops were reentrained inside the mesh, and
the process of reentrainment could not be observed;
it is assumed to be the same as described in the
earlier part of the first method.
81
-------
3. When the liquid load was high, partial flooding
was observed. Reentrainment by methods 1 and 2
took place above the flooded zone. In the flooded
section the air flow rate was low. The flooded
section was partially covered by the falling drops
from above on the downstream side of the mesh.
The reentrainment mechanism was rupture of bubbles,
but it could not be observed properly.
CONCLUSIONS
Based on the data obtained above, the following conclu-
sions can be drawn:
1. Bradie and Dickson's expression in predicting primary
efficiency agrees quite well with the experimental
data.
2. Pressure drop data can be correlated by the ex-
pression
1.65
AP = a., un
I b
where "a" is a constant dependent on "L/A".
3. Pressure drop data are comparable for both hori-
zontal and vertical air flows.
4. Reentrainment velocities are lower in the system
with horizontal gas flows than with vertical gas
flows.
5. The amount of reentrainment is higher in the system
with vertical gas flow than with horizontal gas
flow.
82
A
-------
CHAPTER 6
PACKED BED
Packed beds of standard design with a capacity of
up to 65 m3/sec (140,000 CFM) are available. They can
remove drops as small as 3 ym in diameter at 80-901
efficiency. Superficial gas velocities range from 75 to
250 cm/sec, and pressure drop is generally low, 0.05 -
0.1 cm W.C. per cm of bed length.
Cross flow beds are claimed to have high drainage
efficiency and therefore are less prone to plugging. Up-
stream washing is recommended to avoid plugging if solids
are present in the drops to be removed.
Packing in different materials, shapes and sizes is
available. Various rings are claimed to have high col-
lection efficiency and low pressure drop.
Packed beds are often used for mass transfer because
of their high interfacial area. Thus they are sometimes
employed when simultaneous mass transfer and entrainment
separation are desired.
MATHEMATICAL MODELS
Primary Efficiency
Jackson and Calvert (1966) and Calvert (1968) have
developed a theoretical relationship between particle
collection efficiency and packed bed operating parameters
Their formulation included the following assumptions:
1. The drag force on the drop is given by Stokes
Law.
-------
The number of semicircular bends, "n^", is
related to the overall length, "Z", of the
packed section of the bed, the packing dia-
meter, "d ", and the channel width, "b",
where any consistent units may be used, by:
Z
The gas velocity through the channels, u^, is
inversely proportional to the free volume of
the bed available for gas flow, where any con-
sistent units may be used:
u
Gb
= u,
where u~ is the superficial gas velocity of
the bed (volumetric flow rate divided by total
cross sectional area of the shell), "e" is the
bed void fraction (porosity), and "H," is the
liquid holdup within the bed, i.e. the fraction
of the total bed volume taken up with liquid.
Table 6-1 lists values of bed porosity, e, for
beds using various packing materials.
4. The width of the semicircular channels, b, can
b"e described as a fraction, j, of the diameter
of a single packing element:
b = j dc
These assumptions lead to the following equation for pre-
dicting the particle penetration for a packed bed.
Pt = 1 - exp
- IT
2(j + D (e - Hd)
K - P* ^/G
P 9 He d.
_Z_
c
(6-1)
84
A
-------
where j = ratio of channel width to packing diameter
H, = fractional liquid hold-up in the bed
e = bed porosity
Z = bed length, cm
d = packing diameter, cm
ur = superficial gas velocity, cm/sec
dj = drop diameter, cm
The experimental data of Jackson (1964) were analyzed
to determine appropriate values of "j " to use in Equation
6-1 with all quantities in the equation known except " j " ,
which was calculated. The results are given in Table 6-2
which lists "j" values for various types and sizes of pack-
ing material. For the manufactured packing materials, "j"
is fairly constant at about 0.16 - 0.19. The very low
value of 0.03 for coke may be due to the small passages
within the coke itself, which make each large piece of
coke function effectively as' a number of smaller pieces.
Pressure Drop
Perry (1963) gives a generalized pressure drop and
flooding correlation plot which appears as Figure 6-1,
2
where a dimensional group of function
(centipoise) 0-2 , is plotted against a dimensionless group
T /pr
of function £• I — ) , where "G" and "L" refer to the gas
b \PL /
and liquid mass flow rates respectively. "1"' is the ratio
of water density to entrained liquid density. Values for
the packing factor, "F", for dumped pieces, stacked pieces
and grids are given in Tables 6-3 and 6-4. If "F" is not
o
known, — 3 may be used instead.
85
-------
TABLE 6-1
BED POROSITY, e, FOR VARIOUS PACKING MATERIALS
Name
Size
(cm)
1.27
1.9
2.54
3.8
5.1
Stoneware
Raschig
Rings
0.57*
0.67
0.68
0.68
0.75
Carbon
Raschig
Rings
0.71*
0.75
0.67
—
Steel
Raschig
Rings
(1/16"
thick)
0.92
0.92
—
Stoneware
Berl
Saddles
0.65
0.69
0.70
—
Stoneware
Intalox
Saddles
0.70
0.81
—
Steel
Pall
Rings
0.95
0.94
—
*Treyball (1955)
All other data from Perry (1963)
TABLE 6-2
EXPERIMENTAL VALUES OF
j, CHANNEL WIDTH AS FRACTION OF PACKING DIAMETER
Size (cm)
1.27
2.54
3.8
7.6 - 12.7
Type of Packing
Berl Saddles, marbles, Raschig Rings,
Intalox Saddles
Berl Saddles, Raschig Rings,
Pall Rings
Berl Saddles, Raschig Rings
Pall Rings
Coke
j
0.192
0.190
0.165
0.03
Adapted from Jackson (1964) and Calvert (1968)
86
A
-------
Table 6-3. PACKING FACTORS, "F", FOR DUMPED PIECES (m2/m3)
Nominal size of packing, cm
[0.64] [0.95] [1.27] [1.59] [1.9] [2.5] [3.2] [3.8] [5] [8] [10]
Raschig rings,
ceramic
.16 cm wall
.32 cm wall
.63 cm wall
.95 era wall
Raschig rings,
carbon
.16 cm wall
.32 cm wall
.63 cm wall
.79 cm wall
Raschig rings,
metal
.08 cm wall
.16 cm wall
Lessing rings,
porcelain
.52 cm wall
.63 cm wa11
Less ing rings ,
metal
.OS cm wall
.16 cm wall
5,250 3,280
5,250
2,300 1,280
510
430
210
121
98
1,340
920
525
430
210
118
980
,340
560
950
510 380
720 450 360 272 187 105
(800)
(360)
(1,060)
(630)
(472) (387) (295) (200)
Table 6-3. PACKING FACTORS, "F", FOR DUMPED PIECES (m2/m3) (continued)
\ominal size of packing, cm
[0.64] [0.95]
Partition ring's .
Pall rings,
plastic
Pall rings,
metal
Berl saddles 2,950
Intalox saddles, 2,380 1,080
ceramic
Intalox saddles,
plastic
Super- Intalox,
cerami c
Tellerettes
[1.27] [1.59] [1.9] [2.5] [3.2] [3.8]
318 171 105
230 158 92
790 560 360 213
660 475 322 171
108
200
[5] [8] [10]
262 190
82
66
148
131 72
69 52
100
150
Parentheses denote a value of a/E3, rather than empirical F.
87
-------
Table 6-4. PACKING FACTORS,"F" FOR GRIDS AND STACKED PIECES
(m2/m3)
Nominal size of packing, cm
2.5 3.8 5 8 10 13 14
Wood grid 20 11 8.2 5.9 4.9
Metal grid 8.2
Grid tiles 118
Checker brick,
e=0.55 135
Raschig rings ,
ceramic
.63 cm wall 95 16
.95 cm wall 36 12.8
Raschig rings, 21
metal
Partition rings,
diameter
7.6 cm length (1,200) (725)
10.2 cm length (705)
15.2 cm length
Partition rings,
square set
7.6 cm length (690) (460)
10.2 cm length (450)
15.2 cm length
15
(410)
(375)
(275)
(260)
Parentheses denote a value of a/e3, rather than empirical F.
A
-------
The operation of packed beds is limited by flooding.
The flooding lines for dumped pieces, grids and stacked
rings are shown in Figure 6-1. Pressure drop should be
obtained by using the largest gas and liquid streams.
EXPERIMENTAL RESULTS AND DISCUSSION
Overall Efficiency
Efficiency data for horizontal gas flow through 30 cm
of 2.5 cm Pall rings are presented in Figure 6-2. Runs at
superficial gas velocities lower than 6.0 m/sec did not
show any penetration. There were negligible reentrainment
when gas velocity was higher than 6m/sec. The theory for
primary collection efficiency, shown as a solid line, is
based on equation 6-1 and predicts 1001 primary efficiency
over the range of gas velocities studied.
Overall efficiency data for vertical gas flow is plot-
ted in Figure 6-3. Inlet liquid loading is used as a para-
meter. The mass median drop diameter of the inlet entrain-
ment is 84 ym. It can be observed that the rate of reen-
trainment is increased as liquid loading is increased.
Heavy reentrainment started at a gas velocity of 6m/sec.
The bed when installed in vertical direction (i.e.
horizontal gas flow) has higher collection efficiency than
the bed installed horizontally. This is because the verti-
cal bed has a higher liquid drainage capability.
Pressure Drop
Figures 6-4 and 6-5 show the dry and wet pressure drops
respectively for horizontal gas flow. There is no effect
of liquid load on pressure drop for the liquid loading used
in the present study. Figure 6-6 shows the wet pressure
drop data for horizontal and vertical gas flows. It can
be observed that the gas flow orientation has little effect
on pressure drop.
89
-------
a.
u
_j
a
0.5
0.2
0.1
0.05
cT* 0.02
u
1 0.01
>,
0.005
0.002
0.001
0.01 0.02 O.OS 0.1 0.2 0.5 1
•y ^ -
PGI.V*
10
(dimensionless)
Figure 6-1 - Generalized flooding and pressure drop
correlation for packed beds (Perry, 1963) .
90
A
-------
100
(_J
2
PJ
rj
" 80
tu
U4
2
0
i — i
f-H
w 60
jj
o
u
1 1 i
*-"*> fT\ w* rr"i , (Trf~i--
-v^
o
V
Inlet Drop -
Diameter, ym
V 84
O 380
D 1,230
O>1,230
1 1 1
02468
GAS VELOCITY, m/sec
Figure 6-2. Experimental collection efficiency in
packed bed, horizontal gas flow, Pall
rings.
100
UJ
2
C
i—i
H
v—/
-4
O
U
80
60
50
Figure 6-3
-KD br
^-/~TJ
O 4.7 mjL/cm2 -min
D 2.3 nu/cm2-min
I ( | I
12345678
GAS VELOCITY, m/sec
Collection efficiency in packed bed,
vertical gas flow, Pall rings.
91
-------
I I
I I I I 7 | I
I I I I I I I
<-
O
O
Q;
U
O
'3'M uio '
1/1
c c
•H -H
Ji
P.
O r->.
•t) 03
a,
3 -0
ft CD
(/> _Q
0)
S-i T3
Q, CO
^4
+-> O
(U rt
vO
0)
00
•H
I I I I I I I
O
cdV
OJ
O
O-
•j
0)
I/)
u
O
LO
C G
•i-l -H
Si
a.
o --H
>-i -H
-r) rt
CLi
QJ
f-< •>
3 T3
w 0)
10 ^3
(I)
^ t3
a.
-------
S
o
20
10
0.5
0.4
0.3
0.2
0.1
^T
-O
O vertical gas flow
\ A horizontal gas flow
l
I I I I
1 234 10
SUPERFICIAL GAS VELOCITY, m/sec
Figure 6-6. Wet pressure drop in packed bed,
Pall rings .
93
-------
Figure 6-7 compares the measured pressure drops with
those predicted by the generalized correlation. As can
be seen, the predicted pressure drop is higher than that
measured in the present study. Thus the generalized cor-
relation will give a conservative design.
Reentrainment
Reentrainment was observed to start when the bed be-
came flooded. In the present study, for a liquid loading,
L/C, ranging from 10 to 10 , reentrainment was observed
to start at a superficial gas flow of 6 m/sec for a bed of
2.5 cm Pall rings when the bed was operated in cross flow.
This reentrainment onset gas velocity is higher than the
flooding velocity calculated from the generalized flooding
correlation. Possibly this is due to cross flow bed offer-
ing better drainage capability. Figure 6-8 shows the cor-
relation of reentrainment velocity along with flooding
lines for dumped pieces, stacked rings and drip-point grid.
CONCLUSION
The following conclusions are drawn based on the above
experimental results for a packed bed:
1. The model developed by Jackson and Calvert
agrees well with experimental data in pre-
dicting primary collection efficiency.
2. The effect of gas flow orientation on over-
all efficiency in the packed bed is not
significant.
3. Neither liquid load nor gas flow orientation
has any significant effect on pressure drop
provided there is good drainage.
4. The gas velocity for the onset of reentrain-
ment from a cross flow bed of 2.5 cm Pall
94
A
-------
a.
o
Of!
\a
oi
uj
a:
CL,
X
U4
0
I I I I I I I
/
/ J
u Vertical gas flow
A Horizontal gas flow x
/
/
/ A
f
O
A-
/
/ O
O
P A
i i
i
02468
PREDICTED PRESSURE DROP, cm W.C.
Figure 6-7. Experimental versus predicted pressure
drop across 30 cm of 2.5 Pall rings.
1 .0.
0. 5
0.1
0.05
Cross flow,
Onset of Reentrainment
Stacked
Rings
Flooding Lines
Dumped Pieces
I I I I I i I I
I I K I I I i I
1.005 0.01
0.05 0.1
0.5 1.0
Figure 6-8. Correlation for onset of reentrainment in cross flow beds.
95
-------
rings is 6 m/sec, and is not affected by liquid
load.
Generalized pressure drop correlation (Figure
6-1) predicts a higher pressure drop across the
bed than that measured in this study.
96
A
-------
CHAPTER 7
TUBE BANK
Tube banks made of streamlined struts have been used
as entrainment separators but no experience with round
tubes has been reported. Particle collection efficiency
and pressure drop for round tube banks have been studied
and the characteristics appeared promising for entrainment
separation application. Therefore, the performance of
tube banks for use as entrainment separators was chosen
for study as a possible basis for the development of im-
proved devices
MATHEMATICAL MODELS
Primary Efficiency
Calvert and Lundgren (1970) found that the collection
efficiency for closely packed rods is given by the equa-
tion for rectangular jet impaction. The collection effici-
ency of each stage of impaction can be found in Figure 7-1.
