Twenty- Two y
Particulate Solids Flow and Separation
Through Fluid Mechanics
Many chemical engineering unit operations involve
particulate solids as well as fluids. Often the solids
are an integral part of the material being
processed - for example, conveying spray-dried
product in a gas stream or feeding pulverized coal in
air to a burner. Sometimes the fluid motion is used
to separate a particulate solid from the liquid phase,
as occurs in filtration or sedimentation. Sometimes
the fluid motion is used to separate the solid
particles from each other, as in the classification of
particles of various sizes or densities. In all cases,
however, the equations of fluid mechanics describe
the systems, and momentum transfer is the major
transport process occurring.
The separation methods discussed may be
classified as mechanical separations, as opposed to
those separations requiring vaporization or conden-
sation. For example, salt crystals can be separated
from their mother liquor by filtration or centrifuga-
tion. Several different sizes of crushed ore can be
separated by screening, elutriation, jigging, or
c!assification. Sludges can be separated from a liquid
b·t sedimentation. The mechanical methods of separ-
ation may be grouped into two general classes: those
whose mechanism is controlled by fluid mechanics,
discussed here, and those whose mechanism is not
described by fluid mechanics - for example, screen-
ing.
APPLICATIONS OF THE MECHANICS Of PARTICLE
MOVEMENT THROUGH A FLUID
General Principles
In Chapter 13 the concept of form drag was intro-
duced. For steady flow of a fluid past a solid,
boundary layers are established, and a force is
611
exerted on the solid by the fluid. This force is a
combination of boundary-layer drag and form drag,
and it can be expressed in terms of a drag coefficient.
By Equation 13.40, the drag coefficient is
2Fo
Co=-- (13.40)
v,,2 pS
where F is the force acting on the solid, v,, is the
free-stream velocity relative to the particle, and Sis
the projected area of the solid normal to the flow.
This equation is important wherever momentum
transfer at a fluid-solid boundary must be
examined. Thus, it may be applied to the design of
particle-separation equipment as well as to that of
piping systems. Over the range of flow conditions
likely to be encountered, the drag coefficient may
be determined from Figure 22.1. Note that the
sphericity term is introduced to account for non-
spherical particles. (See Appendix 8 for sphericity
determinations.)
Consider a particle moving throuqh a fluid in
one dimension only, under the influence of an
external force. This external force may be from
gravity or from a centrifugal-force field. The basic
theory of the flow of solids through fluids is based
upon the concept of freely moving bodies,
dv
Fgc =m da (22.1)
where F is the resultant force acting on any body,
dvldl) is the acceleration of the body, and mis the
mass of the body.
In Figure 22.2, the forces acting on the falling
body are the external force (FE), a buoyant force
(F8 ), and the drag force (F0) due to fluid friction
in the direction of the velocity of fluid relative to
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 617
of the particles and a water temperature of 20°C. smallest galena particle in the galena product. By
(b) What is the maximum size range of the Equation 22.20,
galena product? 4g(Pga1- p)µ]
log Co = log NRe + log [
Solution 3p 2 v, 3
(a) For equal-sized silica and galena particles,
log Co= log NRe
the heavier galena will settle faster. Therefore, the

settling velocity of the largest silica particle will 4 x 9.80665(7500 - 1000) (0.001)]
determine the water velocity. A water velocity +log [
3( 1000)2(0.04)3
equal to this settling velocity should give a pure

galena product. By Equation 22.19, using the SI log C0 = log NR, + log 1.33
units This is a straight line on Figure 22.1 passing
through (NR, = 1.0, C0 = 1.33) with a slope of +1.
4gDP 3p(p,;, - p)] This line intersects the ,:, = 0.806 curve at NR e =

log C0 = -2 log NRe + log [

3µ 2 9.0. This Reynolds number corresponds to a
diameter of

The viscosity of water at 20°C = 0.001 N • s/m2.

NR,µ (9)(0.001)
D = -- = = 0.000225 m
P , vp (4.0)( 1000)
log Co = -2 log NRe +
Thus, the galena product size ranges between
lo [4 x (9.80665)(0.007)3 (1000)(2650 - 1000)] 0.225 mm and 0.70 mm. Galena particles smaller
g 3(0.001)2 than 0225 mm are carried overhead along with all
silica.
log Co = -2 log NRe + log 7400. Classification Equipment. The simplest type of
classifier consists of a large tank with provisions for
This is a straight line on Figure 22.1 passing a suitable inlet and outlet. Figure 22.4
through (NA e = 1, C0 - 7400) with a slope of -2. illustrates such a gravity settling tank.
This line intersects the If; = 0.806 curve at NR e = A slurry feed enters the tank through a pipe.
28. Immediately upon entry, the linear velocity of the
A Reynolds number of 28 corresponds to a feed decreases as a result of the enlargement of
settling velocity of cross-sectional area. The influence of gravity causes
NA 0µ · (28) (0.001) the particles to settle, with the faster-settling
x 10-3)(1000) = 0.04 m/s
v, = DvP = (0.7 particles falling to the bottom of the tank near the
entrance. The slower-settling particles will be
This velocity must also be the water velocity carried farther into the tank before settling to the
to ensure a clean galena product, since it will carry bottom. Placing vertical baffles within the tank
all silica overhead. allows for the collection of several fractions. Very
(b) Calculation of the size of a galena partic!e fir.e particles will be carried out of the tank with
that settles at a velocitv of 0.04 m/s fixes the the liquid overflow. The position in the tank at

Fluid in,
----+-
wide rang';!------<
of particle
sizes

Coarse Intermediate Fine

particles particles particles

Figure 22.4. Gravity-settling tank.

618 APPLICATIONS TO EQUIPMENT DESIGN

which a certain size particle may be expected can desired, is fed into the top of the column, The
be calculated on the assumption that the particles large particles, which settle at a velocity higher
quickly reach terminal velocities. The relationships than that of the rising fluid, are collected at the
developed earlier apply to the deposition location bottom of the column, and the smaller particles are
of any given particle size. carried out of the top of the column with the fluid.
The gravity-settling tank is referred to as a Several columns of different diameters in series
surface-velocity classifier. The resulting separation may be used to bring about a further separation.
is not a sharp one, since considerable overlapping A ·double-cone classifier (Figure 22.7) is a
of size occurs. conical vessel inside of which is a second cone. The
The Spitzkasten (Figure 22.5) is another type inner cone is slightly larger in angle and is movable
of gravity-settling chamber. It consists of a series of in a vertical direction so that a variable annular
conical vessels of increasing diameter in the direc- area is available for flow.
tion of flow. The slurry feed enters the top of the The feed material flows downward through
first vessel where the larger, faster-settling particles the inner cone and out at a baffle at the bottom of
are separated. The overflow, including unseparated
solids, feeds into the top of the second vessel,
where another separation occurs since the velocity
is lower than that in the first vessel. Each succeed-
ing vessel will separate a different range of particles
depending upon settling velocities of the particles
and upon the fluid velocity in the vessel. In each
vessel, the velocity of the upflowing fluid is inde-
pendently controlled to give the desired size range
of product from that vessel.
An elutriator (Figure 22.6) is a vertical tube
through which a fluid passes upwardly at a specific
velocity while a solid mixture, whose separation is

Less coarse solids

Fine solids

figure 22.5. Schematic representation of a Spitzkasten.

 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
Particle
feed
Finest
sohds
Water Coarse less Fine
sohds coarse solids
sohds
Figure 22.6. Elutriators.
the inner cone. Rising upward through the annular
space, fluid is fed at a controlled velocity into the
unit in the vicinity of the exit of the inner cone.
The solids from the inner cone and fluid are mixed
and then flow through the annulus whose cross·
sectional area varies. Classification action occurs in
this annular space with the small particles leaving
with the liquid and the larger particles settling to
the bottom for removal.
Another piece of equipment for solid separa-
tion whose mechanism is based upon settling
velocity is the rake classifier (Figure 22.8). The
rake classifier is a tank with an inclined bottom in
which are provided movable rakes. Feed is intro·
duced near the middle of the tank. The lower end
of the tank has a weir overflow over which the
fines that have not settled leave in dilute suspen-
Figure 22.8. Heavy-duty rake classifier. Cutaway shows rake assemblies. (Courtesy
Dorr-Oliver, Inc.)
619
Feed
t Water and
I//J1D tme-sohd
overflow
Movable cone
Fixed cone
Figure 22.7. Double-cone classifier.
sion. The heavy material sinks to the bottom where
the rakes scrape it upward toward the top of the
tank. The stroking action of the rakes is such that
when the stroke is completed, the rakes are lifted
and returned to the starting position. The slurry is
thus kept in continuous agitation. The time of the
raking stroke is adjusted so the heavy particles have
time to settle while the fines remain near the
surface of the slurry in the rake compartment.
Thus, the heavy material is moved upward along
the floor of the tank and removed at the top of the
apparatus as a dense slurry.
For classifying fines material, a bowl classifier
(Figure 22.9) is used. It consists of a shallow
cylinder with a gentle conical bottom. 'The feed
slurry is fed at the center near the surface, and the
liquid flows outward in a radial direction. The
620 APPLICATIONS TO EQUIPMENT DESIGN

Figure 22.9. Bowl classifier 8 by 12 ft in diameter. (Courtesy Dorr-Ofiver, Inc.)

solids settle out according to their particle size
with the fines overflowing at the bowl wall. The
heavy sludge is scraped toward the center for dis-
charge.
Centrifugal Classification
The most widely used of the centrifugal-type separ-
ation equipment is the cyclone separator for separ-
ating dust or mist from gases (Figure 22.10i.
of solids from fluids in a cyclone usually involves
particle sizes small enough so that Stokes's law
may be assumed valid, irrespective of gas velocity.
Equations 22.7 and 22.16 give, for a spherical
particle obeying Stokes's law,
rrD 3 2
----f-rw (p,- p) = 3rrDpvRµ (22.30)
where vR is the radial velocity of the particle. But

The feed enters the cyclone tangentially near rw2 = vf Ir where v,.0 is the tangential velocity
00
a
the top and is given spinning motion as it enters of the particle at radius r. Thus,
the chamber proper. The· tangential velocity of the
particles tends to carry them toward the periphery rrDp 3 Vfun
---,-(p,-p)=3rrDpVRµ (22.30a)
of the chamber. The spiral motion of the fluid
results in some inward radial acceleration of the Solving Equation 22.30a for VR gives
particle, and simultaneously gravitational force
imparts downward acceleration. The result is a Vn -[Dp2. (p,-p) lVEin
downward and spiraling path of increasing radius (22.31)
18 µ J r
until the particle reaches the boundary. Thereafter

the particles continue a spiraling path down the
wal I, and the gas, freed of solids, moves upward in
the central core. At high tangential velocities, the
outward force on the particle is many times the
force of gravity; hence, cvclones.accornpllsh more
but, by Equation 22.15, the bracket term in Equa-
tion 22.31 is the gravitational terminal velocity of
the particle with the constant, g, omitted. There-
fore,
Vt �an

rapid and more effective separation than VR =--- (22.32)

gravitAtional-settling chambers for particles of sizes
g r

down to a few microns. For very small sizes, the
energy represented by the tangential velocity is
insufficient to overcome the centripetal force of
the rotating fluid, and separation is ineffective.
It is assumed that the particles in a cyclone
quickly reach their terminal velocities. Separation
Thus, the higher the terminal velccitv, the
greater the radial velocity, and the easier it should
be to separate the particle. The evaluation of the
radial velocity is complex because it is a function
of terminal velocity, tangential velocity, and posi-
tion from the center of the cyclone. For a given-
 .!'ARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLU!DMECHA res 621
sized particle, the radial velocity is a minimum near
aEAN AIR the center of the cyclone and increases toward the
-- OUTLET wall.
SCROLL OUTLET-

Theoretically, the size of the smallest particle

retained by the cyclone can be calculated, but
experience indicates that agglomeration of particles
and entrainment tend to introduce deviations from
the calculations. If a cyclone is designed to separ-
ate a definite particle size, a cyclone efficiency may
be defined as the mass fraction of the particles of
that size that is separated in the cyclone. Cyclone
efficiency increases as the size of the particles
increases. If a cut diameter is defined as that
diameter for which half the particles by weight will
be separated, the efficiency of a typical cyclone
separator is shown in Figure 22.11.
Cyclones are usually designed to operate with
a specified pressure drop. The pressure drop
through cyclones may be one to twenty times the
inlet velocity head (51 ). Lapple (32) suggests an
inlet gas velocity of 50 ft/sec for ordinary installa-
tions.
Through the use of plastic cyclones, the
effects of size, shape, and air velocity have been
studied by several investigators (19, 56). Signifi-
cant cyclone dimensions are height, diameter, exit-
duct diameter, and the ratio of the diameter of the
cylinder to that of the exit duct.