Each row of tubes except the first represents one stage of
impaction. "8" is used as a parameter in Figure 7-1 and is
defined by:
0 = 2
where b = jet orifice width
£ = distance between orifice and impingement plane
"Kp", the inertia parameter, is defined with drop radius,
"r ", rather than diameter as in Figure 3-3.
Efficiency for the bank of tubes is given by:
E = 1 - (1 - n.)N (7-2)
97
-------
1.0
o
I—I
H
U
<
OJ
Hi
0.5
O.
UJ
H
u
O
U
:er q Chow
Exp.
is Theory
0.5
1.0
u
Figure 7-1- Theoretical and experimental
collection efficiencies of
rectangular aerosol jets.
1.5
98
A
-------
where n. = collection efficiency for a given particle
diameter in one stage of rectangular jet
impingement
N = number of stages in the tube bank
= (number of rows) - 1
If the tubes are widely spaced, the target efficiency,
"n", can be calculated from Figure 3-6. In this case the
efficiency for the entire tube bank is:
where a' = cross-sectional area of all the tubes in
one row
A = total flow area
n = number of rows
Pressure Drop
Pressure drop for gas flow normal to banks of round
tubes can be predicted by means of Grimison's correlations
(Perry, 1973). As an approximation, Lapple (Perry, 1973)
suggests that 0.72 velocity heads are lost per row of tubes
in arrangements of the kind commonly used in heat exchangers
Calvert and Lundgren (1970) found that for closely spaced
tube banks Lapple's approximation agreed satisfactorily
with experimentally determined pressure drops.
Houghton and Radford (1939) studied streamline strut
banks and found that for a center-to-center spacing of 2
strut widths (i.e. open space = strut width) the pressure
drop was about 0.16 velocity heads per row. This can be
expressed as:
AP = 0.16 N pG(5.3 x 10~")(u'G)z cm W.C. (7-4)
where u' is the actual gas velocity
99
-------
U'G =
0.1 2 7 1"c
35 6 a L
0.5
0
G
Reentrainment
Ullock (1956) determined the reentrainment velocity
experimentally for streamlined struts. He found that
the reentrainment velocity was a direct function of the
surface tension and specific gravity of the liquid on
the tube and an inverse function of the density of the
gas flowing around the tubes. The empirical equation
for reentrainment velocity was
(7-5)
where the velocity is in cm/sec, "a" the surface tension
in dyne/cm and"pr" and "Pp" are the liquid and gas densi-
ties in g/cm3 .
EXPERIMENTAL RESULTS AND DISCUSSION
Overall Collection Efficiency
Collection efficiency versus gas velocity data for
horizontal flow through tube banks are plotted in Figures
7-2 through 7-4 for various inlet drop diameters. Pene-
tration due to primary efficiency of less than 100% was
observed for velocities lower than 3.0 m/sec. Figure 7-5
is a plot of overall collection efficiency versus gas
velocity data for vertical flow through tube banks. Liquid
load was used as a parameter. It can be observed that the
onset of reentrainment velocity is as low as 3 m/sec.
Houghton and Radford's (1938) data for strut separa-
tors are also plotted in Figure 7-3. They found a con-
stant collection efficiency of 96.2% for gas velocities
from 1.25 to 17.5 m/sec. The inlet entrainment contained
drops as small as 1 ym. However, no increase in penetra-
tion at lower velocities or reentrainment at higher velo-
100
A
-------
so
rr 60
40
23456
GAS VELOCITY, m/sec
Collection efficiency versus gas velocity
in tube bank with n = 6, d =84 :^. and
o = 1.32. PS
100
40
o o
Q A
Houghton 6 Radford Data (1938)
O Water loading 17.1 £ / m i n
A Water loading 6.8 l/min
p Water loading 11.4 t/min
• Water loading 3.8 £/min
i I _ I _ I
01 23456
GAS VELOCITY, in/sec
7 8
Figure 7-3. Collection efficiency versus gas velocity
in tube bank with d = 580 -jm and - = 1.
101
-------
100
~ 60
o
H 40
a
O
Inlet Drop Diameter, ;jm
V 84
<> 380
O 1,230
O >1,230
I I I I
01
GAS VELOCITY, m/sec
Figure 7-4. Collection efficiency versus gas velocity
in tube bank. Solid line represents theory
I 1 T
100
60
40
20
Water Flow Rate
- O 4.7 i/min
CH 9.4 1,/min
O 27.7 H/min
I I I I I I I
GAS VELOCITY, m/sec
Figure 7-5. Collection efficiency versus gas velocity
in vertical direction in bank of tubes.
102
A
-------
cities was observed. A comparison between the configura-
tion used in the present study and that of Houghton and
Radford is given in Table 7-1.
Pressure Drop
Dry and wet pressure drops through the tube bank are
plotted in Figures 7-6, 7-7, and 7-8. There is little
effect of liquid load and air flow orientation on the
pressure drop although there is an increase of pressure
drop with gas velocity for vertical gas flow. This is
in keeping with the increase in liquid holdup which would
be expected. The pressure drops are about 1.0 velocity
head for 6 rows, which is in agreement with equation (7-4).
Thus, for standard air the experimental dry pressure drops
are given by:
-7
AP - 1.0 x 10 N^'G)2 cm W.C. (7-6)
Reentrainment
Figure 7-9 depicts the value of gas velocity and
liquid load observed as being necessary for reentrainment
from tube banks in cross-flow arrangement. The shaded
area is the operating condition at which drops were ob-
served to tear off the tube by the gas, i.e. reentrained.
However, most of these reenetrained droplets settled out
in the observation section ahead of the sampling point.
Below the shaded area, reentrainment was not present.
Above the shaded area, although reentrainment was detected
at the sampling point (90 cm from the separator) its quan-
tity was negligible. Heavy reentrainment started at a
superficial gas velocity of 7 m/sec. This velocity did not
depend on the liquid loading. This velocity is lower than
that predicted by equation 7-5 (8.9 m/sec based on this
103
-------
Table 7-1 . COMPARISON OF TUBE BANKS
Present
Study
Houghton §
Radford (1938)
Tube (minor-major
axis) diameter, cm
Spacing between
tubes, center to
center, cm
Spacing between
rows, center to
center, cm
Number of rows
Material of
tubes
1.9
3.8
3.3
6
Al
1.25x3.2
2.5
104
A
-------
2.0
1 .0
1.5
0 .2
§ 0.1
.05
.02
.01
o
J I I I I I I
1 510
GAS VELOCITY, m/sec
Figure 7-6. Dry pressure drop in tube
bank versus gas velocity.
2.0-
1 .0
0.5
0.2
§0.1
.05
.02
.01
O
o -
I L I I I I I I
Figure 7-
1 5 10
GAS VELOCITY, m/sec
7. Wet pressure drop in tube
bank versus gas velocity
105
-------
10
5.0
4.0
3.0
2.0
1.0
0.5
0.4
0.3
~? 0.2
; o.i
0 .05
0.04
0.03
0.02
0.01
A
G
A A O
O
O
O vertical gas flow
A horizontal gas flow
I i i i I i l I
1 10
SUPERFICIAL GAS VELOCITY, m/sec
Figure 7-8. Wet pressure drop in tube bank
106
A
-------
equation). This is probably due to the difference
in shades of the tube used.
Figure 7-10 is a similar plot for vertical gas flow.
This graph gives, at a given liquid load, the gas velocity
at which reentrainment increased sharply.
Figures 7-9 and 7-10 indicate that at a given gas
velocity, due to its better drainage characteristic,
vertical tube with horizontal gas flow can handle a higher
liquid load.
CONCLUSIONS
1. Experimental primary efficiency agrees with
the theory.
2. Heavy reentrainment in vertical tube banks
using horizontal gas flow starts around
7 m/sec. Liquid load does not have a signi-
ficant effect on this velocity. However,
the onset of reentrainment velocity of tube
banks with vertical air flow is highly de-
pendent upon the liquid load. Reentrainment
starts at a gas velocity of as low as ? m/sec
3. Pressure drop is predictable by means of
correlations available from the published
literature relating to heat exchanger tube
bundles.
107
-------
< 1x10
Some reentrainment
Reentrainmen
in part duct
onl>
O N'o penetration
Openetration due
to less than 100'
primary efficiency
I
i
Figure 7-9.
3 4 5 ti
GAS VELOCITY, m/sec
Hxperimental results showing the
effect of gas velocity and liquid
load on performance of tube bank
in cross-flow pattern.
10"
10"
10"
I
1 l l
1 5 10
GAS VELOCITY, m/sec
Figure 7-10. Experimental results showing the
effect of gas velocity and liquid
load on reentrainment for tube banks
with vertical gas flow.
108
A
-------
CHAPTER 8
CYCLONE
Commercially available cyclones in standard designs
for entrainment separators have a maximum capacity of up
to 141 m3/sec (300,000 CFM) of gas. Efficiencies of about
951 are claimed for 5 \im diameter drops in a well-designed
cyclone. Some manufacturers use a bundle of small cyclones
(multicyclones), which can efficiently collect drops as
small as 2 ym in diameter. However, this arrangement re-
duces the capacity of the device.
MATHEMATICAL MODEL
Primary Efficiency
Leith and Licht (1971) derived an equation to predict
primary collection efficiency in conical bottom cyclones
as pictured in Figure 8-1. With slight modification it can
be applied to cylindrical cyclones. The following assump-
tions were made:
1. The drag force in the radial direction on the
drop is given by Stokes law.
2. The tangential velocity component of the drop
is related to the radial position by a modified
form of the equation for a free vortex in an
ideal fluid:
ut rn = constant (8-1)
where "r" is the distance from the vertical axis
of the cyclone and "n" is the vortex component
and is defined below in equation (8-3).
3. Backmixing of the drops takes place in the gas
phase .
109
-------
The final equation for predicting primary collection
efficiency is:
In Pt = - 2
ld utg
dc
0.3
where n = l -
(
0.393
o.i
Pt
p,
u
t
T =
1_
2n+l
(8-2)
(8-3)
penetration, fraction
drop density, g/cm3
gas viscosity, poise
drop diameter, cm
tangential velocity, cm/sec
mean residence time of the gas in the
cyclone, sec
gas temperature, °K
The mean residence time of the gas stream in the
cyclone is:
where
t =
A =
QG
effective volume of the cyclone, m3
volumetric gas flow rate, m3/sec
inlet area, cm2
inlet gas velocity, cm/sec
(8-4)
The effective volume of the cyclone, "Ve" is defined as the
volume of the cyclone minus the volume occupied by the exit
duct and exit gas core. The diameter of the exit gas core
can be assumed equal to the diameter of the exit duct.
Leith and Licht (1971) gave the following equations for the
110
A
-------
determination of effective volume of a conical bottom
cyclone :
V = Vi+V2 (8-5)
e
where Vi = annular shaped volume above exit duct inlet
to mid-level of entrance duct
= JCS-f)Cd*-dJ) (8-6)
V2 = volume of cyclone below exit duct inlet
to the natural length of the cyclone
*- ("U C *\ i 3 C / -I . U , U \ C f n T\
v2 = — (Vs) + — n - ( 1+dc +d2J — T- (8-7)
/S+L-h \
where d - dQ - (dQ -b£) f ^- (8-8)
and L = natural length of the cyclone
a, b, d , d , S, h, h are cyclone dimensions defined
L* t.. o
in Figure 8-1. Figure 8-2 shows the theoretical grade
efficiency curve for the cyclone used in the present
study with inlet gas velocity as parameter.
Prsssure Drop
Shepherd and Lapple (1940) derived an equation for
a cyclone with inlet vanes for pressure drop as a function
of inlet gas velocity and the cyclone inlet and outlet
dimensions :
AP = 0.00513 G (8-10)
111
-------
Figure 8-1. Cyclone with tangential gas inlet,
112
-------
1.0
0.5
I I I l I I I I I 1 I I I l I M_
z 0.1
u
•s
o
I—I
C-H
<
0.05
0.01
0.005
0.001
i vi l i i v i i i
5 10
DROP DIAMETER, ym
50 100
Figure 8-2
Theoretical grade efficiency curve of
the cyclone used in the present study
with inlet gas velocity as parameter.
113
-------
where p,, = gas density, g/cm3
Qr = gas volumetric flow rate, cm3/sec
a = cyclone inlet height, cm
b = cyclone inlet width, cm
d = cyclone exit pipe diameter, cm
t
Equation (8-3) can be modified by writing it as a function
of the geometric average of the gas velocity at the cyclone
inlet and outlet:
AP = 0.00513 Pv (8-11)
Shepherd and Lapple also developed an equation for a cyclone
without inlet vanes:
AP - 0.00513 PG ^}feM (8'12)
Reentrainment
Onset of Reentrainment - There is a great disagreement
among results for the onset of reentrainment obtained by
different investigators. This is indicative of the problem
of defining the onset of reentrainment. Zhivaikin (1962)
defined the onset of entrainment as occurring when it is
first detectable. Steen and Wallis (1964) defined the on-
set of entrainment as that air velocity which represents
the extrapolation of the straight line portion of a graph
of entrainment percentage versus air velocity. Since the
increase in entrainment with air velocity is similar to
the exponential function, their results lie considerably
above those of Zhivaikin. Chien and Ibele (1962) defined
the transition on the basis of pressure drop versus gas
flow rate curves. A change in the slope of the curve was
114
A
-------
200
olOO
tu
IT:
u
o
_4
UJ
0 10
t—'
z;
E-M
1 5
1.0
I I I I I I I
O Present study (open channel, 6.5 cm
0° angle)
1. Chien § Ibele (1962)
2. Steen § Wallis (1964)
5. Zhivaikin (1962)
Figure 8-.
i ill
-2.54 cm tube
i l i i i I I
103 10"
LIQUID REYNOLDS NUMBER
Comparison of entrainment onset
velocity by different investigators
115
-------
taken to indicate the onset of gross entrainment. Their
results apply to conditions where a large fraction of
the total liquid flow is entrained.
In view of our need for a correlation for onset of
entrainment, a small scale open channel experiment was
carried out to make observations of the transition from
separated flow to separated-entrained flow. Details of
the experimental set-up and sampling method were presented
in the "Initial Report". It was found that entrainment
velocity depends upon liquid Reynolds number. Experi-
mental data are shown in Figure 8-3 along with other in-
vestigators' results. As can be seen, the present results
are comparable with those of Chien and Ibele's data for
two phase flow in a 2.5 cm diameter tube. Since the liquid
flow in a cyclone can be approximated by open channel flow
with channel width equal to inlet height. Therefore, Chien
and Ibele's line in Figure 8-3 could be used to predict
the reentrainment velocity for a cyclone with liquid
Reynolds number defined as
NRe,L = £-£- (8-13)
o L
where QT , . . n . . , .... . , ,
XL = volumetric liquid flow rate, cnr/sec
u, = kinematic viscosity of the liquid, cm2/sec
d = channel width, cm
= cyclone inlet height
Drop Diameter of Reentrainment - The drop diameter of the
reentrained liquid has a size distribution which varies
with gas flow rate, liquid flow rate, fluid properties and
perhaps pipe diameter. The average drop diameter decreases
with increase in gas flow rate. On the other hand, the
liquid flow rate has only a weak and ambiguous effect.
When the gas velocity exceeds 6,000 cm/sec, high liquid
flow rate has no effect on the drop size distribution.
116
A
-------
Rate of Reentrainment - It is believed that reentrainment
takes place due to penetration of liquid waves into the
turbulent zone of the gas. The amplitude of the waves
increases exponentially with liquid flow rate. Therefore,
reentrainment is assumed to take place in proportion to
expCK^ Npe L) where "K," is a constant.
The rate of reentrainment depends upon gas flow rate,
liquid flow rate and fluid properties. According to
Anderson et.al.(1964) , the rate of reentrainment is approxi-
mately 41 of inlet entrainment for ND , >2,750 and is seen
Ke, L
to increase slightly with NRg G (3.51 for N = 5 x 10\
4°i icr NRe G = 1.6 x 105). Below NR£ L =2,750, the only
data available are for NRg L = 1,150, at which point re-
entrainment is 0.5%.
RESULTS AND DISCUSSION
Overall Collection Efficiency
Figure 8-4 shows the experimental penetration versus
inlet gas velocity. Data were collected with and without
the use of inlet vane in the cyclone. For the case with-
out the inlet vane, the inlet area was 30.5 cm x 15.0 cm
ai.d tne maximum inlet gas velocity was 22 m/sec. M-26
nozzJes were used to generate the drops. When the cyclone
was operated with the inlet vane, the inlet area was
50.';. cm y 7.5 cm and the maximum inlet gas velocity was
61 m/sec. Small garden hose was used to produce the en-
trainmeiit and the maximum liquid flow rate was 1.5 x 10 m3/m3
of gas (11 . 5 gal/MCF) .
In all experiments for gas velocity below 40 m/sec,
collection efficiency was 100%. For gas velocities between
40 m/sec and 60 m/sec, reentrainment was negligible (<0.5%).
Theoretical predictions based on equation (8-2) predicted
100% collection efficiency for all conditions.
117
-------
z
o
CYCLONE INLET AREA
O 50 .5x15 .2 cm2 (no vane)
D 30.5x7.6 cm2 (with vane)
Q E3
- -OOHGKXD D-O-0-^''0^
1 '
_L
_L
J.
0 10 20 30 40 50 60 70 80
GAS VELOCITY IN CYCLONE INLET, m/sec
Figure 8-4. Experimental penetration versus gas
velocity in cyclone inlet with and
without vane.
50
10
o 5.0
1.0
0.5
500
1,000 .
5,000 10,000
GAS VELOCITY IN INLET, cm/sec
Figure 8-5. Experimental dry pressure drop versus
gas velocity in cyclone inlet.
118
A
-------
Pressure Drop
The experimental pressure drop data in the cyclone are
plotted in Figures 8-5 through 8-7. The effect of gas velo-
city on pressure drop is shown in Figure 8-5. Cyclone inlet
width is used as a parameter. The slope of the experimental
pressure drop curves, on log-log graph paper is 2. The effect
of reducing the inlet width of the cyclone is a proportionate
reduction in the pressure drop, i.e., if the cyclone inlet
width is reduced to half, the pressure drop will be reduced
to half provided the gas velocity through the cyclone inlet
is kept constant. For comparison, a straight line for 1
velocity head was also plotted in Figure 8-5.
The effect of volumetric flow rate through cyclone
on pressure drop is shown in Figure 8-6. At a given
volumetric flow rate, pressure drop through the cyclone
inlet increases with reduction in the inlet area.
The effect of geometric average gas velocity in the
cyclone inlet and outlet on pressure drop is shown in
Figure 8-7. All the experimental data falls on a straight
line represented by
AP = 0.000513 p /_\.l 12.8 t-z^l (8-14)
where AP = pressure drop, cm H~0
p~ = gas density, g/cm3
Qr = volumetric flow rate, cm3/sec
a = cyclone inlet height, cm
b = cyclone inlet width, cm
d = cyclone exit diameter, cm
e '
The above equation agrees in form with the pressure
drop equation for cyclone with inlet vane, given by
Shepherd § Lapple (1940), i.e. equation (8-10). However,
119
-------
50
10
i 5.0
0.5
I I ! i I I I
INLET WIDTH, cm
A 15.2 (no vane)
n 11.4
O 7.6
0 3.8
0.1 0.5 l.o 2.0
VOLUMETRIC FLOW RATE, m3/sec
Figure 8-6. Experimental dry pressure drop
versus volumetric flow rate in
cyclone.
INLET WIDTH, cm
A 15.2 (no vane
O 11-4
5.0 -
1.0 _
0.5
500
1,000
5,000 10,000
GEOMETRIC AVERAGE GAS VELOCITY
IN THE CYCLONE INLET AND OUTLET, cm/sec
Figure 8-7. Comparison of experimental pressure
drop data and predicted pressure drop
for cyclone with inlet vane by Shepherd
5 Lapple (1940) .
120
A
-------
predictions by Shepherd § Lappe give 2.7 times higher pres-
sure drop than predicted from equation (8-15).
Reentrainment
From Figure 8-4, it is observed that the onset of
reentrainment gas velocity is between 40-50 m/sec. For
the cyclone, the air inlet duct has a height of 30.5 cm.
It was assumed that all the liquid collected inside the
cyclone flows along the inside surface as a film having
a width of 30.5 cm. During the experiments, the liquid
flow rate was about 810 cm3/sec. Accordingly, liquid
Reynolds number was
N = (4) (810) = 10>600
Re'L (30.5)(0.01)
From Figure 8-3, using Chien and Ibele's correlation for
a liquid Reynolds number of 10,600, the reentrainment gas
velocity is 40 m/sec which agrees with observation.
Seme Observations of Gas-Liquid Flow In Cyclone
It was observed that most of the entrainment was
collected on the cyclone surface near the inlet. The liquid
drained on the cyclone surface as a spiral. It drained
from the top of the cyclone to the bottom during the angu-
lar rotations equal to 2/3 of a circle. The width of the
bend increased with increase in the liquid flow rate. The
width was 50 cm when the liquid flow rate was 8,000 cm3/min
(2.1 gpm) and air inlet velocity was 3,680 cm/sec. Waves,
as shown in Figure 5,were present in the liquid. At the
above flow rate, the wave amplitude was almost equal to
the film thickness (1 - 1.5 mm). A few drops were torn
away from the liquid film at the top and drained down on
the serrated cap on the exit.
At higher liquid flow rates, 6X101* cm3/min (15 gpm)
and the same gas velocity all the inside surface (including
121
-------
top) was covered with water. Liquid drained as jets of
liquid from the corners of the serrated cap on the exit
CONCLUSIONS
The experimental results show that the primary
collection efficiency in a cyclone is approxi-
mately 100%.
Pressure drop data can be correlated by the
equation :
AP = 0.000513 pp - 2.8 * C8-10)
o \ a D/ \ U. Z /
The Chien and Ibele correlation gives a better
prediction of the onset of reentrainment gas
velocity. Thus, the Chien and Ibele curve is
recommended for determining the onset of re-
entrainment in a cyclone.
122
A
-------
CHAPTER 9
ZIGZAG BAFFLES
Baffles can efficiently separate drops greater than
10 ym in diameter, while some of the better designed de-
vices can separate drop diameters of 5-8 ym. Common gas
velocities are 2.0 - 3.5 m/sec, and the pressure drop for
a 6-pass separator is about 2 - 2.5 cm W.C.
The most common baffle shape is zigzag with 3 to 6
passes. These can be fabricated from a continuous wavy
plate or each pass is separated, in which case the sepa-
ration distance is normally smaller than the width of the
baffles. Cross-flow baffles are claimed to have higher
drainage capacity than countercurrent flow baffles.
MATHEMATICAL MODELS
Primary Collection Efficiency
A model to predict primary efficiency was developed
in this study and was presented in the initial report.
Based on turbulent mixing, the primary collection efficiency
of a continuous zigzag baffle section is
exp
u. nw8
57.3 u b tan 9
(9-1)
where n = primary collection efficiency, fraction
u = drop terminal centrifugal velocity, in
the normal direction, cm/sec
u,, = superficial gas velocity, cm/sec
n = number of bends or rows
6 = angle of inclination of the baffle to the
flow path, degrees
w = width of baffle, cm
b = spacing between two consecutive baffles in
same row, cm
123
-------
The drop terminal centrifugal velocity can be deter-
mined by performing a force balance on the drop. The
result is
u
tc =
(9-2)
where d, = drop diameter, cm
p, = drop density, g/cm3
a = acceleration due to centrifugal force, cm/sec2
CD = drag coefficient
pp = gas density, g/cm3
If the drop Reynolds number is low (ND n <0.1),
Ke , u
Stokes' law applies. For this condition, the drag co-
efficient is given by
(9-3)
NRe,D
where NRg D = drop Reynolds number
= dd utc
By combining equations 9-2 and 9-3, we obtain
u . dd pd a (9-4)
tc 18 yG
The acceleration due to centrifugal force is defined by
the following equation
2 (u^)2 2 u2 sin 6
a = £— = ^ (9-5)
w cot 0 w cos3 6
where u' = actual velocity between baffles, cm/sec
Up = superficial gas velocity, cm/sec
124
A
-------
If N' n >0.1, another appropriate drag coefficient
Ke, L)
should be used in equation 9-2. Foust, et al. (1959)
gave a plot of drag coefficient as a function of Reynolds
number in Figure 9-1, which can be used to determine "ut ".
The effect of surrounding drops on the motion of any
individual drop is neglected.
Pressure Drop
Determination of the pressure drop is based on the
drag coefficient, "fD", for a single plate held at an
arule "6" to the flow as presented in Figure 9-2 (Page
and Johanson, 1927) . Neglecting the effect of neighbor-
ire plates, pressure drop may be expressed as:
n 2
AP = I 1.02 x ID"3 £n pr u' A^ (9-6)
i=l L> b b _£
11 2 At
where AP = pressure drop, cm W.C.
A = total projected area of baffles per row in the
direction of inlet air flow, cm2
A = duct cross-sectional area, cm2
The summation is made over the number of rows of
ba '"£les .
The actual gas velocity, "u'", in the baffle section
should be used in Equation (9-6). The actual gas velocity
is related to superficial velocity by
u'G = UG/COS 6 (9-7)
Note that the angle of incidence for the second and subse-
quent rows of baffles will be twice the angle of incidence
for the baffles in the first row.
Rerntrainment
A mathematical model to predict reentrainment in the
zigzag baffles was derived and presented in the "Initial
Report". The models used to predict reentrainment were based
125
-------
10,000
1 ,000
100
10
1
0.1
o.ooi o.oi o.i i 10 100 1000 10,0011 105
Revnolds nurber \'r
Re,n
dd utc r-<:
Figure 9-1. Drag coefficient versus Reynolds number after
Foust et al (1959), with sphericity >'.; as
the parameter.
_-0.8
z:
U.
00.4
u
I I I I I I
20
I \
Plate Inclined to Flow
Angle of
Incidence
I I i I
40
60
80
ANGLE OF IXCIDENCE, degrees
Figure 9-2,
Drag coefficients for flow past inclined
flat plates (data from A. Page q F.C.
Johansen, (1927) .
126
A
-------
on the assumption of film flow and it predicted a lower
reentrainment velocity than that observed in the experi-
ments. In the experiment with zigzag baffles, it was
observed that the liquid flow on the baffles was dropwise.
Therefore, an attempt was made to predict reentrainment
due to tearing off drops from the baffle edges.
An additional factor to consider is that not all of
the reentrained drops appear in the outlet. This is due
to their settling out in the distance between entrainment
separator elements and the outlet where sampling is done.
The effect of gravity therefore reduces the amount of re-
entrainment measured at the sampling point.
Consider a drop hanging on the baffle edge prior to
reentrainment. The necessary condition for the drop to
be torn off from the baffle with vertical gas flow is
when the drag force due to gas flow is balanced by gra-
vitational force and surface tension effect, i.e. when
CT pG (u')2\^ d2
M ; x ~ dd PT § (9'6)
2 /\ 4 / * 6 d L
where C_, = drag coefficient
G = gas density, g/cm3
u' = gas velocity for onset of reentrainment,cm/sec
d, = drop diameter, cm
d, = drop attachment length, cm
G = surface tension of liquid, dyne/cm
p = drop density, g/cm3
g = gravitational acceleration, 980 cm/sec2
Solve equation 9-6 for "u^", to obtain
0.5
(9-7)
llr
UG
"16 o d .
7T C- D
4
-j-
d i 3
Q
dd
PG
PLg 1
CD
127
-------
If the drop attachment length is assumed equal to drop
diameter, i.e. d = d,, then the expression for reentrain-
ment velocity becomes
d
0.5
UG =
16 a
4 d
CDPG dd
D
(9-8)
In reality, the drop oscillates due to drag forces exerted
by gas flow and the drop attachment length may not be equal
to one drop diameter. Another consequence of drop oscilla-
tion is that the drop shape is not spherical and may be
quite "flat", such that the form drag area is increased.
For the case of horizontal gas flow, the reentrainment
condition is that the drag force has to overcome the sur-
face tension effect
D
= 2a d
(9-9)
Rearranging equation 9-9, we obtain for reentrainment
velocity
16 a dn
UG'
11 PG CD dd
0.5
(9-10)
The reentrainment velocity predicted either by equation
9-7 or equation 9-10 is the actual gas velocity in the zigzag
baffles. Depending upon the angle of baffles, the superficial
reentrainment velocity is lower than the actual velocity.
Superficial reentrainment velocity is related to actual re-
entrainment velocity by
ur = u' cos 0 (9-11)
b b
The value of the drag coefficient, "CD", depends upon
the drop Reynolds number and the sphericity of the drop.
128
A
-------
The drop Reynolds number depends upon the gas velocity
relative to the drop. Thus, equations 9-7 and 9-10
shoula be solved by trial and error method.
The predicted superficial reentrainment velocity due
to tearing of drops is shown in Figures 9-3 and 9-4 for
vertical gas flow and horizontal gas flow respectively
with drop attachment length and baffle angle as parameters.