Wet Scrubbers
For gases containing very fine. particles, liquid
scrubbing 'is sometimes used for effective separa-
Figure 22.10. Cyclone separator. (Courtesy the
tion. The dirty gas passes upward through water
Ducon Co.) sprays, which tend to wash out the dirt particles

l.OE==E=:EE:-=+-El:E===;l:=�Rffij
0.9-1-------+---'--+-
0.87 e------l--_.:
+-+-+' --.-
--+---'-
�
l-+-+-+-µ1...j__�,-"'-�:..+---+-f--..f-r-1•-
:? 0. I i I i /
� o.61-----+-·--1--..-1-+....,:f-tV -1- -,,L----,1---1--+--+-+-t--t-+-i
! o.s 1--------+--·----+--+--�-+---++---- -+--,---+-
. � o• '------<--..:�
...... --++-1-/.,{-+----+---+-t-+-+-+--+-H
8 ! 1.,1 '
t 0.3

f
� 0.21------1----'--+
. I/ �f--..---+---+-+-!f--..H-t-+-,
I _4--I-+----,' 1

1
fl 1 I

£ O.IL_ I I I i I ;:
i..;c_-'--'-'--.L.....L.._L-_...,_ _,._
I =
., _,_., ,_., '--'�

0.1 0.2 0.3 0.4 0.5 0.6 0.8 1.0

I 5678910
Particle-size ratio, D,. I D,ut

Figure 22.11. Cyclone efficiency.

622 APPLICATIONS TO EQUIPMENT DESIGN
CLEAN AIR OUTlET
I I I I
SLUDGE OUTLET
figure 22.12. Centrifugal washer. {Courtesy The
Oucon Co.)
and entrain them for removal at the bottom of the
scrubber. Figure 22.12 shows a cyclone scrubber in
which both centrifugal force and scrubbing action
work toward elimination of dirt. Many variations
of this type of scrubbing equipment are available.
Centrifugation
Centrifuges of the solid-bowl type are settling
devices that utilize a centrifugal field rather than a
gravity field to cause separation of the components
of liquid-solid or liquid-liquid systems. Gas-solid
systems are separated by the same effects in the
cyclone separator discussed earlier. The centrifugal
field causes particles of the heavier phase to "fall"
through the lighter phase away from the center of
rotation. This action is exactly what occurs under
the influence of the gravity field in classification.
Centrifuges. There are three main types of
centrifuges distinguishable by the centrifugal force
developed, the range of throughputs normally
obtained, and the solids concentration that can be
handled (20). The first of these is the tubular-bowl
centrifuge. This centrifuge rotates at high speeds
developing centrifugal forces of the order of
13,000 times the force of gravity, but it is built for
low capacities in the range between 50 and
500 gal/hr. Since it has no automatic solids
removal system, it can handle only small concen-
trations of solids. The second is the disk-bowl
centrifuge, which is larger than the tubular-bowl
centrifuge and rotates at slower speed, developing a
centrifugal force up to 7000 times gravity. 'This
centrifuge may be designed to handle as much as
5000 gal/hr of a feed containing moderate quanti-
ties of solids that are discharged continuously in a
concentrated stream. Both of these types of centri-
fuge are primarily designed to separate liquid-liquid
systems. However, the disk-bowl centrifuge can be
adapted to separate liquid-liquid-solid systems, or
liquid-solid systems where the major product is a
clarified liquid. The solid-bowl centrifugal, which
forms the third class, is primarily a solid-liquid
separator and operates like a. thickener (to be
described later). These centrifugals are built to
handle solids at rates up to 50 tons/hr.
The tubular-bowl centrifuge is shown in
Figure 22.13. It consists of a tubular bowl that
rotates within a housing. The bowl stands verti-
cally, hung from a thin flexible, solid shaft sup-
ported by a thrust bearing and is belt driven from
an electric motor or direct driven with an electric
motor or from an air or steam turbine. At the
bottom, the bowl is loosely guided by a spring-
supported bushing. The feed is delivered to a nozzle
at the bottom of the bowl and jets into the bowl
where it is quickly accelerated to bowl speed by
the action of light metal vanes loosely fitted into
the bow!. The heavy phase collects along the walls
of the bowl, the light phase forming a concentric
layer on the inside of the heavy phase. Droplets of
light liquid in the heavy phase move toward the
center of the bowl, and droplets of heavy iiquid
move toward the wal I of the bowl. The 2- to 5-ft
length of the bowl affords sufficient residence time
for the droplets to reach their proper phase layer.
The layers are maintained in the bowl, and their
separate discharge is contro!led by ring dams at the
top end of the bowl. There is no arrangement for
removing solids; if thev are present, they usually
build up on the walls of the bowl until the unit is
stopped and cleaned.
A cutaway view of a disk-bowl centrifuge is
shown in Figure 22.14. This type of centrifuge was
invented by Delaval in 1878 and has been widely
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 623
used since then for such diverse applications as
cream separation, catalyst separation, dehydration
of marine lubricating oils, and refining fish oils. In
the unit shown, feed, which may consist of two
liquid and one solid phases, enters from the
bottom and is accelerated to the centrifuge speed
as it is thrown out toward the perimeter of the
spinning disks. It feeds up into the disks through a
light liquid series of holes in these disks. Here the long flow
outlet path but short distance from disk to disk promotes
easy separation, with the heavy liquid flowing out
on the underside of the disks and the light liquid
flowing toward the center on the top surface of the
disk. Solids collect in the heavy liquid in the area
outside of the disk stack. They are discharged by
extrusion through a series of nozzles around the
perimeter of the spinning bowl along with some of
the heavy liquid. Light liquid and the bulk of the
heavy liquid overflow at the top of the bowl and
are discharged from the flanged pipes shown. In
cases where the feed contains only one liquid
phase, the feed is conducted to the outside of the
disks and flows in toward the hub between the
disks. The solids are then thrown into the outer
area and extruded out through the nozzles.
Centrifuges of the .disk-bowl type are oper-
ated at speeds up to 10,000 rpm producing centri-
fugal forces of 14,000 times gravity force g. They
Figure 22.13. Cutaway view of tubular-bowl centrifuge. offer capacities up to 30,000 gal/hr.

Figure 22. 14. Cutaway view of a disk-bowl centrifuge with top drive, bottom feed,
and peripheral nozzles for continuous solids discharge. (Courtesy Sharples-Stokes
Div., Pennwalt Corp.)
624 APPLICATIONS TO EQUIPMENT DESIGN
Figure 22.15. Typical cross-sectional view of solid bowl centrifugal, right-hand
design. (Courtesy Bird Machine Co., Inc.)
A cross-sectional view of a solid-bowl centri-
fugal is shown in Figure 22.15, and the action of
this centrifugal is diagrammed in Figure 22.16. The
major parts of this machine are a cylindrical bowl
with a truncated cone-shaped end and an internal
screw conveyor for solids that closely fits the
interior of the bowl. These parts rotate together,
but the screw conveyor rotates at a rate 1 or 2 rpm
below the rate ·of rotation of the bowl. In opera-
tion, the feed is admitted through the central
screw and enters the bowl about halfway along the
side. The centrifugal action forces both liquid and
solid phases to the walls of the cylinder. The solids,
being denser, concentrate against the walls and
along the bottom of the pool of liquid held in the
bow: by the position of the fi It rated ischarge parts.
The solids conveyor, however, has a net rotation
toward the small end of the bowl and scrapes the
solids from the wal Is of the cone toward the small
end.· .As the solids move in this direction, they may
be given a freshwater wash with water entering in
the same manner as the feed entered. They are
ultimately discharged at the small end of the coni-
cal bowl section.
These centrifugals are made with maximum
FEED
bowl diameters ranging between 4 and 54 in. The
54-in. machine handles up to 50 tons/hr of solids,
although this rate must be reduced if the particles
are particularly fine or if the liquid phase is vis-
cous. These machines may also be operated as
classifiers, in which case the feed rate and bowl
speed are adjusted so that satisfactorily small
particies will not settle out and leave with the
filtrate. This type of operation might occur with
the centrifugal accepting the product from a wet-
grinding step and feeding the large-particle solids
back to the grinder while the filtrate with its fine
particles passes on fer further processing. These
units develop a centrifugal force up to 3000 ti mes
that of gravity, with speeds up to 6000 rpm, so
that sharp separations can be made in the 1-micron
particle-size range.
Similar centrifugals are made with a perfor-
ated wall on the bowl. These centrifugals act
exactly !ike filters (to be described later) with the
filtrate draining throuqh the cake and bowl wall
into a surrounding collector. As with other centri-
fugal filters, they operate best with coarse-qranular
or coarse-crystalline, free-flowing solids.
Centrifuge Theory and Calculations-Rate of
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 6:5

-TORQl,£ CMJlLCWl
SWITCH

/ ORIV(f\. SHEAVE

=-£TO INLET

I EFF"U..JENT
SQ.JDS DtSO,ARC,E
DSCHAi=r.£

Figure 22.16. Cutaway view of a solid-bowl centrifugal showing internal flow of

liquid and solids phases. (Courtesy Bird Machine Co.)
Separation. The basic force balance around a the unique terminal velocity characteristic of its
particle falling in a centrifugal force field was given posrtron. Thus, for any position, dv/d8 = O; but,
earlier as for any instant in the movement of a sing=
particle, dv/dr is positive. Then considering .e
dv=rw2(p,-p)- Copv2S (22.7) particular position, dv/d8 = 0, and Equation 22.33
d8 p, 2m becomes

In the separation of phases by settling in either a

gravitational or a centrifugal field, the perfection rw2 (p, - p)Dp 2
v -- (22.34
of separation is limited by the rate of fall of the R - 18µ
smallest particles present. In most cases, these
particles fall at rates low enough that laminar flow where
exists and Co= 24/NR,, as shown by Figure 13.5.
Making this substitution in Equation 22.7 and con- vR = terminal falling velocity of spherica!
sidering spherical particles so that particles of diameter DP at radius r in a
centrifugal field rotating at rate w

The radial distance traveled by the particle may be

!'_ D 2
6 obtained by multiplying Equation 22.34 by tre
4 p
differential time (dO).
Then Equation 22. 7 becomes
rw2 (p, - p)D 2
VR d8 = dr = p d8 (22.35
dv = rw2 { p, - p)- 1 Bµv (22.33) 18µ
d8 \ o, p.,Dp 2
which, upon integration, gives
As particles move radially in a centrifugal field, the
field strength changes with their position. Con· v
sequently, the terminal velocity of the particles is a Q
function of radial position. In these developments
a particle at any position is considered to move at (22.36
 626 APPLICATIONS TO EQUIPMENT DESIGN
where where
V = volume of material held in the 2 1
(' - ' ) = effective value of (r2 - r, )Ir to
centrifuge r eff be used in Equation 22.37
0 = volumetric feed rate to the centrifuge when r2 - rv, the thickness of
V/0 = residence time of a particle in the the liquid layer in the centri-
centrifuge fuge, is not negligible compared
The diameter in Equation 22.36 is that of a to either r, or rz

particle falling from r1 to r2 during the residence
time available in the centrifuge. The significance of
this equation is perhaps most readily seen in rela-
tion to the parallel equation, which holds when the
liquid layer within the centrifuge is very narrow
A very useful characteristic of a centrifuge
can be derived by manipulating Equations 22.37
and 22.15. Equation 22.37 is solved for Q and the
gravitational constant inserted.

compared to the radius. In this case, the centrifugal _ (p, - p)gDP '2 Vw2r
field is considered constant, and Equation 22.35 a ----=2v, · :!:

may be written directly in terms of the residence 9µ g(r2-r,)

time (V/0). (22.39)
in which
(p, -
p)g0/2 (22.15)
Vt=
18µ

where where v, is the terminal settling velocity of a

x = radial distance traveled by a particle of particle in a gravitational field and
diameter OP in the residence time avail- 2
able Vw r
:!:= - (22.40)