The drop sphericity factor, "i(>M, is assumed equal to 0.6.
Drop sphericity is defined as the ratio of the surface area
of a sphere of same volume as the drop to the surface area
of the drop. As mentioned earlier, not all drops that are
torn off the baffle are reentrained. For the case of verti-
cal gas flow, only those drops with settling velocities
smaller than the upward gas flow will be carried away by
the gas as reentrainment. Curve 3 in Figure 9-3 shows the
drop terminal settling velocity. If the reentrainment velo-
city lies above curve 3, the drops will be carried away.
Thus, the lowest reentrainment velocity detectable will be
at the point where the reentrainment and curves intersect.
For the baffle test section used in the present study, the
lowest detectable reentrainment velocity will be 5 m/sec
if the drop attachment length is equal to half the drop
diameter and will be 6 m/sec if drop attachment length is
one drop diameter.
When the gas velocity is horizontal, some of the drops
that tear off will be settled out due to sedimentation be-
tween the entrainment separator element and the sampling
point. Curve 4 in Figure 9-4 gives the maximum drop dia-
meter that may be sampled in the pilot plant of the pre-
sent study. The vertical height = 60 cm and horizontal
distance = 90 cm, are used to obtain curve 4. If the re-
entrainment velocity lies above curve 4, the drop will be
present at the sampling point. Figure 9-4a shows the pre-
dicted lowest detectable reentrainment velocity and maximum
129
-------
drop diameter as a function of drop sphericity factor for
baffle section with horizontal flow. Drop attachment length
was assumed equal to half the drop diameter.
The agreement between the predicted reentrainment onset
velocity and the experimental reentrainment velocity was not
known since the drop attachment length and drop sphericity,
which depend on gas velocity, were not measured in the pre-
sent study.
130
A
-------
100.
E-
O
<
'O
o:
u;
10
1
0.01
I I \ I I I I I \
= 45°
0.05 0.1
DROP DIAMETER, cm
0.5 1.0
Figure 9-3. Predicted superficial reentrainment
velocity due to tearing of drops
with vertical flow.
Curve 1 - Reentrainment velocity for drop attachment length
equal to drop diameter.
2 - Reentrainment velocity for drop attachment length
euqal to half the drop diameter.
3 - Drop terminal velocity.
100 ,
50
u
o
«
113
O.
10
^ I I I | II |
6 = 30°
I I I I I
= 45°
0.01
0 .05 0.1 0.5 1.0
DROP DIAMETER, cm
Figure 9-4. Predicted superficial reentrainment
velocity due to tearing of drops
with horizontal flow.
Curve 1 - Reentrainment velocity for drop attachment length
equal to drop diameter.
2 - Reentrainment velocity for drop attachment length
equal to half the drop diameter.
4 - Maximum drop diameter that can occur at sampling
point.
131
-------
u
30
U
2 20
w
w
10
W
w
i-J
*<
i— i
u
w
Cu
D
oo
3
0.1
i i I I r
8 = 30
0 .3
0.2
0.1
_ 0.05
i i i i i i i I o .1
s
o
w
H
CH
O
OS
Q
X
0.5
DROP SPHERICITY, ip
1.0
Figure 9-4a
Predicted superficial reentrainment
velocity and maximum reentrained drop
diameter for horizontal gas flow.
132
A
-------
EXPERIMENTAL RESULTS AND DISCUSSION
Overall Efficiency
Vertical Baffle - The overall collection efficiency for
horizontal gas flow through vertical zigzag baffles was
determined as a function of gas velocity as shown in
Figure 9-5. The separator attains 1001 efficiency for
gas velocities between 3.0 and 6.0 m/sec.
Figure 9-6 is collection efficiency for inlet en-
trainments with mass median drop diameter of 84 ym. The
efficiency falls sharply for gas velocities below 3.0
m/sec. Reentrainment velocity was not reached even at the
maximum flow rate achievable in the present pilot plant.
Experimental results reported by Bell and Strauss
(1973) for zigzag baffles are plotted in Figure 9-7 along
with points obtained in this study for d = 380 ym and a
pg
line representing the data of Houghton and Radford (1938).
The inlet entrainment of the Bell and Strauss experiments
was comparable to this study, but their overall efficiency
was much lower. This is probably due to the differences
in separator design as reported in Table 9-1.
Houghton and Radford's experiments were conducted
under two operating conditions: (1) Liquid flow rate
= 38 cm3/niin and spray drop diameter ranging from about
1 to 60 ym, the predominant size being 40 ym, and
(2) Liquid flow rate = 12.3 £/min and spray drop diameter
ranging from 2 to 800 ym, the predominant size being about
500 ym. The results obtained under both conditions were
similar and were comparable with the present results due
to similarities in the design, as summarized in Table 9-1.
Horizontal Baffles - Experimental penetration as a function
of gas velocity in vertical direction for horizontal zigzag
baffles is shown in Figure 9-8. Liquid flow rate is used
133
-------
Table 9-1. COMPARISON OF BAFFLE TYPE ENTRAINMENT SEPARATORS
Number of rows
6°
Lip to prevent
reentrainment
Staggering of rows
Distance between
rows
Spacing between
baffles in a row
Width of baffles
Present
Des ign
6
30
none
2.5 cm
2.5 cm
6.9 cm
7.5 cm
Bell § Strauss
(1973)
4
45
1.9 cm on 1st
§ 3rd row only
none
3.1 cm between
2nd § 3rd row
only
8.8 cm
6.2 cm
Houghton £T
Radford (1938)
6
30
0.5 on 4th a
5th row only
none
0
2 cm
5 cm
134
A
-------
=x° 100
u
z
u
t_>
E 80
COLLECTION
cr.
o
40
1 1 1 1 1 1 1
Q^D OAOQ^ rA oA (3 -
-
o -
Inlet Drop Diameter ,um
A 84
O 380
O 1,230
D >1,230
1 1 1 1 1 1 1
0246
SUPERFICIAL GAS VELOCITY, m/sec
Figure 9-5. Experimental collection efficiency
for zigzag baffle.
100 -
23456
GAS VELOCITY, m/sec
Figure 9-6. Collection efficiency for vertical zigzag
baffle device.
135
-------
100
80
U
60
PH
W
o
I—I
t_l
w 40
p
20
I I I I T I
Houghton § Radford
Data, n=6, 6=50°
Bell 5 Strauss
Data for 2 "V"
Baffles in
Series
I
_L
pg = 580 pm
J =1.5
g I .. I
23456
AIR VELOCITY, m/sec
Figure 9-7. Collection efficiency for vertical
zigzag baffle.
136
A
-------
as a parameter. The inlet drops have a mass median drop
diameter of 90 ym. As seen in the figure, reentrainment
was not present for water flow rate = 13.5 £/m2-min and
gas velocity up to 7.2 m/sec. At higher liquid flow rate
(28 £/m2-min), reentrainment started at 5.2 m/sec. Pene-
tration increased from 0 to 6.21 with the increase in gas
velocity from 4.6 m/sec to 7.2 m/sec.
Inclined Baffles - Figures 9-9 and 9-10 are plots of over-
all penetration versus vertical gas velocity for baffles
inclined at 45° to the horizontal. Liquid flow rate was
used as a parameter. It was observed that the primary
collection efficiency was close to 100%. Figure 9-9 also
reveals that the reentrainment velocity depends on liquid
loading. The higher the liquid loading the lower will be
the reentrainment velocity.
Figure 9-11 is a plot of overall penetration versus
vertical gas velocity for baffles inclined at 30° to the
horizontal.
By comparing the primary collection efficiency curves
for these different baffle orientations, it indicates that
the baffle orientation has no effect on primary collection
efficiency. However, the gas velocity for onset of reen-
trainment depends heavily on baffle installation method.
Vertical baffle has highest drainage capability, therefore,
its reentrainment velocity is highest. Comparison between
reentrainment velocities for different baffle orientations
will be discussed later.
137
-------
2
O
50
40
30
10
0
Figure 9-8
WATER FLOW RATE
O 42 £/m2-min
[~J 28 £/m2 -min
O 13.5 «,/m2-min
i
i
L 2 3 4 5 6 7 8
GAS VELOCITY, m/sec
Experimental penetration versus gas
velocity in vertical direction in
zigzag baffles.
CXO
2"
0
1-1 4
H H
Pi
H
r4
w
Cu
j 2
0
0
1 1
O 91.5
D 142.6
A 172
«_
-
^
I 1
0 1 2
1 1
£/m2 -min
£/m2-min
£/m2 -min
O
, A,
3 4
1 I I
-
1
1 1
A I
\ I
n -
I -
A J ^
5678
GAS VELOCITY, m/sec
Figure 9-9. Overall penetration versus vertical
gas velocity -for drops having mass
median diameter of 1230 ymA for 45°
inclined baffles.
138
A
-------
z;
a
OH
A
_L
I
I
23456
GAS VELOCITY, m/sec
i-igure 9-10. Overall penetration versus vertical
gas velocity for drops having mass
median drop diameter of 400 urn for
45° inclined baffles.
10
A L
O L
O L
3 L
183 J./m2-min
140
97
59
012345678
GAS VELOCITY, m/sec
Figure 9-11. Overall penetration versus vertical
gas velocity for baffles inclined at
50° to horizontal.
139
-------
Pressure Drop
Experimental dry and wet pressure drop were plotted
against superficial gas velocity for baffles in Figures
9-12 and 9-13 respectively. In both figures the solid
lines represent the theoretical prediction of pressure
drop as presented in Equation 9-6. As can be seen, theory
agrees fairly well with experimental data. However, it
predicts a slightly lower wet pressure drop than those
observed in the experiments. By comparing these two fig-
ures, it reveals that the liquid load does not have a
significant effect on pressure drop in the baffle section.
This should be expected as liquid holdup in the baffles
is small because of the high drainage rates.
An attempt was made to correlate the pressure drop
data by using generalized pressure drop correlations for
packed bed. The generalized pressure drop correlations
are applicable to counter-current flow. In the present
pilot plant, experimental pressure drop data were obtained
using horizontal air flow and vertical air flow. The
vertical air flow is more comparable to counter flow than
to horizontal air flow.
In Figure 9-14, predicted pressure drop from general-
ized pressure drop correlations for packed bed is plotted
against experimental pressure drop for baffles. As ex-
pected the data for vertical flow show better agreement.
In the system with vertical air flow, reentrainment was
observed to start at AP = 0.03 cm W.C./cm length of baffle
section.
140
A
-------
-
r—'
I—1
LJ
O
M-3
>
<
OS
'•Sj
1/1
Q)
00
N
DO
• 3 • M
3
00
OJ
f-< oo
Cu ctt
N
X DO
!-c 'H
3 M
'joyn
0)
i-4
3
DO
141
-------
DO
0.
o
C/5 0.01
a,
a
0.005
I I i i i i i 1 i I 1 i i i i j
•Flooding Line for 'jumped Pieces
0.005
O '• c r t i c a 1 g a s
f 1 ow
j/\ llo r i zon t a 1 gas
f 1 ow
0.01
0.1
EXPERIMENTAL PRESSURE DROP, cm W.C./Vm length
Figure 9-14.
Predicted pressure drop from generalized
pressure drop correlations for packed
bed versus experimental pressure drop
in zigzag baffles.
142
A
-------
Reentrainment
Two very important parameters for determining re-
entrainment are gas velocity and the liquid to gas ratio.
The combination of these two, which results in reentrain-
ment as observed experimentally, is shown in Figures 9-15
through 9-18 for vertical baffles, horizontal baffles,
30° and 45° inclined baffles, respectively.
Drops were first observed to be torn off the baffle
edge, i.e. onset of reentrainment, in the shaded area in
Figure 9-15. Most of these drops were settled out in the
observation section. The reentrainment rate was low
(
-------
10"
o
I 10"
in
u
o
cr
" 10
o
o
Some reentrainment
_. Reentrainment in part of
>-' duct only
D
O
I
Primary efficiency <100?
No penetration
I
I
I
I
I
01234567
SUPERFICIAL GAS VELOCITY, m/sec
Figure 9-15. Effect of gas velocity and liquid
load on performance of vertical
baffles. (Horizontal gas flow)
s
s
o
< - •*
S 10
w
5
o
Q
o
cy
_!
5
10
_ 1 1 1 1 I 1 1
I o
A
A .
o o \
" ^
\ o
\ 00
- ° x^
O Q >~~
A No reentrainment
* *
" O Slight reentrainment
O Heavy reentrainment
1 1 i i i i 1
0 1234567
SUPERFICIAL GAS VELOCITY, m/sec
2
-
_
-
—
-
~
-
_
"
8
Figure 9-16. Effect of gas velocity and liquid load
on performance of horizontal baffles.
(Vertical gas flow)
144
A
-------
e
e
o
i—t
H
&
CO
O
H
10
- 3
o
I—I
p
Q 10
i—i -1 u
-
-
-
—
~o
o
i i
Slight
Heavy
I i
i
O
o
I
O
o
reentrainment
[\
\
0 N
O
i i
-
' f~\ —
&
m°
/ \^^
u \o "
0 ^
reentrainment
I
1
i
i i
% 012345678
j
SUPERFICIAL GAS VELOCITY, m/sec
Figure 9-17. Effect of gas velocity and liquid
load on performance of 45° inclined
baffles (vertical gas flow).
10
-3
z,
*s
O
i—i
00
o
E-1
Q
i—i
:=>
10
-
—
-
_ _
-
i I
1 o1 ; ' '
0 \
0 C\ oo
\ OQ
O \
-
—
-
-
~ Osiight reentrainment \ ~
O Heavy
I i
reentrainment \
I I I I \ I
a 012345678
SUPERFICIAL GAS VELOCITY, m/sec
Figure 9-18. Effect of gas velocity and liquid
load on performance of 30° inclined
baffles (vertical gas flow).
145
-------
Liquid Flow on the Baffles
Flow of liquid on the baffle surface was observed
in some of the runs. As the flow increases, the film
thickness of the liquid near the downstream edge in-
creases. The gas stream forms a wake at the back side
of the baffle, which tears away some of the liquid at
the downstream edge. The approximate shape of the wake
is shown in Figure 9-19. The wake formation becomes more
pronounced with increasing gas velocity. The flow of
liquid film on the back side of the baffle is shown in
the same figure.
If the liquid flow on the baffle surface is small,
only drop flow takes place on the back side of the baffle.
Some of these drops reach the upstream edge of the baffle,
where they are reentrained. The reentrained drops splash
on the adjacent baffle in the same row and disintegrate.
Some of these small drops are reentrained in the air. The
drops normally splash on the third quarter width of the
baffles as measured from the upstream. The drops flowing
on the back side of the baffles are 3-4 mm in diameter.