If x is taken as half the thickness of the liquid layer ---

(r2 - r1 )/2, half the particles of some diameter Op'
will settle to the wall, whereas half of them will
still be in suspension when the fluid leaves the
centrifuge. DP', is the "critical diameter." Particles
of diameter larger than DP' will predominantly be
settled from the fluid phase, and particles of
diameter smaller than Op' will predominantly
remain in solution. Solving Equation 22.35a for
Op', and substituting (r2 - r, )/2 for x, the critical
particle diameter is
g(r2 - r,l
where :!: is a characteristic of the centrifuge itself
and not of the system being separated. The :!:
factor can then be used as a means of comparing
centrifuges (1 ). It is the cross-sectional area of a
settler that will remove particles down to the same
diameter as those separated in the centrifuge when
its volumetric feed rate equals that of the centri-
fuge. If two centrifuges are to perform the same
function,

a,= 02 (22.41)
9µ0 rz= r, :f.1 Z2
D '= ---- · -- (22.37)

p �(p, - p)<,.,2 v ,. The quantity :!: can be determined for com-

mercial centrifuges, although in some cases the
where determination requires approximation methods.
For the tubular-bowl centrifuge, applying Equation
r2 - r1 = thickness of the liquid layer 22.38 gives
OP'= critical particle diameter
1rw2/ (r/ - r, 2)
Comparison of this equation with Equation :!:=--,---- (22.42)

22.36 shows that for the case where the liquid- g

ln-
r/
layer thickness is great enough that the variation of r,2
- centrikgal tieid with radius must be considered, where I= bowl length

the effective value of (r2 - r1)lr is

For the disk-bowl centrifuge Ambler (1) gives
-- =21n- (22.38) 2n1r(r23 - r, 3)w2
:!:=------- (22.43)
rz - '') rz 3g-tan n
( r eff r1
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 6:!;
where Example 22.3. A liquid-detergent solution of
n = number of spaces between disks in
100-centipoise viscosity and 0.8 g/cm3 density
to be clarified of fine Na2S0 4 crystals (p, =
the stack

rz, r, = outer and inner radii of the disk stack 1.46 g/cm3) by centrifugation. Pilot runs in a
Q = conical half angle laboratory supercentrifuge operating a
23,000 rpm indicate that satisfactory clarificatio,:,
Table 22.1 gives values obtained for several types obtained at a throughput of 5 lb/hr of solutior.
of centrifuges based both upon calculations from This centrifuge has a bowl 7t
in. long intema.
geometry and upon laboratory and plant data. The with r2 = t in., and (r2 - r1) = 19/32 in.
table shows both the comparative performance of (a) Determine the critical particle diame-.2'!"
the centrifuges and the effectiveness of the for this separation.
machines in comparison to what is calculated on lb) If the separation is to be done in
the basis of theirqeometrv. Note how much more plant using a No. 2 disk centrifuge with 50 disks
effective the disk centrifuge is than any of the 45° half angle, what production rate could oe
other centrifuges listed. This is the result of the expected.
high residence time and short disk-to-disk separat- Solution. (a) The critical particle diamerer
ing distances built into this machine. (DP') can be determined directly from Equati
In the development of Equations 22.33 to 22.37. Since the liquid layer is thick, Equati
22.43 given above, the diameter (Dp) was defined 22.38 is used to get the effective value of lri - r,
as a particle diameter, with the inference being that
the particle would be solid. If a liquid-liquid separ- 5
9x100x2.42x( )x1728
ation is to be made, the mechanism is no different 62.4 x 0.8
from that of solid-liquid separations. Droplets --------------- x 2 -
62.4( 1.46 - 0.80) x (2,r x 23,000
of liquid rather than particles of solid migrate,
x 60)2 7 .75 x ,r(0.875 2 - 0.2812)
and they migrate across one liquid phase and
coalesce into the other phase instead of migrating Dp' = 0.0000033 ft or 1.08 microns
through the fluid phase to the wall. The rate of
migration is still calculable through use of
(b) Table 22.1 gives L = 1290 for the labor?-
Equation 22.37 properly modified for the centrifuge
tory supercentrifuge and L = 72,600 for the o. -
being used.

Table 22.1 COMPARATIVE PERFORMANCE OF CENTRIFUGES (I)

r Values, ft2

From Experimental Extrapolation 011

Calculuted from Data Clarifying Ideal Commercial Sys:ta:;:5
Georr.etry Systems (Supercentrifuqe TlSS.

Laboratory supercentiifuge (tubular bowl 1! in. I.D. x 7! in. long) operating at:
10.000 rpm 582 532 582'
16,000 rpm 1,485 ;,435
23,000 rpm 3,0:0 3,070 1,290
50,000 rpm 14,520 14,520 not used

No. 16 supercentrifuge (tubular bowl 4! in. 1.0. x 29 in. long) c,perating at:
15,000 rpm 27 .150 27 .150 27,150
No. 2 disk centrifuge, 1i in. r1 x 53 in. rz on disks
52 disks, 35° half angle, 6000 rpm 178,800 98,000 89 ,400 to 178,BOO
50 disks. 45° half angle 134,000 72,600 67.900 to 134.()00
Super-0-Cantor (solid-bowl centrifugal}
Pl�· 14 (conical bowl). 3250 rpm
ID= 14 in.-8 in., L = 23 ln.) 4,750 2,950 2,950·
PY-14 (cylindrical bowl). 3250 rpm
ID= 14 in., L = 23 in.) 8,940 5,980 5,980.

* For relatively low throughput rates.

628 APPLICATIONS TO EQUIPMENT DESIGN
disk centrifuge. Since these centrifuges are to per- final radius (r1). Upon simplification
form identical functions, Equation 22.41 gives

01 5
p= J'' pw2r dr (22.44)

02 = E, · �2 = x 72,600 = 281 lb/hr r; Ye

1290
which, upon integration, gives

Centrifuge Theory Calculations-Outlet Dam pw2

P= -- (r,2 - r;2) (22.45)
Settings. In liquid-liquid separations, the position 2gc
of the outlet dam becomes more important than it

is for solid-liquid separations, for, instead of Applying Equation 22.45 to the physical situation
merely controlling the volumetric holdup in the of Figure 22.17b where the pressure must be the
centrifuge and the critical particle diameter, the same on either side of the liquid-liquid interface at
position now also determines whether a separation r2 results in
can be made at all. Figure 22.17 shows the physical
2 2
situation in a centrifuge arranged for clarifying a Phw .., p1w
liquid phase of entrained solids and that in a centri- - - (r2· - r 4 2) = - - (r22 - r1 2) (22.46)

2 gc 2gc
fuge arranged to separate two liquid phases. In
Figure 22.17b the distances have the following or
significance:
r/-r/ Pt
(22.47)
r1 = radius to top of light-liquid layer r/ - r12 Ph

r2 = radius to liquid-liquid interface where

r3 = radius to outside edge of dam
r4 = radius to surface of heavy-liquid down- Pt= density of the I ight phase
stream from the dam. Ph = density of the heavy phase

The location of the interface is fixed by a balance
of forces arising from the hydraulic heads of the
two liquid layers. Expressing these forces as
pressures gives
J'f dP = J't dF = J't a dm = J'f p( w2 r)(2rrr/ dr)
r; r; A rt gcA rt
where the general case is used in which the liquid
layer extends from any inltial radius (r;) to any
In order for the centrifuge to separate the two
liquid phases, the liquid-liquid interface must be
located at a radius smaller than rs but greater than
that of the top of the overflow dam (r4 ). More-
over, it is usually found that one of the phases is
more difficult to clarify than the other. To com-
pensate for this, the volume of this phase must be
(2rrr/)gc made iarger than the volume of the easily clarified
phase. This can be done by adjusting the height of
the two overflow dams. Note that in the liquid-
clarifying centrifuge of Figure 22.17a, only one
dam is used, and the sole function of this dam is to

I�
control the volume of liquid rnaintained in the
centrifuge.

I !�""-��· Example 22.4 In the primary refining of

vegetable oils, the crude oil is partially saponified
I with caustic and the refined oil separated immedi-
ately from the resulting soap stock in a centrifuge.
I solids
In such a process, the oil
0.92 g/cm3 and a viscosity of 20 centipoises, and
has a density of

the soap phase has a density of 0.98 g/cm3 and a

viscosity of 300 centipoises. It is proposed
la) (bl to separate these phases in a tubular-bowl
centrifuge
Figure 22.17. Overflow-dam arrangements in a tubular· with a bowl 30 in. long and. 2 in. I.D. rotating at
bowl centrifuge. (a) Tubular-bowl centrifuqe arranged for 18,000 rpm. The radius of the dam over which the
clarifying a liquid of entrained solids. (b) Tubular-bowl light phase flows is 0.50 in., whereas that over
centrifuge arranged for separating two liquid phases. which the heavy phase flows is 0.510 in.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 629
(a) Determine the location of the liquid- Sedimentation
liquid interface within the centrifuge. The separation of a dilute slurry by gravity settling
(b) If this centrifuge is fed at a rate of into a clear fluid and a slurry of higher solids
50 gal/hr with feed containing 10 volume percent content is called sedimentation. The mechanism
soap phase, what is the critical droplet diameter of of sedimentation may be best described by obser-
oil held in the soap? vation of what occurs during a batch settling test as
solids settle from a slurry in a glass cylinder. Figure
Solution. (a) The interface location can be
22.18a shows a newly prepared slurry of a uniform
found directly from Equation 22.47.
concentration of uniform solid particles through-
0.92 rl - osio2 out the cylinder. As soon as the process starts, all
P1
particles begin to settle and are assumed to approach
Ph 0.98 rl - 0.5002 rapidly the terminal velocities under hindered-
settling conditions. Several zones of concentration
0.939(ri2 - 0.250) = rl - 0.260
will be established (Figure 22.18b). Zone D of
rl = 0.426, r2 = 0.654 in. settled solids will predominantly include the heavier
faster-settling particles. In a poorly defined
(b) The volume of soap phase held in the transition zone above the settled material, there are
centrifuge is channels through which fluid must rise. This fluid is
forced from zone D as it compresses. Zone C is a
30 x rr(12 - 0.426) region of variable size distribution and nonuniform
3
1728 =0.0313ft
concentration. Zone B is a uniform-concentration ·
zone, of approximately the same concentration and
This represents a residence time of distribution as initially. At the top of region B is a
boundary above which is clear liquid, region A. If
0.0313 x 3600
169 s=--
v the original slurry is closely sized with respect to
50 x 0.10/7.48 0 the smallest particles, the line between A and B is
sharp.
The critical droplet size, of which 50 percent will As sedimentation continues, the heights of
be separated from the soap phase in the 169-sec each zone vary as indicated in Figure 22.18b, c, d.
residence time, can be determined from Equation Note that both A and D grow larger at the expense
22.37. of B. Eventually, a point is reached where Band C

� 9 x (300 x 2.42) x 3600 1.0

2
Dp' � .J �.92)62.4 x (2rr x 18000 x 60)2 x 169 In 0.654
o;' = 26.6 x 1 o-6 ft or 8.1 microns

A
Clear-t-
liquid

8
�---j
r---- Fl
'J·
{<) {di 1,1
Figure 22.18. Batch sedimentation.
630 APPLICATIONS TO EQUIPMENT DESIGN
disappear and all the solids appear in D; this is
referred to as the critical settling point ( Figure
22.18e) -that is, the point at which a single
distinct interface forms between clear liquid and
sediment. The sedimentation process from this point
on consists of a slow compression of the solids, with
liquid from the boundary layer of each particle
being forced upward through the solids into the
clear zone. Settling rates are very slow in this dense
slurry. The final phase is an extreme case of
hindered settling. Equation 22.22 may be used to
estimate settling velocities. It accounts for the
effective density and viscosity of the fluid but does
not account for agglomeration of particles, so that
the calculated settling rate may be in considerable
error.
In a batch-sedimentation operation as
illustrated, the heights of the various zones vary
with time. The same zones will be present in
continuously operating equipment. However, once
steady state has been reached (where the slurry fed
per unit time to the thickener is equal to the rate
of sludge and clear liquor removal), the heights of
each zone will be constant. The zones are pictured
in Figure 22.19 for a continuous sedimentation.
Industrial sedimentation operations may be
carried out batchwise or continuously in equip-
ment called thickeners. The batch thickener oper-
ates exactly like the example cited above. The
equipment is nothing more than a cylindrical tank
with openings for a slurry feed and product draw-
off. The tank is filled with a dilute slurry, and the
slurry is permitted to settle. After a desired period
of time, clear liquid is decanted until sludge
appears in the draw-off. The sludge is removed
from the tank through a bottom opening as indi-
cated in Figure 22.20.
Continuous thickeners (Figure 22.21) are
large-diameter, shailow-depth tanks with slowly
revolving rakes for removing the sludqe. The slurry
is fed at the center of the tank. Around the top
edge of the tank is a clear liquid overflow. The
rakes serve to scrape the sludge toward the center
'lr-----f-------.<'," Clear ltQuor
overflow
Thickened sludge outlet
Figure 22.19. Settling zones in continuous
thickeners.
/
Feed Overflow
Lo, 3
ft /hr V, ft3/hr
3 c11,• OJ no solids)
Co, lb/ft
Underflow
Lu, ft3/hr
c,t, lb/ft
3
Figure 22.20. Schematic diagram of thickener.
of the bottom for discharge. The motion of the
rake also "stirs" only the sludge layer. This gentle
stirring aids in water removal from the sludge.
Continuous-Thickener Calculations. The
purpose of a continuous thickener is to take a slurry
of some initial concentration of solids and through
the process of sedimentation produce a slurry of
some higher concentration. The calculations
necessary for the design of a continuous thickener
are governed by the settling characteristics of the
solids in the slurry. The design of a thickener
requires a specification of a cross-sectional area and a
depth. It is possible through the use of batch-
settling information to design a unit to produce a
specified product in a continuous manner. The next
few paragraphs will indicate the calculation pro-
cedures.
Laboratory measurements of the type dis-
cussed earlier in connection with Figure 22.18 are
the most satisfactory tests for determining the
settling characteristics of a given slurry or sludge.
Although these are batch tests, their analysis is
useful for the design of continuous thickeners.
Figure 22.22 shows the results of this test where
the height of the liquid-solids interface is obtained
as a function ot time. Slopes of this curve at any
point of time represent settling velocities of the
suspension at that instant and are characteristic of
a specific solids concentration.
The, early portion of the .curve tends to be
linear corresponding to a constant-rate settling of
the sludge at the initial concentration. In thicken-
ing, this initial region represents a negligible part of
the total thickening time. As time increases, the
settling velocity decreases. Coe and Clevenger ( 14)
in 1916 and in 1952 Kynch (30) proposed a model
to explain this occurrence by assurning that the
settling velocity is proportional to the suspended
solids concentration. Once the constant settling
rate region of Figure 22.22 is passed, every point
on the curve corresponds to a different solids con-
centration.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
Figure 22.21. Continuous thickener. (Courtesy Dorr-Oliver, Inc.)
It must be emphasized that the settling curve
determined in the laboratory applies only to the
slurry tested and that the results may have some
uncertainties. For example, the largest-diameter
test cylinder possible should be used to minimize
wall effects. Tests should also be carried out at
depths comparable to the depth of the proposed
unit.
Design Calculations
Nearly all of the current methods for deslgn calcu-
lations stem from the work of Coe and Clevenger.
As mentioned earlier, their primary assumption
was that the rate of descent of the sohds-lrquid
Time, 8
Figure 22.22. Batch-settling results.
631
interface was a function of local concentration.
Batch-test results clearly show that the settling
velocity decreases with increasing concentration.
But the decrease is less rapid than the increase in
concentration. This balance of decreasing linear
velocity and increasing mass flow as a consequence
of greater density of solids in the compacting layer
is unpredictable but is important in the design
process.
A necessary condition tor functioning of a
continuous thickener is that the rate at which
solids settle through every zone must be at least
fast enough to accommodate the solid beinq
delivered to that level. In the upper part of the
thickener. the slurry is quite dilute, and settling is
ouite rapid. In the bottom, the density and solids
concentration are extremely high and settling
velocity is low.
In a thickener operating continuously, solids
pass through layers of all concentrations between
that of the feed and that of the final underflow, as
indicated in Figure 22.19. If solids entering any
layer do not pass through at a rate equal to that of
the arrival of solids into the layer, the layer will
increase in thickness, and it will grow upward until
solid ultimately appears in the overflow. Thickener
design is based upon identifying the concentration
of the layer having the lowest capacity for the
632 APPLICATIONS TO EQUIPMENT DESIGN
Figure 22.23. Selected concentration layer
in thickener.
passage of solids through it under operating condi-
tions. This particular layer is called the rate-
limiting layer. Sufficient area must be provided to
ensure that the specified solids flux does not
exceed the capacity of the rate-limiting area.
Figure 22.23 depicts such a layer of concen-
tration c in a batch test. This layer is assumed to be
the rate-I imiting one, so it may be viewed as rising
at a velocity vc. Solids settle into this layer from
just above, having concentration (c - de) and
velocity (v + dv) with respect to the column and
(v + dv + iiL I with respect to the layer. Solids settle
out of this layer at a velocity v with respect to the
walls and v + vL with respect to the layer. If the layer
is assumed to have a constant concentration of
solids, then by material balance
(c -dc)S8(v+dv + v c I= cS8(v+ iid
(22.48)
where S is the area normal to the solids· flow.
Solving Equation 22.48 for iiL gives
dv
iiL = c--v-dv (22.49)
de
and by neglecting dv,
dv
VL = c- - v (22.50)
de
It has been assumed (14, 30) that settling
velocity is a function of concentration (that is,
v = f(c)), and it follows that dv/dc = f(c). Equation
22.50 therefore becomes
v c =cf(c)-f(c) (22.51)
Since c is constant for this layer, f (c) and f(c)
are also constant, and ii L is therefore also constant.
The constancy of iiL in the rate-limiting zone may
be used to determine the concentration of solids at
the upper boundary of the layer from a single
batch-settling test. Let c0 and z0 represent the
initial concentration and height of the suspended
solids in a batch-settling test. The total weight of
the solids in the slurry is cLzoS, where S is the
cross-sectional area of the cylinder in which the
test is being performed. If a limiting layer exists, it
must first form at the bottom and move upward to
the clear-liquid interface. If the concentration of
the limiting layer is cL and the time for it to reach
the interface is BL, the quantity of solids passing
through this layer is cL SB L (vL + ii L). This quantity
must equal the total solids present, since the layer
having this limiting· concentration started forming
at the bottom and moved upward to the interface.
Therefore
cLS8dvL+iid=c0z0S - (22.52)
If zt is the height of the interface at OL with
iiL being constant in accord with Equation 22.51,
then
(22.53)
Substituting the value of iiL from Equation
22.53 into Equation 22.52 and simplifying gives
CL= (22.54)
zt+vL8L
The laboratory-test data may be treated by
plotting the height of interface as a function of
time, as in Figure 22.24. From this plot, the value
of ii L is the slope of the curve at 8 = 8 L as shown
by Equation 22.55. The tangent with the curve at
8 L intersects the ordinate at Z;. The slope of this
line is ·
Z;-ZL
---;: VL (22.55)
BL
or
(22.56)
,,
Time {Bl
Figure 22.24. Batch-settling results.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 633

Combining Equations 22.56 and 22.54 yields

CLZ; = CoZo (22.57)

It follows, therefore, that z, is the height the slurry

would occupy if all the solids present were at
concentration cL . In terms of the model postulated
above, cL is the minimum concentration at which
boundary layers interfere.
The settling velocity as a function of concen-
tration determined from a single settling test as
follows. For several arbitrary B's, the slope of the
tangent and its intercept at B = 0 are determined
from a plot of z as a function of B. The value of
the intercept is used in Equation 22.57 to obtain 2 6 8
Time, hr
10 12
" 16

the corresponding concentration. From this, vL as Figure 22.25. Height of interface as a function of time for
a function of c is obtained.
Example 22.5.

Example 22.5. A single batch-settling test velocity at that time is the slope of the curve,
was made on a limestone slurry. The interface dz/dB = v = 2.78 cm/hr, and c = 425 gm/I. Other
between clear liquid and suspended solids was points are obtained in the same way, tabulated in
observed as a function of time, and the results are Table 22.2, and plotted in Figure 22.26.
tabulated below. The test was made using 236 g of
limestone per liter of slurry. Prepare a curve show- Table 22.2 SOLUTION TO EXAMPLE 22.5
ing the relationship between settling rate and solids
concentration.
8, hr z;. cm v, cm/hr c, g/1

0.5 36 15.65 236

TEST DATA
1.0 36 15.65 236
1.5 23.8 5.00 358
Time, hr Height of Interface, cm 2.0 20 2.78 425
3.0 16.2 1.27 525
a 36.0 4.0 14.2 0.646 600
0.25 32.4 8.0 11.9 0.158 714
o.so 28.6
1.00 21.0
1.75 Thickener Area
14.7
3.0 12.3 The required thickener area is fixed by the layer
4.75 11.55 requiring the maximum area to pass a unit quantity
12.0 9.8 of solids. Therefore, it is necessary to determine
20.0 8.6 the concentration for such a layer.
Consider the batch-settling test in which the
total flux of solids is the sum of the flux of the
Solution. Using the test data, the height of settling particles and the flux to provide bulk flow
the interface (z) is plotted as a function of time (0) to the underflow. This total flux is
(Figure 22.25). From the solid concentration of the
L"c
u A
initial slurry, F = F8 +F = cv + -- (22.58)

c0z0 = 236 x 36 = 8500 g cm/I where

From Equation 22.57, c � layer composition
v = settling velocity of the solids at c
8500 Lu = underflow volumetric rate
c=--g/1

Z; , A = cross-sectional area normal to f-lux

The tangent to the curve at B = 2 hr is found The first term on the right of Equation 22.58
to have an intercept of z, = 20 cm. The settling is the flux of the settling particles called the batch
 634 APPLICATIONS TO EQUIPMENT DESIGN

16
'
I I
14

12
I I

I
-
\1 - -� -
Cr.
Solids concentration
c,
Figure 22.27. Solids flux for continuous
\ thickener.

:\1 The area calculated from Equation 22.59 is

. 2
-. I the rrururnurn area that must be provided for a
continuous thickener operating at steady state.

Example 22.6. For the batch data of

��-
Example 22.5, determine the area and required
underflow rate for a feed rate of 3. 785 x 106
0 liters/day (1 x 106 gal/day). Feed concentration is
200 300 400
I 500 600 700 800
236 g/1 (grams/liter) and the thickened sludge is
cl, Solids concentration, g/t
specified at 700 g/1. The underflow velocity is to
Figure 22.26. Settling-rate-concentration relationship, for be 500 cm/day.
Example 22. 5.
Solution. The solution is shown in Table 22.3.
flux and is obtained from the batch test data. The cm day g
second term is the flux associated with the with- Fu = CVu = 500 - X -- X C - X ----
drawal of solids due to the underflow. A plot of day 24 hr I 1000 cm3

g
Equation 22.58 showing the relative contributions 0.0208c, -2 -
of the two fluxes is shown in Figure 22.27. cm hr
In Figure 22.27 the solids flux FL at concen-

tration cL is less than solids flux at other concen- The total flux, F = F8 + Fu, is plotted against
trations between the feed and underflow concen- composition in Figure 22.28. From Figure 22.28,
trations. Therefore, FL is the limiting solid flux FL = 11.5 g/cm2-hr. Thus,
capacity, and an area must be provided so that the 6 -a)
day- ){2-3-
3. 785 x 1 os ·-I - x -
flux does not exceed FL. Thus, the required area (
L0cc , day 24 hr , I
foe the thickener is A=--=

Loco FL 11.5 g/crn2-hr

A=-- (22.59)
h A = 3.24 x 106 cm2 (= 3483 tt2)

Table 22.3 SOLUTION TO EXAMPLE 22.6

c, g/1* v, cm/nr " Fe, g/cm2-hr F, g/cm2·hr

236 15.7 15.71 4.91 20.6

358 5.0 17.9 7.45 25.4
425 2.78 11.8 8.84 20.6
525 1.27 0.667 10.90 11.6
600 €).646 0.388 12.5 ·12.4
714 0.158 o.113 14.9 15.1

"From Table 22.2.