Reentrainment from the downstream edge of the baffle
was more significant compared to reentrainment from the
upstream edge. If the liquid flow on the baffle surface
was drop flow, some of these drops reached the downstream
edge and (1) were reentrained, (2) were turned to the back
side of the baffle, (3) fell down at the edge due to gra-
vity, or (4) stayed at the edge of the baffle until they
grew by coalescing with other drops. Most of the drops
were collected by the third or fourth step. If the liquid
was flowing as a film on the baffle, part of the film was
torn and reentrained at the downstream end. The drops re-
entrained from the downstream edge of the baffle were 3-5
mm in diameter. These drops were normally collected on
146
A
-------
;ormation of wake
Wall acting as collector
Pulsating
liquid flow
on the back
side of the
baffle
Figure 9-19
Some observed phenomena in entrainment
separator (a) formation of wake
(b) liquid flow on the back side of
the baffle (c) wall effect.
147
-------
the baffles of the second row, i.e., drops reentrained
from the second row were collected on the baffles in the
fourth row.
Some wall effect was observed in the baffle section.
There were four baffles in a row and the side walls of the
test section acted as collectors for the entrainment. This
effect is shown in Figure 9-19.
The liquid flow pulsated whenever reentrainment took
place and occurred in film flow and in drop flow. It was
difficult to determine the amplitude of the pulsating film
which may have been of the order of 0.05 cm. The frequency
of the wave was not measured.
CONCLUSIONS
1. The theoretical model based on turbulent mixing
agrees quite well with the experimental results.
2. The dry pressure drop in zigzag baffles can be
determined from drag coefficients for inclined
plates held in the flow. The effect of liquid
load on pressure drop is small.
3. Wet pressure drop for vertical gas flow in zig-
zag baffles can be predicted from generalized
pressure drop correlation for packed bed.
4. The onset of reentrainment velocity depends
upon the drainage capability of the baffles.
Vertical baffles with horizontal gas flow has
the highest reentrainment velocity at a given
liquid loading.
148
A
-------
CHAPTER 10
AIR-WATER-SOLID EXPERIMENTS
The purpose of this study is to determine the effects
of solids suspension on separator performance and plugging
due to solids deposition. The type of entrainment separator
studied includes the cyclone and the zigzag baffles.
Experimental Set-up
Necessary modifications were done in the pilot plant
to study entrainment separation with suspended solids pre-
sent .
Figure 10 shows the revised system for air, water,
and solids. This system incorporates a wash water system
which is used to cut off the tanks holding slurry so the
rest of the system may be washed. The wash system may be
operated after observing the test section for scaling and
plugging. This wash system has two advantages, 1) it keeps
all the lines clean and 2) the slurry can be re-used in
the experiments.
Pure MCal 0" (CaCO-) were used as solids. The parti-
cles have a mass median diameter of 1.9 ym and a geometric
mean derivation of 1.3. The solids concentration varies
between 10% and 20% by weight. These concentrations are
in the range used in industrial scrubber's.
Experimental Data and Observations
The Cyclone - The first set of experiments were made with
the cyclone separator. The experimental results are sum-
marized in Table 10-1. The results indicated that the
presence cf the solid did not affect the collection effi-
ciency of the cyclone. However, the solid caused solid
deposition problem.
149
-------
Table 10-1. EXPERIMENTAL RESULTS FOR CYCLONE (AIR-WATER-SOLID SYSTEM)
Exp .
No.
181
182
183
184
185
186
187
Test
Section
Cyclone
Cyclone
Cyclone
Cyclone
Cyclone
Cyclone
Cyclone*
Air
Velocity
cm/sec
5,280
2,920
2,400
880
2,400
880
4,800
L/G
3.072x10-"
3.084x10""
3.45x10-"
4.37x10-"
3.08x10""
4.37x10-"
2.83x10'"
Hours
Of
Operation
2
1
2
2
16**
16**
16**
Collection
Efficiency
1
100
100
100
100
100
100
100
Pressure
Drop
cm H20
15.57
12.60
9.29
1.28
8.59
1.16
13.4
Reentrainment
-
-
-
-
-
-
-
* Inlet vane present, inlet area = 30.5 cm x 7.5 cm
** Cyclone washed prior to experiment
Table 10-2. EXPERIMENTAL RESULTS FOR BAFFLE (AIR-WATER-SOLID SYSTEM)
Exp.
it '
188
189
190
192
193
194
195
Test
Section
Baffle
Baffle
Baffle
Baffle
Baffle
Baffle
Baffle
Air
Velocity
m/sec
3.0
4.4
5.4
6.0
2.2
1.2
3.6
L/C
volumetric
4.26 x 10""
2.95 x 10""
3.068 x 10""
3.4 x 10""
4.63 x 10""
1.58 x 10""
5.02 x 10""
Hrs. of
operat ion
16*
16*
16*
16*
16*
16*
32*
Collection
Efficiency
1
99
99.13
99.05
99
99
97.04
99.3
Pressure
drop
cm IV . C .
0.91
1.53
2.11
3.47
0. 54
0.21
1.53
Reentrainment
little
little
little
little
—
—
—
* separator was washed xvith water prior to experiment
150
A
-------
Spray Section
Test Section
J Slurry Preparation Tank
1 Slurry Feed Tank
1 Fresh Water Tank for Washing
1 Feed Pump
5 Catch Tanks
5 Circulation Pumps
Figure 10.' Air-water-solid system.
151
-------
The cyclone was opened for visual obser\ration of
solid deposition pattern. It was discovered that most of
the solid deposition occurred in the upper half of the
cyclone. The thickness of the solid layer was 0.2 cm
after 16 hours of operation. In the area close to cyclone
inlet, solid deposition was not present. This is because
most of the entrainment was collected near the cyclone in-
let. The scouring action of the collected liquid prevented
the deposition.
It was observed during experiment that some slurry
drops were torn away from the liquid film on the cyclone
surface and some drops did not form liquid film when they
were collected. These drops were more susceptible to creep
along the wall in the direction of gas flow. As these drops
travel along the wall, solids were deposited at the wet-dry
interface. The solids that deposited were not washed away
by the slurry as the flow on this surface was not continuous.
An attempt was made to wash the cyclone with fresh water
during the experiment. However, it was discovered that the
fresh water was not flowing in the same area where solids
were deposited. So at the end of one hour of washing, the
cyclone inlet velocity was increased (from 24 m/sec to 50
m/sec) . The cyclone was then found to be nearly clean after
30 minutes of washing. The total fresh water added during
washing time was 9.41 by volume of slurry flow.
Zigzag Baffles
Vertical baffles were used in the experiment. The
experimental data are presented in Table 10-2. Seven ex-
periments were conducted at air velocities ranging from
1.2 m/sec to 6.0 m/sec. Each of the first six experiments
was 16 hours long and was conducted in two 8-hour segments.
Run #195 was conducted continuously for 32 hours. The
collection efficiency was close to 991 in all the experi-
ments except where it was 971 in experiment number 94,
152
A
-------
and the air velocity was 1.2 m/sec.
The solids deposition was observed after each experi-
ment. It was observed that the solids deposition increases
as the air velocity is increased through the entrainment
separator.
The solids deposition near the edges was more than on
the center of the baffles. The last three rows had more
deposition than the first three. The deposition on the
back of the baffle was thicker than the frontal surface.
The deposit was as thick as about 6 mm on leading edges
and up to about 1 mm on the flat surfaces. There were
heavy solid depositions on the side wall and ceiling of
the test section after the baffle. The cake deposit pat-
tern was very irregular and showed a strong influence of
eddies and wake flow patterns, which caused deposition on
downstream surfaces.
The overall performance was comparable with the air-
water system, i.e., the presence of solids did not affect
the collection efficiency or pressure drop.
CONCLUSIONS
1. The presence of solid in the entrainment does
not affect the collection efficiency of the
cyclone and baffle as long as the deposited
solids do not change the geometry of separator
considerably.
2. Solids will deposit on the wet-dry interface.
3. There appears to be a minimum slurry flow rate
when scoring can occur. Below this minimum,
solid deposition can occur even though the
surface is wet.
A. The washing method is important. For the cy-
clone washing, the gas velocity should be
different than in normal operation.
153
-------
Page Intentionally Blank
-------
CHAPTER 11
SOLIDS DEPOSITION
Solids deposition is a big problem in entrain-
ment separators. Either suspended or dissolved
solids in entrained drops can deposit in an entrainment
separator and cause plugging, a deterioration in per-
formance, and eventual inoperability of the scrubber
system. The precipitation of dissolved solids depends
on temperature, concentration, and nucleation condi-
tions which are unique to any specific system and it
is, therefore, to be controlled by the appropriate
physical chemical conditions rather than a general design
approach. Suspended solids deposition, on the other
hand, appears to be amenable to a general treatment
and it has been selected for study in this program.
MECHANISM OF SOLIDS DEPOSITION
There has been ample demonstration that suspended
solids will deposit in any type of entrainment separa-
tor so our attention can be given to how it happens,
how to predict its behavior, and how to prevent or
minimize it. The mechanisms of suspended solids depo-
sition can include the following:
1. Settling to non-vertical surfaces
2. Impaction due to:
A. Surface curvature
B. Liquid flow direction changes, including
turbulence
3. Diffusion
L, , 155
Preceding page blank
-------
4. Electrophoresis
5. Liquid loss from slurry drops due to:
A. Drop running down a surface
B. Evaporation
C. "Blotting" by a partially dry surface,
such as previously deposited material.
Once solids have deposited on a surface, the ques-
tion is why they adhere to it. Adhesion of particles
may be caused by:
1. Gravitational force on non-vertical surfaces
2. Trapping in surface roughness due to:
A. The original surface
B. Previously deposited solids
3. Electrostatic forces
4. Surface tension forces due to moisture in the
spaces between particles.
5. Cementing due to the precipitation of slightly
soluble materials.
6. Bridging of deposit between elements of the
separator.
There have been a few adhesion studies dealing
with the adhesion of solid particles in pure gases or
liquids , but no studies have been known so far about
the rate of deposition of suspended solids on an
entrainment separator. H. Uno and S. Tanaka (1970) con-
ducted a study on the adhesion of the suspension of
particles on the wall, and they considered that wetting
of the wall is the most important factor relating with
the adhesion of particles on a surface.
There are three types of wetting, namely, adhesion
al wetting, immersional wetting, and spreading wetting.
156
-------
Adhesional wetting is a state of water drop remaining
on the water repellent surface. Immersional wetting
is the state of particles trapped on the wall when the
wall is immersed in a liquid medium, and spreading
wetting is the state of water spreading freely on a
clean surface. Among these three types of wetting, the
spreading wetting has a predominantly strong trapping
effect.
When a liquid film containing a suspension of
particles flows down a surface, some of the suspended
particles are trapped on the surface. The driving force
for particle trapping is the surface tension of the liquid
film acting upon the water line of the particle surface.
When the thickness of the liquid film becomes less
than the diameter of the particles as in Figure 11-1
but not too thin like Figure 11-2, the pressure "P" on
the particle due to surface tension can be expressed as:
P = 2ir a sin a (2T6-62)1/2 (11-1)
where r = radius of the particle, a = surface tension,
a = angle made between suspension surface and contact
angle of the medium against the particle, 6 = liquid film
thickness. If "6" is smaller than "2r", the particle is
pressed against the wall and trapped. When the liquid
film becomes very thin, as in Figure 11-2, the pressure
at this stage is expressed as:
7' - I C11'2)
where "R" is the radius at the water line along the
particle surface made by the remaining water and "r1" is
the radius of curvature between the particle surface and
the wall .
157
-------
Liquid
film
Figure 11-1 - Trapping of particle by thick
liquid film.
Figure 11-2
Trapping of particle by thin
liquid film.
158
A
-------
Removal of deposits can be accompanied by the
elimination of attractive forces. Washing is the most
common way of overcoming the attractive forces. From
equation (11-1) it can be deduced that if the liquid
film thickness is larger than the diameter of the
particle, the particle is free from the attractive
forces . The degree of freedom of the particle in-
creases as the thickness of the liquid film is increased,
Based on this, it appears that the factors which
affect the deposition of solids on a surface will be:
1. Particle properties, such as size, density,
and shape
2. Slurry flow per unit area of collection surface
3. Liquid film thickness
4. Slurry drop size
5. Slurry concentration
6. Collection surface orientation.
EXPLORATORY EXPERIMENTS
Observations of air-water-solid experimental
systems show that suspended solids will deposit and
adhere to smooth vertical surfaces, and even on the
underside of horizontal surfaces, under conditions
where there is little or no evaporation of water and
no cementation. Thus, one can conclude that there is
less chance of finding a general means of stopping
deposition and adhesion than of learning how to scour
deposits away. The apparatus used to determine the
minimum flow rate that scouring occurs is shown in
figure 11-3. It consists of a constant head reservoir,
through which an overflowing device gives a constant
slurry flow. The air jet sprays the slurry onto the
baffle. Three layers of hardware screen were used to
159
-------
COMPRESSED
AIR
Y
»'
OVERFLOW
CONSTANT
HEAD TANK
,
M
*
6
SCREEN
BAFFLE
SLURRY SUMP TANK
PUMP
Figure 11-3.
Experimental set-up for
solid deposition test.
160
-------
knock out large droplets and to control the amount of
slurry reaching the baffle. The pump recirculates the
slurry back to the head tank and thus the experiment
can be run continuously.
The baffle is divided into eight sections - four on
each side of the baffle. Each section is bounded by silicone
rubber to prevent the slurry flowing from the above sec-
tions (See Figure 11-4). An aluminum foil of 7 cm
diameter is clipped to each section. The slurry is
then sprayed onto the baffle by the air jet. The
flow rate at each section is determined by placing a
7 cm diameter filter paper in front of that section
for about 60 seconds and measuring the increase in
weight of the filter paper. The concentration of
CaCO, in the slurry was calculated from the residual
weight after evaporating the water away from a known
quantity of slurry. The deposition rate was calculated
from the dry weight gain of the aluminum foil after
each run.
During this study, the effects of particle prop-
erties, slurry drop size, and collection surface
orientation on slurry deposition were investigated.
Particle Properties
Calcium carbonate particles were examined under the
microscope. They appeared to be irregular in shape.
However, their size distribution is quite uniform.
161
-------
o
Q
(5
=^
O
ALUMINUM FOIL
SILICONE RUBBER
Figure 11-4. Baffle structure.
162
-------
Figure 11-5 shows the particle size distribution of
calcium carbonate particles. They have a number median
diameter of 1.5 ym and a geometric standard deviation
of 1.3. This corresponds to a mass median diameter of
1.9 ym with the same geometric standard deviation.