 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 635

26 section of this angle bisector line with the settling

24
curve gives an estimate of the time Be at which the
solids enter the compression zone, and the con-
22 centration ate c is Cc.
During the test, the mass of solids in the
20 cylinder is constant. A material balance on the
18
solids yields
(22.60)
o 16
or
1' 14 Cozo = Cozo = CuZu (22.61)

where the subscript u denotes underflow values. If

j 12
-c
'6 IO
·
FL

---, I
I
the thickening process is thought of as going from
concentration Cc to Cu, an amount of water must
"' I be removed to accommodate this change in con-
I centration. The volume of water that is squeezed
I
I
I out is
I
I (22.62)
I
I and the time required to discharge this volume of
I
I
I water is (Bu - Be). So the volumetric flow is
L = V A (Ze - Zu)
o������������_c_��
0 200 400 600 BOO (Bu - Be) (Ou - Oe)
Concentration. g/l
and solving for (e u - Oe)
Figure 22.28. Solution to Example 22.6.

A(ze-Zu)
Wu-Oe)= L (22.63)
Use of the above procedure is straightforward,

but it may be of interest to know at what time, From Figure 22.29, the settling velocity at ee
and at what concentration, this critical flux is is obtained from the slope of the curve ate c , or
realized. Talmadge and Fitch (55) presented a
method for this calculation, as well as a means to z, -zc
ve = --
Be
(22.64)
determine the minimum area.
The batch cylinder test is again the basis for Under continuous flow conditions, the clear
the procedure. At the start of the test, the solids liquid upflow cannot exceed ve if thickening is to
are uniformly dispersed throughout the cylinder at occur. Therefore, the flow at Be when thickening
a concentration c0• The total mass of solids in the starts is
cylinder is c0Az0, where A is the cross-sectional
area of the cylinder and z0 is the initial height of (22.65)
the interface, which is, in this case, the liquid
depth. Let ee be the time needed to reach the
critical concentration c2. Physically, this concen-
tration is reached at a time when the settling
valocity decreases rapidly.
Once again, the test data are plotted as in
Figure 22.21. This curve is used to determine 62
and c2 in the following way. The early portion of
the curve represents "free settling" at neariy con-
stant velocity. Draw a tangent to this portion of
the ·curve. At the very end of the test, where
concentrations are high and velocities low, the
curve also shows nearly constant velocity. Draw a
tangent to th is portion of the curve. The two Time.e
tangents are extended until they intersect. At the Figure 22.29. Determination of critical concen-
intersection construct the angle bisector. The inter- tration.
636 APPLICATIONS TO EQUIPMENT DESIGN
Substituting L" from Equation 22.65 into Equation 60
22.63 gives
� 50
(22.66) 1i
Ou - Oc Oc � 40
.� 30
Equation 22.66 is used to calculate Ou from Figure
22.29 as follows: zu can be calculated from Equa- � 20 le - -
tion 22.60. "'
� 10

(22.67)
o���
0
-��� ----�-�-�
8 IO 12 14 16 18 20 22
Time.mm
On Figure 22.29 locate Zu, and Ou can be calcu-
lated from Equation 22.66 or obtained from the A horizontal line through Zu = 11.7 intersecting
plot where the horizontal line through Zu intersects the tangent to the settling curve at Oc = 6.5 min
the tangent to the curve through Oc. gives Ou = 11.2 min. So,
The Talmadge and Fitch procedure applied to Lo Ou
the batch-test analysis described above is used to A=--
zo
determine the minimum area for a continuous
thickener. The average mass rate at which the layer
of concentration Cu forms in the thickener is
6
(4.5 x 10 -day
1
-)l/ 1440day.min )(11.2 min)(1000 cmI
3
)

CuZuAl8u. But, since CuZu = cozo according to 51 cm

Equation 22.60, the average mass rate 'is c0z0Al0u.
For continuous operations, the rate at which the = 6.92 x 105 cm2 (7 43 ft2)
layer of concentration Cu is formed must equal Thickener Depth. Comings ( 15), in a series
that at which solids enter the unit. of tests on a continuous thickener, determined the
L0co = c0zoAIO u effect of underflow rate on thickener capacity.
or Essentially clear overflow was obtained from all
LoOu runs. The depth of the thickening zone increased as
A=-- (22.68) the underflow rate was decreased. He concluded
zo that the depth of the thickening zone for incom-
Example 22.7. A biological sludge from a pressible slurries is less important than the reten-.
secondary waste treatment facility is to be concen- tion time of the particles within the thickener.
trated from 2500 mg/I to 10,900 mg/I in a continu- That is, the area of a thickener may be dependably
predicted from the settling limit described above
ous thickener. Flow to the unit is 4.5 x 106 I/day.
and from the volume of the compression zone
Determine the area required from the batch-test
determined to give necessary retention time. The
data given in Table 22.4.
retention time may be determined from laboratory
Solution. The test data are plotted as shown data in a batch test. Therefore, since depths repre-
in the accompanying illustration. Tangents are sentative of those expected in the fuli-scale
drawn and angle bisector constructed as described thickener were used in the batch tests, no separate
above, from which Zc = 20 and Oc = 6.5 min. The calculatlon for the thickened-sludge volume is re-
tangent to the curve at Oc gives a value of z1 = quired.
32.5 crn. in addition to the depth of the settling zone,
From Equation 22.67, some provision must be made for inventory and
mechanical design features. The total depth of the
z0c0 (51)(2500) thickener may be estimated 17) by adding to the
Zu = -- = ---- = 11. 7 cm
Cu (10900) depth of the settling and thickening zones.

Table 22.4 TEST DATA

0, min 0 2 3 5 8 12 16 20 25

Height of
interface, cm 51 43.5 37.0 30.6 23.0 17.9 14.3 12.2 11.2 10.7
 PARTICULATE SOLIDS FLOW AND'SEPARATION THROUGH FLUID MECHANICS
Bottom pitch 1 to 2 ft
Storage capacity 1 to 2 ft
Submergence of feei:l 1 to 3 ft
It is important to note that if the required
underflow concentration is less than the critical
value determined from the batch test, area calcula-
tions are based upon the unhindered settling rates.
APPLICATIONS OF THE MECHANICS OF FLOW OF
FLUIDS THROUGH PARTICULATE SOLIDS
In many industrial operations, a flu id phase
flows through a particulate-solid phase. Examples
include filtration, heat transfer in regenerators and
pebble heaters, mass transfer in packed columns,
chemical reactions using solid catalysts, adsorption,
and flow of oil through the reservoir toward an oil
well. In many cases, the solid phase is stationary, as
it is in a packed distillation column; in some cases,
the bed moves countercurrent to the gas stream, as
it does in a pebble heater or in some catalytic
reactors. In some cases, the fluid velocity is great
enough that the momentum transferred from the
fluid to the solid particles balances the opposing
gravitational force on the particles and the bed
expands into a fluidlike phase; in still other appli-
cations, the fluid phase carries the solid phase with
it, as it does in pneumatic conveying.
The rate of momentum transfer from the fluid
to the solid particles, and therefore the pressure
drop for flow through the bed, is related to the
physical mechanisms by which flow occurs. In a
packed bed, the flow path is made up of many
parallel and interconnecting channels. The channels
are not of fixed diameter but widen and narrow
repeatedly, and even twist and turn in varying
directions as the particles obstruct the passageway.
The channels do not even have the same average
cross section er total length. In flowing through
these passaqes, the fluid phase is rspeated'v acceler-
ated and decelerated and experiences repeated
kinetic-energy losses. In addition, the rough sur-
faces of the particles produce the usuai form-drag
and skin-friction losses. The flow through large
open channe s will be at higher velocity than the
1
flow through parallel narrow constricted channels,
because the pressure drop per foot of bed length
must be constant regardless of the channel under
consideration, For this reason, the transition from
laminar to turbulent flow will occur at a much
lower bulk flow rate in open passages than it will in
restricted channels. Moreover, at the convergence
of two channels, eddy currents and turbulence will
be promoted because of the inequality of velocity
637
in the two channels. The flow behavior in
expanded, or fluidized, beds will be very similar to
that for packed beds, except that the flow passages
will be more open and almost continuously inter-
connected. In cases where the bed is falling down
through the rising fluid or where the fluid is con-
veying the solid phase, the mechanism is more like
that of the hindered settling of particles through
the fluid which has been treated above.
It is apparent from the physical mechanism
discussed above that the momentum transfer from
fluid to particles arises from a form-drag and a
kinetic-€nergy loss, that is,
(7gc)total::: (rgc)form drag + (rgclkineticenergy
and skin friction
(22.69)
At low rates of flow through very small passages,
the kinetic-energy losses are small compared to the
form-drag losses, but for high rates of flow through
large passages or fluidized beds, the kinetic-energy
losses may completely overshadow the form-drag
losses. However, the transition will not be sharp,
because the wide variety of parallel passages
permits a wide variety of flow conditions.
Flow Through Packed Beds
Fluids are forced to flow through stationary beds
of particulate or porous solids in a wide range of
practical situations including moisture assimilation
by soi Is, adsorption, ion exchange, and many of
the examples noted in the introductory section. In
these applications, the form-drag losses can be
related to the flow conditions through modifica-
tion of the relations previously given for friction
losses in ducts. For this case,
t=
2(-t.P),gcD
_2, =¢(NR,)=q,
(fJiip)
--
v PL 1 µ
(22.70)
where -c.P, = pressure drop due to friction and
8(rygc),
f=--- (13.9a)
;,;2p
have been previously presented. For noncircular
ducts, an equivalent diameter must be used.
4S
D b
=-
eq (13.201
This defines the equivalent diameter of a duct as
four times the cross-sectional area divided by the
wetted perimeter. In terms of a packed bed, Equa-
638 APPLICATIONS TO EQUIPMENT DESIGN

tion 13.20 can be modified to In the region of low flow rate and small
particle size or, in other words, at low NA e
4S L total volume of voids (laminar flow), Equation 22.74 can be used to

-x- = -----------
b L total surface area of particles express the entire pressure drop, because the
4ENVP kinetic-energy losses are small. Under these condi-
(13.20a) tions, the drag coefficient and hence the friction
(1-E) · NAP factor are inversely proportional to the Reynolds
where number.
porosity, fraction of total volume that
E =
(13.10)
is void
N = number of particles
NVP = total volume of solid particles where k1 is a proportionality constant.
NAP= total surface area of solid particles Combining this equation with Equation 22.74
yields
Inserting Equation 13.29a into Equation 22.70, c3 (-!:,.P),gcDp (1 - E)µ
----2-� = k2 --- (22.75)
(1 - E)v, pl Dpv,p

which rearranges into

It is usually convenient to express this equation in

terms of the superficial velocity, that which would
occur with the actual mass-flow rate flowing
through the container holding the bed if this con-
tainer were empty. By definition, then, v, = eo ,
where v, is the superficial velocity. Also, it is
common practice to replace the VP and Ap terms
with a "particle diameter." The particle diameter is
defined in terms of the diameter of an equivalent
sphere. (See Appendix B.)
For a sphere,
Ap = ,r(O,p )2
Vp ,r(D,,)3/6
from which
where O,p = diameter of a spherical particle
For an irregular particle of known Ap to VP ratio,
there is only one size of sphere having this same
ratio. The diameter of this sphere is taken as
characteristic of the particle.
6
DP =-=D
Ap
sp
Vp
(22.76)
where k , and k2 are constants. Equation 22.76 is
known as the Carman-Kozeny equation and has
been successfully used to calculate pressure drop
for laminar flow through packed beds. It was
originally derived by Kozeny (28), who used the
simplified model of a number of parallel capillary
tubes of equal length and diameter to describe the
packed bed. If this model were exact, k 1 would be
64, as given by Equation 13.16. Carman (13)
applied this equation to experimental results on
flow through packed beds and found that k2 =
180.
If the Ne « is high, kinetic-energy losses
become significant. These losses may be found by
(22.72) modifying the kinetic-energy term in the general
relation, Equation 20.16a.
-&. ii2
--=-
P 2gc
But, if the energy loss is to occur repeatedly in a
unit channel length,
(22.73) (-1'.P)kgc n _
L =2pv 2 (22.77)
where

Writing Equation 22. 71 in terms of v, and Dp gives n = number of repetitive kinetic-energy

3 losses in a unit lenqth
8(-t,.P)rgcE Dp ( 4Dp v,p ) -!:,.Pk = pressure drop due to kinetic-energy
f= =¢ (22.74)
6v/pL(1 - E) 6(1 - E)µ losses
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 639

In the channels under consideration, the expan- Equation 22.81 and the superficial velocity, (v,)
sions in channel width probably occur at distances may be inserted into Equation 22.78.
roughly equivalent to the channel diameter. This PVs2
follows because the particulate bed has been

replaced by a model consisting of many parallel (-l!,.P)kgc E2 pv/ (1 - E)

circular ducts. The diameter of these ducts will be
---- = k3--- = k4 -- . ---
L

proportional to the particle diameter, and the fact

4eDP Dp e3
of one expansion occurring for each particle is thus
approximated. Using n proportional to 1/0c gives 6(1 - E) (22.82)

(-fiP)kgc = k3pv2 This relation was first derived by Burke and

(22.78) Plummer (11) to express the pressure drop result-
L De ing from turbulent flow through packed beds.

were k3 is a constant and De is the channel Equation 22.76 for the pressure drop caused by
diameter. As with the pressure drop due to form form drag and Equation 22.82 for the pressure
drag, it is convenient to convert this equation to an drop caused by kinetic-energy losses may now be
expression in terms of DP and v,. As before, added to obtain the total pressure drop resulting
v, from flow through the bed (18, 36).
17=-
E
(1 - E)2 µv,
�-Op
(-fiP)gc (1 - E) pv/
--L--=k2-E_3_ 0/ +k4

(22.73) (22.83)
and This equation has been developed under the
assumption that the velocity is constant through-
(22.79) out the length of the bed. With gas flow at high
2
L7fD- ( 1-E )
- pressure drop, this wou Id not be the case, and a
4 differential form of Equation 22.83 would have to
be written and integrated for the full bed depth. If
where
such an integration is carried out by assuming
Ne = number of channels in the area of the
isothermal expansion of an ideal gas, Equation
bed
22.83 becomes
D = bed diameter
(-fiP)ge (1 - E)2 µv,m (1 - E) Gv,m
In Equation 22.79 the bed has again been likened ---=k -----+k ----
L 2 €3 Dp 2 4 E3 Dp
to a large number of capillary tubes of fixed length
and diameter. The particle surface area (Ap) has (22.84)
been taken as equal to the wall surface of these where
capillaries. Also,
v,m = superficial velocity at the average of
s, «o; 2 nD2 inlet and outlet pressure
---=--·c (22.80)
4 4 G = mass rate of flow (pv,) based upon
total bed cross section
so that
Equation 22.84 can be rearranged into an equation
NeD/
D 2 =-- (22.80a) relating a modified friction factor and a NR, as
e follows ( 17):

Combining Equations 22.79 and 22.80.a gives (-fiP)gc 0/

--------=k 2 +k 4 --
E
3
NA,

Ap = A_e_D_e L µv,mi1-e)2 (1-e)

Vp NeD/ Dc(1 - E) (22.85)

--· (1-E)

4E Here, the term on the left side of the equation is an

(22.81) averaged friction factor modified by the inclusion
640 APPLICATIONS TO EQUIPMENT DESIGN
of the porosity terms. Using this relation, the con- of these equations. Typical units might be as
stants k2 and k4 can be readily determined from follows:
experimental data by planing
-L'J' = pressure drop through the packed bed,
D/ E3
(-L'J')
--a------
lb,/ft2
L = bed length, ft
L s c µv,ff (1 - E)2
DP = particle diameter, ft
as a function of NR ,!( 1 - E). Since Equation p = fluid density, lb/ft3
22.85 is linear in these groups of variables, straight v,m = superficial velocity at a density aver-
lines result where the slope is k4 and the intercept aged between inlet and outlet con-
at NR ,/( 1 - E) = 0 is k2. A large amount of experi- ditions, ft/s
mental data on flow through beds of granular e = bed porosity, dimensionless
solids show that k2 = 150 and k4 = 1.75 (17). The NR, = average Reynolds number based upon
modified friction-factor term of Equation 22.85 is superficial velocity (Dpv,mp/µ). dirnen-
the ratio of the total pressure drop to the pressure sionless
drop due to form drag and skin friction. Usually,
For liquids, v,m may be taken at any point in the
the friction factor is given as the ratio of the total
bed, but gas-density variations require the use of
pressure drop to the kinetic-energy losses. Re-
the average velocity as specified above.
arranging Equation 22.85 in this way by dividing In Figure 22 .30 is shown the mean curve
by Wpv,m Pm/µ)/( 1 - el and inserting the values obtained when experimental data from several
of k2 and k4 found experimentally gives sources ( 11, 17, 43, 45) are plotted as
(-L'J')gc Dp E3 (1 - e) [ (-!:,.P)gcDp/Lpv,m 2] [e3 ( 1 - e)]
-------= 150--+1.75
L pV,m (1 - e) NR, as a function of [NR ,/(1 - e)]. The solid line is a
(22.86) plot of Equation. 22.86, and the dotted lines·
marked Carman-Kozeny and Burke-Plummer are
Both Equation 22.85 and Equation 22.86 are plots of Equations 22.76 and 22.82, respectively.
dimensionless, and hence any set of consistent The data points themselves have been removed for
units may be used throughout any complete term clarity. They scatter smoothly around the line

!Qr-
8-
- 6 t:-
-
-
4L
" 2r

1.0 '-----..L--'---.Ll....LJw..! _L L_ _:__.LLJ_LlJ

IC 'Z 3 4 567810 20 40 e: 100 200 400 1000

NH,•
Figure 22.30. Pressure drop for flow through packed beds.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 641
plotting Equation 22.86 showing no systematic bed porosity (E), and the mean superficial velocity
deviation, and a maximum deviation of about 20 must first be determined.
percent if a few erratic points are eliminated. This For the !-in. cubes,
is excellent agreement in view of the wide range of
variables covered by Figure 22.30. Note the simi- 6Vp 6 x 0.253
D = -- = = 0.25 in.
larity in shape between the curve shown here and P Ap 6 x 0.25 2
the drag-coefficient-Reynolds-number plot The porosity can be determined using Figure
(Figure 13.5). The gradual transition between 8-12, Appendix B, which gives porosity as a function
laminar and turbulent regions, which covers a range of sphericity (\ii). By Equation 8-26,
of NA,/( 1 - e) between 5 and 2000, results from
the widely differing flow passages through the bed, 3
with the resulting wide variations. in actual velo-
city.
\i,=---
6
1T( :Pr "(;t 3
(0.253)2/3
Other relations for the pressure drop through
Ap 6 x 0.252
packed beds have been found by empirically corre-
lating experimental data. In this way, Leva and - ("3/2 . 6)2/3 -
Grummer (36) obtained - 53/2 • ,r - 0.806

0.0243G'-9 µ0· 1 A,., (1 - E) From Figure 8-12, this gives

--L-- = DP,., p _E_3_
E = 0.44
(22.87)
Fundamentally, v,m must depend upon the
pressure drop, since the gas specific volume is
to which they applied different constants for
pressure dependent. Here, however, as in many
particles of different surface roughness. Here A is a
pressure-drop computations, the pressure drop will
shape factor obtained by the relation probably be small relative to the total pressure, so
that the effect of this -tlf' on gas density will be
small. With this simplification, v, 1 and v,2 can be
(22.88) fixed directly. Otherwise, a trial-and-error solution
would be necessary to find v2• Ori this basis,
Leva's experimental data were all in the range of
PM i 100x29
relatively high NA, and relatively large particles.
P1 =--= 0.500 lb/ft3
Data on both liquid and gaseous fluids were in- 1 RT _� 10. 73 x 540
cluded in the correlation so that G can be either ---.....,._
PM 100x29
Gv or GL. 0.314 lb/ft3
Brownell and Katz (10) correlated results P2 = RT2 = 10.73 x 860
obtained mainly for very fine particles by using the
and
standard f versus NA, plot by multiplying f and
NA, by factors dependent upon the particle spheri- 1000 1000
--+---
city and the bed porosity.
0.500 0.314 I
v,m = = 2595 ft hr
2
Example 22.8. A bed of ,i-in. cubes is to be
used as packing for a regenerative heater. The Us,ng· these values, the group NA,/( 1 - e) can be
cubes are poured into the cylindrical shell of the calculated, and the modified friction factor can be
regenerator to a depth of 10 ft. If air flows through read from Figure 22.30.
this bed, entering at 80° F and 100 psia, leaving at
400°F, and flowing at a mass rate of 1000 lb/hr ft2 DpG
of free cross section, determine the pressure drop
across the bed. NA,= � = 0.25 Y. 1000 =
669
1 -E 1 -E 12(1 - 0.44) X (0.023 X 2.42)
Solution. This problem can be solved by
applying Equations 22.85 and 22.86 or Figure (-t;.P)gcDp E3
22.30. In any case, the particle diameter (DP). the --- = 2.00 (from Figure 22.30)
Lpm v,m 2 1 - E
642 APPLICATIONS TO EQUIPMENT DESIGN
This equation may be solved for (-tJ.P). Here,
Pm v,m is replaced with G.
2.00 x 1 ox 1 ooo x 2595 x 0.56 I
-tJ.P = = 40.4 lb, ft 2
_ 0.25 -
(32.2 x 36002 ) x U x 0.443
-tJ.P = 0.28 psi
The assumption of a relatively small -tJ.P has then
been justified for this case. Larger relative pressure
drops can be expected if the particle diameter is
small.
Bed Fluidization
The packed bed expands when the pressure drop
due to the upward flow of fluid through a granular
unrestricted bed equals the weight of the packing.
As the bed expands, it retains its top horizontal
surface with the fluid passing through the bed
much as it did when the bed was stationary. Now,
however, the porosity is much greater, and the
individual particles move under the influence of
the passing fluid. The bed has many of the appear-
ances of a boiling liquid and is referred to as being
"fluidized." Writing the force balance on a section
of bed of length L when the pressure drop equals
the gravitational force gives
(-tJ.P)gc
---= (1 - e)(p,-p) · g (22.89)
where p, = solid-particle density. The extent of bed
expansion may be obtained by eliminating
(-tJ.P)gciL between Equations 22.84 and 22.89.
(1 - e) µv,m Pm v,m2
150----+1.75---=(p -p)g
e3 D/ o, s
(22.90)
Since pressure affects the density and hence the
velocity of a gas in the fluid phase, this equation
should be solved riqorouslv bv successively insert-
ing v, varying between inlet and outlet. The bed
porosity would show equivalent gradation. This
effect is usually small, and the use of mean values
of p and v, is usually satisfactory. In most indus-
trial applications involving expanded or fluidized
beds, the particle diamet€r is small, and v,m is also
sma!I. in these cases, the second term of Equation
22.90 is negligible compared to the first, so that
150µv,m e3
(22.91)
(p, - p)gD/ 1 -e
For a given bed, Equation 22.91 may be written
for both the unexpanded and the expanded state at
the pressure drop required for bed expansion.
From the difference in these two states,
150µ
0)=-----
€3 eO'
-----(v -V
g(p, _ p)O/ sm sm 1 _€ 1 _ €0
(22.92)
where the superscript O refers to the condition
where the fluid velocity is just insufficient to
expand the bed. Equation 22.92 relates the bed
porosity and therefore the extent of bed expansion
to the mean superficial velocity of fluid passing
through the bed. It applies only where the pressure
drop is constant and balances the force of gravity
acting on the solid particles and where Dp is small.
The pressure-drop behavior of a packed bed as
the velocity of flow up through it increases is
illustrated in Figure 22.31. Between points A and
8, the bed is stable, and the pressure drop and
Reynolds number are related by Equation 22.86.
At point 8, the pressure drop essentially balances
the bed solids weight. Between points 8 and C the
bed is unstable, and particles adjust their position
to present as little resistance to flow as possible. At
point C, the loosest possible arrangement is
obtained in which the particles are in contact.
Beyond this point, the particles begin to move
freely but collide frequently so that the motion is
similar to that of particles in hindered settling.
Point C is referred to as the "point of fluidiza-
tion." By the time point D is reached, the particles
are all in motion, beyond this point, increases in
c
IL _
log N 1 ,.,,
Figure 22.31. Fluidization of a bed of par·
ticulate solids (61 ). (By permission of Chem.
Eng. Prog., copyright© 1948.)
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
NR e result in very small increases in -1:,.P as the
bed continues to expand and the particles move in
more rapid and more independent motion. Ulti-
mately, the particles will stream with the fluid, and
the bed will cease to exist. This occurs at point E.
Two main types of fluidization have been
noted experimentally (61). In cases where the fluid
and solid densities are not too different, where the
particles are small, and therefore where the velo-
city of flow is low, the bed fluidizes evenly with
each particle moving individually through a rela-
tively uniform mean free path. The dense phase has
many of the characteristics of a liquid. This is called
particulate f/uidization. Where the fluid and solid
densities are greatly different or the particles are
large, the velocity of flow must be relatively high.
In this case, fluidization is uneven, and the fluid
passes through the bed mainly in large bubbles.
These bubbles burst at the surface spraying solid
particles above the bed. Here, the bed has many of
the characteristics of a liquid with the fluid phase
acting as a gas bubbling through it. This is called
aggregative fluidization. In addition to these two.
main classifications, fluidized beds show consider-
able bypassing and slugging - that is, intermittent
and unstable flow of large gas bubbles through the
bed. Both these behavior patterns are strongly
affected by the design of the distributor that holds
up the packed bed and distributes gas flow into it.
The presence of particulate or aggregative
fluidization must be the result of the influence of
gravitational. forces on the phases present in a
fluidized bed as well as the bulk fluid mechanics of
the system. Thus, the Froude number (v2 /Dpg).
which is the ratio of kinetic to gravitational energy,
would be expected to be one of the criteria. Orig-
inally, the Froude number was taken as the entire
criterion (61 ), but more extensive information has
indicated that if
particulate fluidization occurs, whereas if
(IVF,)(,VR,) (p,; p)(t) > 100 (22.93b)
aggregative tluidization exists (48). In Equations
22.93 the NF,, NR ,, and L should all be taken at
the point of minimum fluidization.
As indicated above, the minimum fluidization
point, point C in Figure 22.31, will occur when the
pressure drop due to flow balances the bed weight,
643
and after the particles have had a chanceto adjust
to a position offering minimum pressure drop.
Thus, at point C,
-t,.Pe g
-- = (1 - Ee)(p, -pG) - (22.94)
Le ge
The porosity at minimum fluidization Ee will be
slightly greater than that for a packed bed, and
hence Le will also be greater than the height of the
bed without flow. Ee will also depend upon the
shape, surface characteristics, and size of the solid
particles. For example, for mixed round sand
(,JI= 0.86) Ee"" 0.42, for absorption carbon
Ee"" 0.72, and for anthracite coal (i/1 = 0.63)
Ee"" 0.60 (35).
As the fluid velocity is increased from C
toward E (Figure 22.31) the bed expands, some-
times up to 2.5 times Le. The exact amount of bed
expansion depends upon the distributor design and
bed diameter, and hence is difficult to predict on
the basis of laboratory or pilot results (29).
Equation 22.92 gives an estimate of this expansion
for particulate fluidization in a bed where wall
effects and recirculation are absent. In designing a
fluidizedbed. the engineer must allow not only for
this expansion but also for the height above the
top of the bed required to get adequate solid-fluid
separation, the transport disengaging height (TDH).
This also depends upon the characteristics of
the fluid-solid system and has not yet been
adequately correlated. Figure 22.32 gives an
empirical correlation for estimating TD H for fluid
catalytic cracking catalyst (64) and shows typical
TDH values. At heights above TDH, there will still
be catalyst losses, especially where there is particle
-;;
:,:
....
Vessel diameter d1 (cm)
Figure 22.32. Empirical correlation for estimating
the transport disengaging height for- fluid catalytic
cracking catalyst (64). (By permission of A.I.Ch.E.,
copyright© 1958.)
644 APPLICA TlONS TO EQUIPMENT DESIGN
breakage or widely varying particle sizes within the
bed, but these losses will not be greatly reduced by
further free space.
Industrially, it has been found advantageous
to carry out many solid-catalyzed reactions
in fluidized beds. The circulation of the bed
and uniform agitation within it prevents the
occurrence of hot spots and dead regions. It
also makes possible the continuous circulation of
the catalyst between the reaction vessel and a
regeneration vessel. The major disadvantage of this
system is
that the catalyst is eroded. and broken by its con-
stant motion, so that there is continual attrition of
the particles requirinq continual makeup of fresh
catalyst.
The largest-scale industrial application of
fluidization is in the fluid catalytic cracking of
heavy crude-oil fractions to give gasoline compo-
nents. This process was developed during the early
part of World War 11 when there was a critical need
for high gasoline yields from crude oil and has been
very widely applied since then. Figure 22.33 shows