Collection Surface Orientation
Ten runs were conducted to investigate the effects
of collection surface orientation on deposition rate.
The first five runs were conducted with the baffle
kept in a vertical position. Runs 6 through 8 were
conducted with the baffle inclined and the slurry
sprayed on the upper surface, and runs 9 and 10 were
conducted with the baffle inclined and slurry sprayed
at the lower surface. The weight percent of CaCO, in
the slurry and the duration of each run are listed in
the tabulation below:
Run
No .
1
2
3
4
5
6
7
8
9
10
CaCO, concentration
\ by weight
6
8
12
9
7
9
6
9
6
6
.0
.5
.3
.6
. 7
.5
.3
.1
. 7
.0
Duration of run,
min .
435
420
465
285
385
275
370
438
795
345
163
-------
Pi
PJ
<
I—(
Q
U
i—i
H
< 1.0
^ 0.9
0.8
0.7
0.6
0.5
I I I I I
i i i i i i i
10 50
CUMULATIVE NUMBER UNDER SIZE,
90
98
Figure 11-5. Particle size distribution for CaCO, particles.
164
A
-------
Figures 11-6, 11-7, and 11-8 are plots of experi-
mental data in the form of percent of solids in slurry
deposited versus the total mass of slurry (liquid plus
solid) flowing to a unit area in a unit time. Both
figures show that solid deposition rate is small at
high slurry flow rate.
Figures 11-6, 11-7, and 11-8 are plotted in Figure
11-9 for comparison. It can be seen that slurry sprayed
on the upper surface of an inclined surface has the
highest deposition rate. This phenomenon might be
due to the higher settling rate of solids on inclined sur-
faces, as is reported by Eli Zahavi and Eliezer Rubin (1975)
The solids deposition data can also be plotted in
the form of Figure 11-10, which shows deposition rate
as a function of slurry flux. It is striking to see the
sharp maximum at slurry flux less than a few tenths
mg/cm2-sec. For comparison with traditional engineering
units, 0.1 mg/cm2-sec corresponds to about l.SxlO"3
gal/ft2-min and an entrainment rate of 1 gal/MCF would
correspond to about 0.1 gal/ft2-min for a zigzag baffle
of the design we used. Thus, if the inlet entrainment
rate were 1 gal/MCF the most rapid deposition rate, and
the place where plugging would first occur, would be
where the entrainment has been reduced to roughly II of
the inlet loading. This is based on the assumption that
the separation efficiency per baffle has dropped to 50%
or less because the larger drops have been removed.
Drop Size Effect
It has been observed that cake formation at the
back surface of the baffle is sometimes thicker than on
the frontal surface. When the baffle is vertical and
165
-------
0 0.1 0.: 0.3 0.4 0.5
SLURRY FLUX mg/cm2-sec
Figure 11-6. Solid deposition rate versus slurry flow rate
for vertical baffle at an angle cf 30° with
the direction of gas flow.
100
i r I ( i i
i i i i
0 0.1 0.2
0.4 0 . b 0
SLURRY FLUX mg/cm2-sec
Solid deposition rate versus slurry flow
rate for inclined baffle with slurry
sprayed at the upper surface.
166
A
-------
I I
O Run '9
A Run -10
I I I I
Inclined baffle with slurry
sprayed on lower surface
I i
_l_
j.l 0 . . 0.5 0.4 0.5 0 . l< 0 ." 0.8 0.9 1.0
= ;.URRV Fi.UX itg/cm: -sic
ij.::ri 1 : - 0 . I'orparison of Figures 11-6, 11-", and 11-8.
167
-------
x 4
E
•J
o«3
T
Inclined Baffle
Vertical Baffle
0 0.1 0.2 0.3 0.4 0.5
SLURRY FLOW RATE, mg/cm2-sec
NOTE: Slurry deposition rate must be multiplied by fraction
solids to get cake deposition rate.
Figure 11-10. Slurry deposition rates for
inclined and vertical baffles.
1,000
500
E
a
a.
•s.
<
100
50
40
A Front Surface
O Back Surface
10 50
CUMULATIVE NUMBER PERCENT
90
99
Figure 11-11. Drop size distribution plot for Run fflO.
168
A
-------
the slurry is sprayed normal to the baffle surface,
deposits at the back surface are scarce. However,
when the baffle is inclined at an angle to the vertical
or if the slurry is directed at an angle to the baffle
surface, as it will be in a zigzag arrangement, heavy
deposits are obtained on the back surface. Sometimes
deposition on the back surface is thicker than that on
the frontal surface.
During run number 10, drop size measurement was
taken fron both surfaces and compared and the obser-
vations are summarized below. Figure 11-11 shows the
drop size distribution from both surfaces.
Front Surface Back Surface
Slurry mass median drop 830 ym 170 ym
diameter hitting the
surface
Drop geometric standard 1.9 1.4
deviation
Amount of deposition Varies along Heavier than the
the surface front, no sign of
washing is observed
Thus, it was suspected that the slurry drop size might
have an effect on cake formation on baffle surfaces.
Five experiments were then conducted to investigate the
effect of drop size on the deposition rate on a baffle
surface. Different drop sizes were generated by varying
the orifice size of the air nozzle. Listed below is a
summary showing the weight percent of CaCO., in the slurry,
O
duration of each run, and drop sizes. The slurry sprays
for runs 11 and 12 were generated by a 0.22 cm air
169
-------
nozzle, while those of runs 13, 14, and 15 were gener-
ated by an air nozzle of 0.46 cm diameter.
Run CaCO_ Concentration Duration
No. (1 by weight) of Run
(min.)
11
12
13
14
15
3
10
9
7
7
.8
.5
.5
.5
.7
211
360
294
310
420
Drop Size
Mass Median Geometric
Drop Dia- Standard
meter Cum") Deviation
190
160
390
440
420
1.
' 1.
1.
1.
1.
6
6
7
7
7
Figure 11-12 is a plot of experimental data in the
form of slurry deposition rate versus slurry flux for
small drops while Figure 11-13 shows the same relation-
ship for large drops.
The two graphs are plotted in Figure 11-14 for
comparison. It can be concluded that small drops have
a slightly higher deposition rate than large drops at
high flow rates.
Some Observations on Solids Deposition Experiments
During the exploratory experiments, the mechanisms
of drop deposition on the baffle, particle adhesion,
and washing were carefully observed. This was in order
to develop a means of incorporating the data obtained
during the experiments into a prediction of where depo-
sition occurs most in the pilot plant. The following
phenomena were observed:
170
A
-------
h- O
171
-------
10
z
o
0.01
I
I
I
0 0.2 0.4 0.6 0.8
SLURRY FLUX mg/cm2-sec
Figure 11-14. Deposition rate versus slurry
flux for big and small drops.
1.0 1.1
1 .0
20.5.
20 30 50 100
DROP DIAMETER, ym
Figure 11-15. Predicted penetration versus
drop diameter for zigzag baffles.
172
A
-------
1. Fine drops depositing on the surface will first
stick there. As more drops accumulate, they
aggregate to a bigger drop.
2. As the aggregate of drops gets heavy, it
slides down the surface, sweeping the other
drops as it goes, forming a thin film.
3. As the film slides down, some of the particles
are washed away, while some are left behind
on the surface.
4. The top part of the collection surface always has the
heaviest deposition. The deposition thickness
gradually decreases at the lower end.
Conclusions on Solids Deposition Experiments
Based on the solids deposition experiments, the
following conclusions can be drawn:
1. The solids deposition rate depends largely
on particle properties, such as size, density,
and shape, etc.
2. Deposition rate decreases as the slurry flux
is increased.
3. Deposition rate decreases as the liquid film
thickness is increased.
4. Deposition rate is higher on an inclined baffle
than on a vertical baffle due to the increase
in settling rate of solids suspensions.
5. Small drops are more susceptible to being
caught in eddies which would bring them to the
back surfaces of the baffles.
6. Small drops have a higher deposition rate than
large drops.
173
-------
SOLID DEPOSITION PREDICTION
The deposition experimental data can be correlated
by the following empirical equation:
RS = W$ exp[-(0.13 + 0.53$)S] (11-5)
where Rg = deposition rate of CaCO, on a vertical
flat surface, mg/cm2-sec
W = weight fraction of solid in slurry
4> = slurry flux, mg/cm^-sec
6 = liquid film thickness, ym
To gauge the realism of the deposition experiment,
the correlation was used to predict the behavior of
our pilot plant zigzag baffle separator. Based on
Equations (9-1) , (9-2) and (9-3) , the grade efficiency
curve can be constructed. Figure 11-15 shows the grade
efficiency curves for zigzag baffles with number of rows,
n, as parameter. The following values of parameters
were used in the calculations.
W = 7.5 cm
b = 7.25 cm
6 = 0.524 rad (30°)
- 4
r = 1.8 x 10 poise
^ = 3.6 m/sec (same as Run #195)
174
-------
The overall collection efficiency for the
zig-zag baffle can be obtained by the cut diameter
method reported by Calvert (1974) . Equations (9-1)
through (9-3) can be combined to obtain the following
equation in the appropriate form for using Figure 12-4.
where A =
Pt = exp [-A cT]
p , an W 9
Vr u,, b tan 6
b li
(11-6)
Based on this method, the following results were
obtained. The cut diameters can be computed from
equation (11-6) by setting Pt = 0.5.
n
}
~)
J>
4
5
6
dp50> um
57
40
5 5
28
2 5
2 5
Pt
0.0087
0.00255
0.0011
0.00062
0.0004
0.0003
E
0.9913
0.9975
0.9989
0.9994
0.9996
0.9997
A mass median of 400 um and geometric standard
deviation of 2 were assumed for the slurry drop size
distribution in the above calculations. This is
equivalent to the distribution generated by an M-26
spraying nozzle of Spray Systems Company.
175
-------
The solid deposition rate in each row of the baffle
can be calculated if the entrainment flow rate, gas velo-
city and weight percent of solid content are known. The
following is an example of the calculation for the third
row:
Assume entrainment flow rate = 190 cm3/sec (3 GPM)
CaCCU concentration = 10% by wt .
•J
CaC03 density = 2.7 g/cm3
Then, the entrainment mass flow rate is 227 g/sec. The
amount of slurry collected by 3rd row
= 227 (E , - E 0)
v n = 3 n = 2J
= 0.318 g/sec
= 318 mg/sec
It is assumed that slurry is uniformly spread over
the baffle surfaces (both front and back) . Then slurry
flux, , is:
= °-087 m/cm2- sec
Slurry deposition rate can be calculated from equation
(11-5) once the liquid film thickness "6" is known. Calcu-
lation method for "6" was presented in "Initial Report" for
both horizontal and vertical baffles. As an illustration,
at the leading edge or top edge of a vertical baffle, film
thickness approaches zero. According to equation (11-5), the
solid deposition rate will be
RS = (0.1)(0.087)
= 0.0087 mg/cm2- sec
If the deposited cake has a porosity of 401, then the cake
density is (2.7)(1-0.4) = 1.6 g/cm3. For a 32 hour experi-
mental run, the cake thickness at the leading edge will be
(0.0087) (5600) (32)
(1000) (1.6)
= 0.63 cm
176
-------
Following the same method, cake thickness at other
location can be calculated. Figure 11-16 is a plot of
cake thickness versus the horizontal distance from the
leading edge along a surface 50 cm from the top and
Figure 11-17 is a plot of deposit thickness versus the
vertical distance from the top of the baffle. The deposit
thickness is predicted to vary between 1 to 5 mm on the
inside surfaces and 6 mm at the leading edge. This is
what was observed in the pilot plant experiments.
These calculations indicated that equation (11-5)
can be used to predict the most likely location for solid
deposition to occur in the baffle and minimum amount of
washing liquid required. Once the location and liquid
requirement are known, one can design a washing system
to wash clean this area.
For the baffle test section used in the present study,
equation (11-5) predicts that solid deposition will start
on the third row of the baffle. Thus spray nozzles for
cleaning purpose could be installed between second and
third row. Also fine spray should be used. This will
allow the gas turbulence to carry some washing liquid
to the back side of the baffle.
177
-------
DISTANCE FROM LEADING EDGE, cm
Figure 11-16. Predicted deposit thickness along a
baffle surface 30 cm from top.
0.5
0.4
0.3
0.2
10 20 30
DISTANCE FROM TOP EDGE, cm
Figure 11-17. Predicted deposit thickness versus
distance from top edge of baffle
at 3 cm from leading edge.
178
-------
CHAPTER 12
DESIGN APPROACH
Design equation for knitted mesh, packed bed, tube bank,
cyclone and zigzag baffle are presented in Chapters 5 through
9. In this chapter, we will clarify and show the application
of these equations in the design and selection of a proper
entrainment separator. The following is a brief outline of
topics that are covered in this chapter.
I. Requirement.
A. Performance requirement
B. Capacity requirement
C. Process and physical limitations
II. Entrainment information needed for design and
selection.
A. Liquid phase
B. Gas phase
III. How to select the type of entrainment separator.
A. Choose possible type(s) for detailed study
B. Predict characteristics
REQUIREMENTS
Before one can design or choose an entrainment separator,
he must first study the process and source of entrainments
in order to specify the performance requirement, the capa-
city and the limitations. The following is an outline of
the requirements needed to be considered in the design and
selection of entrainment separator.
Performance requirement
The performance requirement of an entrainment separator
could be defined in terms of:
1. collection efficiency
2. maximum outlet loading
5. behavior of the emitted entrainment
179
-------
Collection efficiency is the overall efficiency of
the separator or separators if several units are arranged
in series. The overall collection efficiency is the differ-
ence between primary efficiency and reentrainment. The pri-
mary efficiency is the collection efficiency an entrainment
separator would have if reentrainment were not present.
Primary efficiency includes only the collection of drops
present in the original entrainment. The reentrainment
of these collected drops or the subsequent collection of
these reentrained drops does not affect the primary collec-
tion efficiency. Reentrainment is the mass ratio of drops
entering the gas from the liquid collected in the entrain-
ment separator, to drops present in the original inlet en-
trainment. Due to reentrainment,. the overall collection
efficiency is always lower than the primary efficiency.
In specifying the efficiency requirement, one should always
define it in terms of overall efficiency.