Figure 22.33. A 23,000 bbl/day fluiC: .catalytic cracking unit. The large vessel
just to the right of the elevator shaft is the regenerator. The reactor is the r.ext
vessel. to the. right of the regenerator. This is an Essa Model IV unit built at Port
Jerome, France. Units of this type first went on stream at the end of 1952 and
have been built in sizes from 5,000-55,000 bbl/day. (Courtesy Exxon Research
and Engineering Company.)
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 645
Example 22.9. A catalyst having spherical particles
_ To product of Dp = 50 microns and p, = 1 .65 g/cm3 is to be
rrecnonaueo
used to contact a hydrocarbon vapor in a fluidized
reactor at 900°F, 1 atm pressure. At rest the bed
has a porosity of 0.35 and a height of 3 ft. At
operating conditions, the fluid viscosity is 0.02
centipoise and its density is 0.21 lb/ft3. Determine
the superficial gas velocity necessary to fluidize the
bed, the velocity at which the bed would begin to
flow with the gas, and the extent of bed expansion
when the gas velocity is the average of the velo-
cities previously determined. The porosity at mini-
mum fluidization velocity is 0.42. Does aggregative
of particulate fluidization occur?

Solution. The mean superficial velocity at the

Figure 22.34. Schematic diagram of fluid catalytic-cracking
start of fluidization can be obtained directly from
process.
Equation 22.91. Thus,

a typical fluid-cracking unit, and the catalyst and (v,m O) =� (p, - p)gDP 2
fluid flow through this unit is diagrammed in
Figure 22.34. 1 -e 150µ
Regenerated catalyst at about 1050°F is
2
blown by the vaporized feed into the reactor 5
(1.65x62.4-0.21)32.2( 0.0 0 )
vessel. The hot catalyst preheats the feed and is 0423
. 2 x 2
5.4 1
conveyed at a rate of about 7 lb of catalyst per =- -x---------------
0.58 150 x (0.02 x 0.000672)
pound of feed as required by the system energy
balance. Cracking takes place in the reactor at
about 900° F with coke depositing on the surface 0.128 x 102.8 x 32.2 x (0.0269 x 10-6)
of the activated natural clay or synthetic silica- 0.002014
alumina catalyst. Spent catalyst collects in a stand-
pipe and flows through a valve into an airstream = 0.0056 ft/s
that conveys it back into the regenerator. In the
regenerator, the coke is burned off the catalyst,
whereupon the regenerated catalyst collects in a is the velocity at which fluidization will begin.
second standpipe for return to the reactor." A The bed will disintegrate and stream with the
25,000 bbl/day unit such as the one shown in flowing gas when the gas velocity equals the velo-
Figure 22.33 contains about 500 tons of catalyst city of free fal! of the particles. For these small
of 20- to 400-micron particle diameter circulating particles, the flow is laminar. and the settling
at a rate of 20 tor.s/min and transferring velocity can be deterrni ned directlv from Equation
100 x 106 6tu/hr of heat from the reqenerator to 22.15.
the reactor. Only about 1 percent of the circulating
catalyst is entrained by the fluids, and it is separ-
ated in cyclones before the fluids leave the con- v = (p, .,- p)g D/ = 102.8 x 32.2 x (0.0269 x 10- 6)
tacting chambers (25). t 18µ 18 x (0.02 X 0.000672)
Although the fluid-cracking process is the = 0.37 ft/s
largest-scale application of fluid-bed reactinq,
fluidized reactors have been developed for many
processes including ore roasting, cement manu- The extent of bed expansion at anv inter-
facture, extraction of oil frqm shale and bitumi- mediate flow rate can be determined from
nous sand, production of phthalic anhydride by Equation 22.92 by solving for e. This was not
oxidation of naphthalene, and oxidation of ethy- possible when finding the maximum value of v,m
lene to form ethylene oxide (50). because Equation 22.92 becomes indeterminate
646 APPLICATIONS TO EQUIPMENT DESIGN
when e r+ 1. Solving Equation 22.92, sizes at flow rates high enough to fluidize all the
3
bed particles. At these flow rates, a large propor-
150µ e0 tion of the fines would be expected to be entrained
2 (v,m - v,m 0 ) + ---
g ( p, - p )D p - € in the gas phase. For example, cracking catalyst
0.002014 normally ranges between 20 and 500 microns in
diameter (40) and is fluidized at superficial
32.2 x 102.8 x 0.0269 x 1 o- 6 velocities of 1 to 2 ft/sec. As seen in Example 22.9,
0.37 + 0.0056 ) 0.423 these velocities are great enough to entrain
x ( 0.0056 + -- particles somewhat greater than 50 microns. How-
2 0.58
ever, in operation, relatively little entrainment of
solids occurs. This apparent violation of Stokes's
= 23.5(0.187 + 0.0027) + 0.128 = 4.18 + 0.128
law probably results from the flocculation of the
= 4.308 small particles into agglomerates perhaps held
e = 0.855 at v,m = 0.190 ft/s, the average of the together by static electricity. Such agglomerates
previously determined velocities. have been observed and seem to be particularly
large and stable in particles of about 10 microns
The type of fluidization that will occur can be and smaller diameter. Similar effects were noted by
inferred from the use of Equations 22.93. Morse (44) in examining the available experimental
data on pressure drop and porosity of fluidized
p, - p) L v2 De vp Ps - p L beds. Below NR • of about 10, he found that the
NF,xNReX = x x x

( --- - -- --- - modified friction-factor term of Equation 22.86

p D Dpg µ p
was lower for fluidized beds than for fixed beds,
v3(p, - p)L but, above NR • = 10, the modified friction factor
for fluidized beds was greater than that for fixed
gDµ 1-0.35
(0.0056)3(1.65 x 62.4 - 0.21) x 3 � beds. In the low NRe range, this deviation is
explained on the basis of flocculation, since the
0.050 experimental data in this range were obtained by
32.2 x 0.02 x 2.42
25.4 x 12 Leva and coworkers (37) with small particles. The
data of NR • > 10 were obtained by Wilhelm and
= 0.237
Kwauk (61) with larger particles. The deviation
Thus, we would expect particulate fluidization, in here was explained as the result of the kinetic
accordance with Equations 22.93. energy imparted to the solid particles by the fluid
Even though we do have particulate fluidiza- stream. The comparison between fixed- and fluid-
tion, it is probably necessary to redo the calcula- ized-bed pressure-drop data given by Morse is
tion on the bed expansion at the stated conditions. shown in Figure 22.35 (44).
An examination of that result leads to that con-
clusion. At the porosity calculated, the bed height
would be

L 1 - e0 0.58
-=--=--=4.00
Lo 1 - e 0.145
Thus, the bed is four times its original height. Such
an extreme expansion is sure to produce bypassing
c
and slugging so that the bed would be much less .2

expanded than calculated. R

Equations 22.91 end ?2.fl? \Mere ohtainsd by 0.1 ......--i�--------�

applying to the Carman-Kozeny equation the re- o.i 1.0 10 100 iOOO 10.000 100.000
striction of constant pressure drop equivalent to Mocl1f1ed Reynqlds number N�/1-E