In a wet scrubber, the scrubber liquor usually contains
suspended and dissolved solids. These solids could be the
separated particulates or the chemicals added to the scrubber
liquid. Entrainment carryover will cause the solids in the
drops to be re-suspended in the gas stream. Thus, the
efficiency of the scrubber decreases and the emission loading
increases. In order to set the maximum allowable outlet en-
trainment loading, one should determine the maximum allow-
able contribution of pollutants in the entrained droplets
to the total emission. For example, one may specify that
the acceptable contribution of entrainment to particulate
emission is 5%. If the emission rate is 4.54 kg/hr
(10 Ib/hr), then 5% allowable contribution corresponds to
227 g/hr. If the solid concentration in the scrubber liquor
is 10%, then the maximum allowable outlet loading of the en-
trainment will be 2.3 kg/hr or 2.3 1/hr if the liquid den-
sity is 1 g/cm3. Of the two requirements just mentioned,
180
A
-------
one should always choose the one that is more stringent
as design basis.
Besides these two performance requirements,the beha-
vior of the emitted entrainment droplets should also be
specified. For example, there must be no "rain-out" of
liquid drops in the vicinity of the emission point.
Capacity Requirements
Capacity requirements can either be defined in terms
of gas flow rate or liquid flow rate, depending upon which
one is the limiting factor. One should specify the maximum
and minimum gas and liquid flow rate, in order to design an
adequate entrainment separator that can cover the whole
range of scrubber operating conditions, not only normal
gas flow rate and liquid entrainment flow rate information.
Liquid flow rate has great effect on the onset of re-
entrainment. Data obtained in the present study showed
that the higher the liquid flow, the lower will be the
onset of reentrainment gas velocity and the higher will
be the chance of flooding.
Process and physical limitations
Several physical and process limitations should be
spelled out before the design of the separator. Some of
the limitations are:
1. Pressure drop. What is the maximum pressure
drop available for the operation of the en-
trainment separator? In the case of mechani-
cally aided separator, the question is what
is the maximum allowable power input.
2. Space. If the separator is to be installed
inside the scrubber, then one should have the
knowledge beforehand regarding the volume,
181
-------
height, etc., inside the scrubber that is
suitable for the installation of the separator.
If the separator is to be installed as an
independent unit of the pollution control
system, then one should have information
about the space available.
3. Materials
4. Maintenance
5. Susceptibility to plugging
6. Orientation
ENTRAINMENT INFORMATION
In order to design a proper entrainment separator, or
to predict the collection efficiency of an entrainment
separator, certain information on liquid phase and gas phase
is needed. This includes
A. Liquid phase
1. Entrainment drop size distribution. This
is the most important single factor in the
design and selection of an entrainment sepa-
rator. Different entrainment separators are
limited to certain drop diameters, below
which their efficiency falls off sharply.
2. Entrainment loading. If drop size distri-
bution and entrainment loading are not known,
they can be estimated based on method des-
cribed in Chapter 3.
3. Suspended and dissolved solids
4. Densities
5. Vapor pressure
6. Nature of the entrainment, i.e. is it sticky,
corrosive, oily, etc.
B. Gas phase
1. Temperature
2. Pressure
182
A
-------
HOW TO SELECT THE TYPE OF ENTRAINMENT SEPARATOR
Preliminary selection
After analyzing the process and limitations, one should
summarize all available information such as called for in
the information sheet shown on Figure 12-1. Next, the poss-
ible type(s) for detailed study can be chosen after ranking
by:
1. Efficiency capability
2. Maximum gas velocity for onset of reentrainment
3. Liquid capacity
4. Plug-ability
5. Installation and operating costs
Figure 12-2 shows the approximate application ranges
for several common entrainment separators. From the drop
size information, this figure can tell what types of en-
trainment separator are available that might be suitable.
However, this figure does not give any information about
the collection efficiency of these separators.
Table 12-1 lists other important limitations for these
common separators.
For cases where drop collection efficiency requirements
are stringent, the prediction of efficiency must be precise.
The "cut diameter" method provides a convenient approach
to the definition of separator efficiency.
The "cut diameter" method, first described in the
"Scrubber Handbook" (Calvert et al. 1972) and further dis-
cussed by Calvert (1974), can be used as a convenient method
for entrainment collection efficiency prediction. This
method is based on the idea that the most significant single
parameter to define both the difficulty of separating en-
trainments from gas,and the performance of entrainment sepa-
rator, is the drop diameter for which collection efficiency
is 0.5(501) .
183
-------
Figure 12-1. ENTRAINMENT SEPARATOR DESIGN AND SELECTION
INFORMATION SHEET
I. Application: (Describe service application of unit when
possible)
2. Operating Conditions: Maximum Minimum Normal
Gas Flow Rate
Entrainment Flow Rate
Temperature
Pressure
3 . Entrainment Phase
Source of Entrainment
Density Viscosity Surface tension
Composition or Nature of Entrainment (Corrosive,oily)
Drop Size and distribution
Solids Content (Composition and Quantity)
Dissolved
Suspended
Performance
Allowable Total System Pressure Drop
Allowable Separator Pressure Drop
Allowable Entrainment
5. Special Conditions:
184
A
-------
I I I I I I I I I
ELECTROSTATIC PRECIPITATQRS ^-GRAVITY SETTLER
^
CYCLONE
STF.VF COLUMN
BAFFLE
MESH
PACKFD RED
TUBE BANK
I I I I I 1 I
0.01 0.05 0.1 0.5 1 5 10 50 100 500 1,000
DROP DIAMETER, ym
Figure 12-2. Entrainment separator approximate operating range
185
-------
n)
a
4-*
O
s
o
u
QJ
O ' 7)
cd rH Ti
q cj o
+J O
q co
•H -3 q
3) q -H
S 33 4J
VI
q
0
4-J
cd
4_)
•rH
6
-1
4-J
IH q
s q u
E X-p-i CJ
•r-t 4-J 73 (f>
X •-< SH -\
O U 4-J E
r- o q
CL,<—I QJ
DJ QJ QJ
< > OS
tJ
CTj
O
—1
"~3
3
—3
q x
o u CD £ -H
Cd "~t -rH CO
>"d 'rH Q,
CliH-H In O
QJ '-H O S-.
tO W "-H Q
0>
6 .H
3 co
e
c o
S Q
X
s +-•
3 •""•*
E U
•H Cd
X &
oj ctf
s o
o
4-1 r-H
U cti
3 -H
j_J }_,
4-> 0
V) 4-1
q oi
CJ
o
4-J
CO CJ
r-. D,
03 X
pL,E— i
QJ
CO
O
00 i i X
SH -d --H q o -H x
ni q o 3 u --H u
-H Cd O O 73 71 £
— i 'J t—> Cd 73 ,q
O 73 > O 3 J "H
x o co o o ~q
S — « V. O "3 CH
73 M G >< X
— ' *3 M 73 • I- CJ r-
<— 3 — 71 O O >
o cr oc o r< x1—
Vj ._ ._ ._i 0 73 -H VI
2
o
X
i-l
QJ
>
71
o la
q o X-H
— i ^ !-i
V) "3 'JO
o q -H +j
O 73 4-J rf
O m '-/I E
0
^
VI ,-1
erf ..
QJ O
rH O O ^r
q 4-j c/i
73 13 ca at
-q -H s-.
3 JZ
q cr tri oo
73 -H 73 -H
U rH Dfl^
0\Q
LD
en
S m
o
X
SH
O
>
o
V) 1 .-H
*3 .-H 4-*
O CU QJ
a. -M u
o q
Q q QJ
O -3
O
LTJ
V)
"d
X.H
•H t-1
fl Qj
e
X M
4-» O
• r-* i — I
73 +J
in QJ
O If,
p- q
— i tx o
"D 3 cd 'J 4-1
QJ '_ • —
30 O — i u- E
3^-5 2 0
— U Vi Cw
ex, o q
• 4-* O '•/>
o >, q •— ' q
.-i 73 '-
X V. r- 'J 4-*
73 r3 O •-- •-
^
O
]
S- 4-t
o 1/1 q
'•— r3 CJ
.C -H 4-1
4-j oC1 — i q
0 TH 0 0
2: x: -ys u
LO
I'l
N
m U
' O 1
-H "J
X O
u") i/i
-H _o q
E S SH S 0
zi a QJ o •-<
(-0 LH v) u 4-J
^
i
U J2
CJ (/)
V) QJ
\ E
S ^
(.n o
i O
SH 4-J
o 4-j q
4-1 73 -H
td X CJ
E 4-- .a q
S 1J
X--H q CTl r-
q 31 73 rH -H
< .H 'J T) 2
U X
r-1 71
•H OJ
S E
O
CJj
^ 2
q o
73 '—
M
"3 '/!
71
J~* O
O (H
4-J 'J
4-J
o q
u
73
I* 4J
o 7i q
-H "3 O
•i-j ^£ — ' q
0 •- 0 0
LD
-i
— i
^
73
!H
O i
O O O
CO J-H
4-J
S "3 cd
O CJ • — 1
>-. M QJ
U, — J-i
D.=>
CO
E
J3-
Kv
O
QJ
V)
•v^.
m
E
u-l
0
*-«-(
00 73
q o -H
H _C IH
^ O
rj q x 4-j
d. U 73 S
•3
QJ
0 OJ
73 . 0
0.
._
T-, 0
OJ J^ 0 X
'-— 1 ^ —
'— 73 q 73
J2 o *J "J -^
•rJ O
D o 3fj • q
'_J JZ 73 — •-
4-1 ^_> ._ 4_. — 3
SH > 73 q o
O '-i « O v,
i
o
-J
r-~-
oj
i V)
• -H 7}
rH 00— 1
E O — (
3 4J
= 0 u
•- 1 -3 --^ :•:
>- •-< -t-i
S CT SH J3
=VO
C7i
CO
E
^1
— <
^
1
o 4-J q
o o o
V, 3 71 TH
\ "3 VI 4J
m O U
E <-H rH QJ
[-- O 'J 71
1 — I
X -H
q SH
4_1
6
o
< — !
U-t
cd
—
o >,
3 —
73
o q
o
73 q
V
O
1
•J",
"3 -3
q --H
rH O 4-t
o x 7i q
'— -4 0
4-j "^ o£ q
O 4-J •— i O
^. T. ^=. U
^
— i
f)
^
• O
X 71
M O
ci 4-J
O t— i 73
•-1 O J-
^
LO
s
U
CJ 73
'/I QJ
rvi 73
O '— .
-H O
73
X-i-*
q rH
4J
73
_Q •"•
3 cd
186
-------
When a range of sizes is involved, the overall
collection efficiency will depend on the amount of each
size present and on the efficiency of collection for that
size. We can take these into account if the difficulty
of separation is defined as the diameter at which collection
efficiency (or penetration) must be 50%, in order that the
necessary overall efficiency for the entire size distribution
be attained. This particle size is the required "separation
cut diameter", "dRr" and it is related to the required over-
all penetration, Ft, and the size distribution parameters.
The number and weight size distribution data for
most entrainment from scrubbers follow the log probability
law. Hence, the two well established parameters of the
log-normal law adequately describe the size distributions
of the drops. They are the geometric mean weight diameter
"d " and the geometric standard deviation "a ".
Pg g
Penetration for many types of inertial collection
equipment can be expressed as a function of constants "A"
and "B".
Pt = exp (-A d!*) (12-1)
d
Packed bed, baffle, mesh, tube bank, cyclone and
sieve plate columns follow the above relationship. For
the packed bed, mesh, baffle, tube bank and sieve plate
column "B" has a value of 2. For cyclone, "B" is about
0.67.
The overall (integrated) penetration, P~tf, of any
device and size distribution will be
W ,
Pt = (^)Pt (12-2)
o
187
-------
The right-hand side of the above equation is the
integral of the product of each weight fraction of drop
times the penetration on that fraction. If equation (12-2)
is solved for a log-normal size distribution and collection
as given by equation (12-1) , the resulting equation can be
solved to yield Figures 12-3 and 12-4.
Figure 12-3 is a plot of "Ft"" versus (d /d )B with
& F P50 Pg
"B In a " as a parameter• For a required "PT" one can
find the value of d^p when "d ", "a0", and "B" are given.
K-L, Pg »
For convenience, Figure 12-4 is presented as a plot of "Pt"
versus (d rn/d ) with a as the parameter when B = 2.
p ou pg g
To illustrate the use of the separation cut diameter,
assume that 951 collection efficiency (51 penetration) is
needed for drops with mass median diameter, d , equal to
100 ym and geometric standard deviation, a_ , = 3 . If an
o
entrainment separator such as baffle is to be used, Figure
12-4 shows that (d 50/d ) = °-15- Thus, the required cut
XT r &
diameter, dnn, must be (0.15)(d ) = 15 urn. If the separator
KL pg
is capable of a smaller cut diameter, that is good; so "dRC,,
is the maximum cut diameter acceptable.
Prediction of separator's cut diameter
Selecting an entrainment separator with the proper
cut diameter requires some knowledge of its performance
characteristics. The most important of these are primary
efficiency, gas pressure drop, and capacity limitations.
The energy required for entrainment separation is
generally a function of the gas pressure drop. Figure 12-5
is a plot of performance cut diameter, d , versus gas pres-
sure drop. Theoretical energy consumption is also plotted
on the same figure. This figure was constructed based on
188
A
-------
z
o
ir
i-
u
z
UJ
Q.
a
u
s
0.001
0.01
0.001
Figure 12-3. Integrated (overall) penetration as a
function of cut diameter, particle
parameters and collector characteristi
o
er
CC
LJ
O.
<
tr
UJ
§
.001'
0.001
0.01
Figure 12-4. Overall penetration as a function of cut
diameter and particle parameters for common
scrubber characteristic, B = 2.
189
-------
design equations and experimental correlation presented
in Chapters 5 through 9.
For the example mentioned earlier, from Figure 12-5,
for a required cut diameter, dRC, of 15ym, the required
pressure drop across the separator is 0.01 cm W.C. for
knitted mesh and 2.5 cm W.C. for a six-row baffle. Suppose
the maximum allowable pressure drop across the entrainment
separator is less than 1 cm W.C., then this quick calcula-
tion indicates that baffle is not suitable for installation.
In some occasions, some entrainment separator manu-
facturers only give pressure drop versus gas velocity re-
lations in their sale literature. In this case, Figure
12-5 can be used to predict the collection efficiency of
the separator. For example, suppose a packing material
manufacturer says that the pressure drop is 2.5 cm W.C. when
the gas velocity is 3m/sec (10 ft/sec), then from Figure
12-5, the expected performance cut diameter is 3.5 ym if
this material is used as packing.
For the same drop size distribution as mentioned
earlier, then
,P50 = lil = 0.035
pg 100
From Figure 12-4, the expected collection efficiency
of the packed bed is 99.8% (i.e. penetration = 0.002).
To estimate the penetration for drop diameters other
than the cut size, under a given set of operating conditions,
one can use the approximation of equation 12-1 with B = 2.0.
Alternatively, one could use more precise data or design
equations for a given separator. Figure 12-6 is a plot of
the ratio of drop diameter to cut diameter versus penetra-
tion for that drop size on log-probability paper.
190
A
-------
:oo,
TI:EORETICAL POKER CONSUMPTION, Kw/Mm'/mi
0.050.1 0.5 1 5
50 100
50
10
Baffle, n = b, 6=30°
Baffle, n=6, 9=45°
Tube bank, n=6, b = l
Tuhe bank, n=6, b=0.
I I I I I I I I I
I I i i i I i I
0.01
0.05 0.1
0.5
1
10
50
100
PRESSURE DROP, cm W.C.
Figure 12-5. Performance cut diameter as a function of pressure drop
for several entrainment separators.
99.9
99
COLLECTION EFFICIEN'CY,
90 50
10
3.0
1 .0
0.5
0.1
J 1—I L
J L
_L
J L
_L
10 :o so
PENETRATION FOR dd, »
80
90 95 98 99
Figure 12-6. Ratio of drop diameter to cut diameter as a function
of collection efficiency.
191
-------
Note that the cut diameter method only gives an
approximate collection efficiency, the exact characteristic
of the entrainment separator could be predicted by the
method described in next section.
Predict Characteristics
Table 12-2 is a summary of design equations and
figures for common entrainment separators. The general
steps in utilizing this table to predict the performance
characteristics of an entrainment separator are as follows:
1. Based on process condition and separator
configuration, construct the grade efficiency
curve for the separation. Equations for pri-
mary efficiency can be used for this purpose.
In case the gas velocity is higher than the
reentrainment onset velocity, reentrainment
should be subtracted from the primary efficiency.
2. Compute the collection efficiency for the whole
population of the drops. This can be done either
graphically or mathematically. For graphical
solution, plot Pt. versus fraction smaller than
dj.(where Pt. is penetration for drop size d ,.).
The area under the curve is the overall pene-
tration. Outlet loading is equal to inlet
loading times overall penetration.
3. Compute expected pressure drop.
In the process of designing an entrainment separator,
the steps should be repeated for different proposed sepa-
rator configuration. The final configuration can then be
selected after optimization analysis.
192
A
-------
Table 12-2. SUMMARY OF DESIGN INFORMATION
Type of
Separator
Mesh
Packed Bed
Tube Bank
Cyclone
Baffle
Gravity
Settler
Sieve
plates
Primary
Efficiency
Eq. 5-1
Eq. 6-1
Eq. 7-2, 7-3
Eq. 8-2
Eq. 9-1
Eq. 3-7
Eq. 3-10
Pressure Drop
Eq. 5-3
Fig. 6-1
Eq. 7-4
Eq. 8-15
Eq. 9-6, Fig. 6-1
Eq. 3-12
Reentrainment
Velocity
Eq.5-4,
Fig. 5-13
Figure 6-8
Figure7-9,7-lQ
Fig. 8-3
Fig. 9-15
193
-------
Page Intentionally Blank
A
-------
CHAPTER 15
FUTURE RESEARCH AND DEVELOPMENT RECOMMENDATION'S
The primary objectives of the present research, i.e.
to evaluate the technology on scrubber entrainment separa-
tors, advance theoretical development and solids deposition
have been achieved in the present study. It is also im-
portant to define the areas where additional work is needed,
The following paragraphs give an account of these areas.
REEN7RAINMENT
One of the problems which present day entrainment
separators suffer is their large size which is due to low
operating velocities. The gas velocities are limited by
reentrainment velocities and flooding conditions. Re-
entrainment may take place due to various mechanisms,
depending on flow rates and geometry.
While the present program will provide information
on the conditions under which reentrainment occurs in
several separator configurations, it would be helpful to
have more detailed knowledge of these phenomena. It is
quite possible that a fundamental study of the mechanisms
of reentrainment from different geometric arrangements
would enable one to develop more efficient separator de-
signs. At least the results of such a study would deli-
neate the limits of possible performance and save effort
which might otherwise be expended in unprofitable directions
The study needed is onset of reentrainment conditions,
.rate of reentrainment, equilibrium constant between
Preceding page blank
-------
entrainment and liquid in film, drop size distribution,
smooth and shock type contact of gas and liquid, effect
of duct dimensions, etc. The application to entrainment
separator will include improving design methods to det-
ermine reentrainment under operating conditions, effect
of higher gas velocities and improvements in design to
reduce reentrainment.
SOLIDS DEPOSITION
Solids deposition and consequent plugging is a
major operational problem in scrubber systems. While this
study introduces the minimum flow rate required for
washing, it would be helpful to have more research on
the methods of washing.
As can be deduced from the results of the solids
deposition studies, increasing the flux in the form of
a fine spray will eliminate cake deposition on the backs
of baffles as well as on sheltered regions of the duct
walls. On the other hand, increasing the flux will
lower the collection efficiency of the entrainment
separator. Thus, research on finding the optimum flow
rate required and the feasibility of intermittent washing
would be required.
One other method of eliminating cake deposition would
be increasing the liquid film thickness on the baffles.
However, increasing the liquid film thickness will also
increase the reentrainment rate. Thus, it would be help-
ful to have more knowledge of the degree of increase
in reentrainment rate due to the increase in liquid film
thickness.
196
A
-------
FIELD TESTING OF INDUSTRIAL ENTRAINMENT SEPARATORS
Performance data on industrial entrainment separators
are generally not available. The industrial data are col-
lected to evaluate the overall performance of the scrubber
and it is assumed that the entrainment separators have
100% efficiency. Also, all the liquid introduced in the
wet scrubber is assumed to be removed by entrainment
separator. The effects of sedimentation, bends in the
duct carrying entrainment, etc. are neglected. The dis-
tance between sampling point and entrainment separator
elements is important. Also, the effects of industrial
operating conditions on performance of entrainment separ-
ators should be determined.
The aim of development of entrainment separators is
to improve performance of separators under industrial
conditions. Thus, it is necessary to collect data on
industrial separators. The data, when compared with
theoretical models, will represent possible problems
resulting from industrial conditions and will help in
designing future entrainment separators.
DEMONSTRATION PLANT
From the present contract work, it is felt that we
can predict the performance of an entrainment separator
with reasonable accuracy. It is possible to obtain im-
provement in the performance due to better design. We
would like to move from the present research and develop-
ment to a demonstration of an improved design in the field.
197
-------
The capacity of the present pilot plant is 85 m3/min.
Therefore, the next size should be around 1,000 m3/min
(35,000 CFM).
The demonstration plant operation will involve
selecting an organization which operates a suitable plant
having entrainment separation problems and which is willing
to participate in the demonstration plant program. The
design effort will include obtaining the necessary data
concerning the source of entrainment, preparing overall
design and selecting a final design. The fabrication and
start up will involve selection and negotiation with sub-
contractors, procurement of components and supervision of
subcontractor efforts. The test program will be to deter-
mine performance, observe the effect of change in variables
and compare the performance with theoretical developments.
STUDY OF COMBINATIONS OF ENTRAINMENT SEPARATORS
It is possible that if more than one entrainment
separator is used in series, the combined unit will provide
a synergistic effect. One can combine two different en-
trainment separators to include the best features of each.
Some examples are as follows:
1. The maximum gas velocity in the entrainment
separator is limited to the onset of reentrainment velocity.
It is generally the case, however, that a separator which
has high primary drop collection efficiency will have a low
reentrainment velocity, while one with high reentrainment
velocity will have low primary collection efficiency. If
a combined unit is used with the first unit being used for
198
A
-------
primary collection and coalescence of drops and the second
one for collection of large drops while being below the
onset of reentrainment velocity, increased capacity will
result. The combination of efficiency and capacity will
exceed what either unit can do alone.
Because the size of the entrainment separator will be
smaller, the initial capital cost will be lower. The mini-
mum drop size that can be separated in the entrainment
separator is limited by the operating velocity. This prob-
lem can be solved by using a combination of entrainment
separators.
2. Sometimes the entrainment load is high and con-
stituted of particles in a wide size range. A single
entrainment separator may be inefficient, flooded or may
present reentrainment in this situation. A combined unit
may be used in this case. The first separator is a pre-
cleaner with low pressure drop, which removes large particles
constituting a significant fraction of the entrainment. The
second separator will be an efficient device.
COLLECTION EFFICIENCY FOR SMALL DROPS
The drop size used in the present study was over 100 ym.
Based on our sampling data on various scrubbers, it was dis-
covered that there were substantial amount of entrainment
droplets smaller than 10 ym in diameter. It would be helpful
if more tests were performed to determine the collection
efficiency for drops smaller than 10 ym.
199
-------
Page Intentionally Blank
A
-------
REFERENCES
Anderson, J.D., R.E. Bellinger, and D.E. Lamb. Gas Phase
Controlled Mass Transfer in Two Phase Annular Horizontal
Flow. AIChE Journal, 10: 640,1964.
Atteridge, et al. AIChE Journal 2_: 3, 1956.
Bell, C.G.,and W. Strauss. Effectiveness of Vertical
Mist Eliminators in a Cross Flow Scrubber. APCA Journal
23: 967-9, November 1973.
Bradie, J.K., and A.N. Dickson. Removal of Entrained
Liquid Droplets by Wire Mesh Demisters. Paper 24 in
Fluid Mechanics and Measurements in Two-Phase Flow Systems.
(A joint symposium of the Inst. of Mech. Engr. and the
Yorkshire Branch of the Inst. of Chem. Engr.) 24-25,
London. September 1969.
Brooks, et al. Petroleum Engineering. C-52, August 1955.
Buerkholz, A. Drop Separation on Wire Filters. Chemie
Ingenieur Tecknik. 4_2: 21, 1314-1321, 1970.
Calvert, S. Air Pollution. Stern, A.C. (ed.). _3, Academic
Press, New York. 1968.
Calvert, S. Engineering Design of Fine Particle Scrubbers.
APCA Journal, 2±: 929-934, 1974.
Calvert, S., J. Goldshmid, D. Leith, and D. Mehta.
Scrubber Handbook. Prepared for EPA Contract No. CPA-70-95.
Vol. I and II, 1972.
Calvert, S., and D. Lundgren. Particle Collection in Closed
Packed Arrays. Presented at AIHA. 1970.
Calvert, S., et al. Entrainment Separators for Scrubbers -
Initial Report. NTIS Pub., PB-241-189.
Carpenter, C.L., and D.F. Othmer. Investigation of Wire
Mesh as an Entrainment Separator. AIChE Journal, p. 549,
1955.
Chien, S.F., and W. Ibele. Pressure Drop and Liquid Film
Thickness of Two Phase Annular and Annular-Mist Flows.
ASME Paper. 62-WA170.
Preceding pap blank
201
-------
Davis, R.F. Proc. Inst.Mech.Engrs. 149: 148, 1940.
Page, A., and F.C. Johansen. Proc. Roy. Soc. (London).
116A: 170, 1927.
Foust, A.S., L.A. Wenzel, C.W. Clump, L. Maus, and
L.B. Andersen. Principles of Unit Operations. Toppan
Company. Tokyo. 1959.
Fuchs, N.A. The Mechanics of Aerosols. The Macmillan
Company. 1964.
Golovin, M.N., and A.A. Putnam. Ind. Engr. Chem. Fund.
3^:264, 1962.
Houghton, J.G., and W.H. Radford. Trans. Am. Inst. of Ch.E.
.35: 427, 1939.
Hunt, Hanson, and Wilke. AIChE Journal,.!: 441, 1955.
Jackson, S., and S. Calvert. AIChE Journal, 12: 1075,
1966.
Jashnani, I.L. Coalescence and HTU in Foam Fractionation
Columns. Ph.D. Dissertation, U. of Cincinnati. 1971.
Jones and Pyle. Chemical Engineering Progress. 51: 424,
1955.
Kitchener, J.A. Foams and Free Liquid Films in Recent
Progress in Surface Science. Academic Press. New York.
1964.
Lane, W.R. Shatter of Drops in Streams of Air. Ind.Engr.
Chem. _4_3: 1312, 1951.
Leith, D., and W. Licht. The Collection Efficiency of
Cyclone Type Particle Collectors - A New Theoretical
Approach. Paper presented at San Francisco meeting of
AIChE. December 1971.
Mercer, T.T., and H.Y. Chow. J. of Coll. and Interface
Sci. 27_: 75-83, 1968.
Nukiyama, S., and Y. Tarrasawa. Trans. Soc. of Mech. Engr.
(Japan). <4, _5, £, 1938-1940.
Perry, J.H. Chemical Engineering Handbook. 4th Edition.
McGraw-Hill. New York. 1963.
Perry, J.H. Chemical Engineering Handbook. 5th Edition.
McGraw Hill. New York, 1Q63.
202
A
-------
Poppeile, E.W. Master Thesis. Newark College of Engineering,
Roberts, B.C., and D.E. Hartley. A Correlation of Pressure
Drop Data for Two Phase Annular Flows in Vertical Channels.
Queen Mary College. (London). Nuclear Research Memorandum,
No. Q6.
Satsangee, P.D. Master's Thesis. Polytechnic Institute
of Brooklyn. 1948.
Shepherd, C.B., and C.E. Lapple. IEC Chem. 31, 1246, 1940.
Sherwood, T.K., and R.L. Pigford. Adsorption and Extraction.
McGraw-Hill. New York. 265, 1952.
Sherwood, T.K., G.H. Shipley, and F.A.L. Holloway. Ind.
Engr. Chem. 3£, 765, 1938.
Schurig, W.F. D.Ch.E. Dissertation. Polytechnic Institute
of Brooklyn. 1946.
Stearman, F. and G.J. Williamson. Spray Elimination in
Processes for Air Pollution Control. Nonhebel, 2nd ed.,
CRC Press, Cleveland, 1972.
Steen, D.A., and G.B. Wallis. The Transition from Annular
to Annular-Mist Cocurrent Two-Phase Down Flow. NYO-3114-2,
1964.
Taheri, M. and S. Calvert. APCA Journal. 18, 240, 1968.
Uno, H., and S. Tanaka. Adhesion of Suspension Particles
on the Wall Surface of the Container. Kolloid-Z. U Z
Polymere 242, 1186-1195, 1970.
Wallis, G.B. The Onset of Droplet Reentrainment in Annular
Gas-Liquid Flow. General Electric Report No. 62 GL127, 1962.
York, O.H. Performance of Wire Mesh Demisters. Chem. Engr
Prog. Vol. 50, No. 8, 421, 1954.
York, O.H., and E.W. Poppeile. CEP. _59_, 45, 1963.
Zhivaiking, L.Y. Liquid Film Thickness in Film Type Units.
Int. Chem. Engr. 2, 237, 1962.
203
-------
------- |