the bed weight (Equation 22.89) and allowing a Figure 22.35. Comparison between fixed- and fluidized-
resulting variation in porosity. Fluidized beds are bed momentum transfer (44). (By permission of Amer.
normally operated with a wide range of particle Chem. Soc .• copyright© 1949.)
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
Since fluidized beds are used for such diverse
purposes as catalytic reactions, coal combustion,
and particulate drying, the rates of heat and mass
transfer between the particles and the fluid are of
special interest. Determination of these transfer
rates has been the subject of much work (29, 64).
A correlation for the particle-fluid heat transfer
coefficient is ( 4)
hG µ213
k2/3 Cp 1 /3 0.043
[d
P
(1-E)2
G2 gp(p, - p)
]1/4
(22.95)
where all the properties are those of the fluid
unless indicated.
Fluid-Solid Conveying, Pneumatic Con-
veying. When the fluid-phase flow rate exceeds
the free-settling velocity of the particles, the fluid-
ized bed loses its identity because the solid
particles are conveyed in the fluid stream. This
method of conveying is frequently used through-
out industry usually with air as the fluid phase; for
instance, to unload grain ships, to convey the
product from spray dryers, to fill and empty
cement silos, etc. It has the advantage of cleanli-
ness, low loss, and the ability to move large quanti-
ties of solids rapidly. On the other hand, significant
breakage of solid particles may occur, and pipe
erosion may be excessive.
As the vertical gas-phase flow rate increases
beyond the fluidization region, the solid particles
are initially moved with the fluid almost as a
moving fluidized bed. This flow region is usually
called moving bed flow. At higher gas rates, the
solids are carried in gas slugs, much as occurs in
aggregative fluidization. This is called slugging
dense phase flow. Finally, at still higher gas rates,
the solid bed loses all cohesion and the particles
stream with the gas as dispersed individuals. This is
called dilute phase flow (38).
The transition from slugging dense phase flow
to dilute phase flow is referred to as the choking
point and usually, but not always, is an abrupt
transition. Conditions at which this transition
occurs have been correlated by Yang (62) to about
a ±30 percent accuracy
v, = kev- v,)(1 - Eel (22.95a)
kev-v,)2/2gD = 100(Ee-0.47 -1)
(22.95b\
where v, = superficial solids velocity
v, = terminal velocity of a solid particle in
the gravity field '
v = gas velocity at the choking point
Ee= fraction voidage at choking
647
Equations 22.95a and 22.95b can be used to
obtain the unknown velocities and void fraction at
choking with two different sets of known condi-
tions. In one situation, the gas velocity and void
fraction at choking can be obtained by simul-
taneous use of these equations when the superficial
solids velocity is known, as would be the case if the
solids were fed into the gas flow at a measured
rate. In the other situation, the void fraction at
choking can be determined directly from Equation
22.95b if the gas velocity itself is known.
Where particles of a distribution of sizes are
being conveyed, as would almost always be the case,
Equations 22.95a and b can be modified by in-
cluding a volume fraction distribution of the
particles (38)
x,;v, = x,;( 1 - Ee)(VEe - v,;) (22.96a)
[�x,;(VEe - v,;)] 2 = 200 gD(Ee-o.47 - 1)
(22.96b)
(22.96c)
where xi;=volume fraction of feed particles with
terminal velocity v,;
where x,; = volume fraction of particles in the riser
with terminal velocity v,;
The transition from moving bed flow to slug-
ging dense phase flow occurs when the voidage in
the conveying tube equals that at incipient fluid-
ization (39). If a typical void fraction at initial
fluidization is taken as 0.45, then
0.55 V"Ec - 0.45 V5 = 0.55 Vm f (22.97)
The use of Equations 22.95 and 22.97 allows
the phenomenon of gas-solid vertical conveying to
be broken into three regimes, as shown in Figure
22.36. Since each of these exhibits a different
phvsical situation, it wou!d be expected that each
must be treated separately in predictir.g the
pressure drop. This is the case, and hence separate
pressure drop correlations for these regions are
presented below.
1. Dilute Phase Flow Regime. Since this
region shows steady, independent flow of gas and
solid phases, the energy balance can be written for
the two phases and simplified to give the needed
relationships. From such a qeneral expression (53)
2]L
t;,P = [ pEV2 +
p v
! !..._ + g f L
((1 - E)p, + ep]dZ
1 -E o 0
+4,L/D (22.98)
648 APPLICATIONS TO EQUIPMENT DESIGN
Moving bed
flow
Log (dimensionless gas velocity), log (v/umt>
Figure 22.36. Quantitative flow regime dia-
gram in vertical pneumatic conveying for a
choking system (38). (By permission of
Amer. Chem. Soc., copyright© 1976.)
where the three terms on the right-hand side of the
equation represent pressure drop due to accelera-
tion, gravity, and wall friction, respectively, and
where r is the shear stress at the wall. Assuming
that t:.P is small relative to the pressure, that the
acceleration section is short, and that the gas
density is very small, Equation 22.98 reduces to
2 v
PsVs
t:.P=--+ (1 - E)p,gl +4rl/D (22.99)
-€
The wall friction term is typically broken up into a
term due to fluid friction, and one due to solids
flow (53).
4rl/D = t:.P,, + t:.P,, = fl v2f2gcD + 2f,Lp,v,2D
1-€
In this equation f is the friction factor as defined in
Chapter 20, and f� is a particle-wall friction factor
that must be determined empirically. Since the
particle-wall friction loss is usually a small factor.
Leung and Wiles- (33) recommend a simp!e expres-
sion that roughly correlates most of the low-
pressure conveying data
f, = 0.005(1 - E)!v,
Since the constant in Equation 22.101 is dimen-
sional, v, must be expressed in m/s. Data on
particle-wall friction loss in high pressure con-
veying have been correlated by the equation (27)
f = [o.02515(w /pv10-041 s( v )-o.ass]
s ' ' (1 ;- E)V
- 0.03
where w, is the mass flow rate of solids, kg/s m2
2. Dense Phase Slugging Flow. Here the
pressure drop is almost entirely a result of the
weight of the solids, and Equation 22.99 can be
simplified to
t.P = P, (1 - E)gcl (22.99a)
Use of even this simple equation requires deter-
mination of the void fraction. The following
equation has been proposed (41)
1 - e
--- = [vb + w,/[p,(1 - Emrll] I
- Emf
(v+ vb - Vmr+ w,lp,) (22.103)
where vb = bubble rise velocity in a fluidized bed
vb= 0.35# for slugging conveying
3. Moving Bed Flow. This flow regime has
practical application in conveying systems in
limited situations, such as for short-distance solids
moving. In this regime, the pressure drop cart be
calculated by modifying Ergun's equation
(Equation 22.86) by the use of an appropriate slip
velocity (63) '
t:.PD €3
= 150/NA,,/ + 1.75
Lp (v---E)(1 - E) (22.104)
1-€
where NR, , is the Reynolds number of gas relative
1
to the solia particles, and is defined as
Dp(v-_.".!_)PE
N - 1 -E
Re,r - µ(l _ €) (22.105)
(22.100)
The equations and the approach presented
here offer a logical attack on the design of vertical
solid.gas conveying. However, the wide variation in
solid properties and the tendencies of solids to
agglomerate, and to adhere to wall surfaces, makes
the use of these equations hazardous except when
dealing with free-flowing solids of nearly uniform
size and high sphericity. Conveying fibers and
(22.101) solvent-refined coal are but two examples of appli-
cations for which these equations would be inade-
quate. Thus, a test of the actual convevioq situa-
tion should be carried out if at all possible.
Moreover, horizontal conveying has not been
considered. This might be attacked by writing
Equation 22.99 without the term for solid particle
weight. In fact, for short horizontal runs such a
procedure is feasible and results in pressure drops
(22.102)
that are small compared to those in vertical
conveying. Problems occur where low gas velocities
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
are used, for here the tendency of solids to settle in
horizontal sections limits the velocity that can be
used.
Bend losses must also be recognized. Bend
loss estimates ranging from 0.5 to 1.5(v2 Pm /2g)
have been recommended with the constant de-
creasing as the bend radius increases (38). Here Pm
is the density of the flowing mixture,
Pm=p,(1-e)+pe.
The design of a pneumatic conveying system
requires careful consideration of the properties of
the material to be conveyed - such as caking
tendencies, ease of particle breakage, and any
tendencies of the powder to explode - and of the
operating conditions such as pressures, tempera-
ture, sol ids/gas ratios, etc.
The physical arrangement requires, at a mini-
mum, an air blower, a method of feeding the solids
into the airstream, and a method of later removing
the solids from the air. The household vacuum
cleaner is a simple example of such a system. Use
of such a vacuum system is often convenient on a
large scale, too, especially where the solids are
friable and where both feed solids and ultimate
solids location are at atmospheric pressure. Unload-
ing barges or trucks is an example. Often the solids
are fed into a moving gas stream by a mechanical
device such as a star feeder. This is a set of paddle-
vanes that rotate in a horizontal cylinder allowing
solids to flow into the cylinder at the top, to be
carried between the vanes to the bottom of the
cylinder, and to fall out at that point into the
conveying air. Separation of the solids from the
airstream is usually done in cyclone separators. In
vacuum systems, it is usually necessary to protect
the blower with an air filter placed between the
cyclone and the blower.
If a pressure system is used, either because the
blower is located upstream from the solids addi-
tion point (thus ensuring that solids do not enter
the pump), or because the solids are taken from or
delivered to a pressurized chamber, solids addition
becomes much more difficult. Feeding into a
Figure 22.37. Schematic diagram of a blow-through solids
feeder.
649
slightly pressurized system, such as would be the
case with an upstream blower, can probably be
done with a star feeder. Problems can occur as the
pressure drop caused by solids flow builds up,
however. Another possibility is the use of a flow-
through feeder. This operates like the star feeder
except that the air flows through the feeder
cylinder (see Figure 22.37). Another possibility is
the mechanical screw feeder shown in Figure
22.38. Here the material enters the screw from a
small hopper and is carried through the barrel by
the screw action. The screw is of decreasing pitch,
which compacts the materials and forms a
pressure-sealing zone. The material forces the
check valve open as it leaves the barrel, and is
broken up and picked up by the air nozzle that
forces it into the conveying line.
Feeding solids into a high-pressure chamber
on a continuous basis is one of the more difficult
requirements of chemical processing. In fact,
feeding coal into high-pressure reactors has been,
and continues to be, a major barrier to the develop-
ment of coal conversion processes. Most successful
systems have been semicontinuous. For instance,
the solid is fed into a vessel, sealed in and pres-
surized, and then a valve is opened that permits the
solids to flow into a high-pressure conveying line.
If continuous feed is desired, two such tanks are
needed, both continuously weighed, with one
being filled and pressurized while the other is
feeding the conveyor. Other feeding mechanisms
fluidize the solid in the pressure chamber and
remove the solids from the bottom and/or top of
the fluidized bed.
For such flow systems, there is always the
danger of static buildup and spark ignition of a
combustible dust-air mixture. Safety precautions
appropriate to the system are mandatory.
Example 22.10. In manufacturing synthetic
detergents, the spray-dried bead· is pneumatically
ccnveved from the bottom of the spray drier to
cyclone separators at the top of the manufacturing
building. In a plant producing 3000 lb/hr of
Figure 22.38. Schematic diagram of mechanical screw feed·
feeder.
650 APPLICATIONS TO EQUIPMENT DESIGN
product, the conveyor is a 10-in. I.D. tube running
vertically for 60 ft and horizontally for 30 ft. The
detergent has a mean particle size of 600 microns
and a packaged bulk density of 20 lb/ft3. The
blower pulling air through this conveyor has a
rating of 500 standard ft3 /min. What is the
pressure at the blower inlet if atmosnheric pressure
exists at the bottom of the spray drier? Ignore
pressure drop through the cyclone and assume the
conveyor operates at a constant 80°F.
Solution. First the flow regime will be deter-
mined by use of Equations 22.95 and 22.97. Here,
as is typically the case, the superficial solids
velocity is obtainable. The solid density is obtained
from the bulk density in the package.
20
From a consideration of the form of Equations
22.95a and b, both as shown in Figure 22.36 and
as seen by noting the equations themselves, it
seems clear that choking does not occur, that is,
flow is in the dilute phase region. Thus Equations
22.99, 22.100, and 22.101 will be used to obtain
the pressure drop. To use Equation 22.101, the
void fraction e is needed. This can be obtained
from the superficial gas and solids velocities, and
the terminal velocity
v,=v,*(1 -e)
v-v5•=vr
where v, • = actual sol ids velocity
v, • = -5.05 + 16.77 = 11.72 ft/s
1 - e = 0.0473/11.72 = 0.00404

p, = _ _ 32.3 lb/ft3
1 0 38 which confirms the presence of dilute flow. Then,

This gives the superficial solids velocity from Equation 22.101,

3000 1 1 ' f, = 0.05 x 0.00404/0.0473 = 0.00427
v = -- x x -- = 0.0473 ft/s
' 32.3 rr/4 x (10/12)2 3600
The friction factor at the wal I can be obtained
The gas velocity is given by the blower capacity from Appendix C-3.
500 1 540
v=-x x-=16.77ft/s . !2!!e_ (10/12) x 16. 77 x 0.074
60 rr/4x (10/12)2 492 77,950
NRe = µ - 0.02 X 0.000672

and the terminal velocity is obtainable from Figure

f= 0.019 if e = 0.0002 as for commercial steel pipe
22.1 via Equation 22.19
4gDe 3 p(p, - p) - Equation 22.100 can now be used to get the
3 2 - wall friction term of Equation 22.99.
µ 29 492
4 x 32.2(600 x 10- 6/0.305)3 x - x- (32.3- p) 2

359 540 4
T/D = fv + 2f,p,v,2 D

3 x (0.000672 x 0.02)2 '2/JeD 1 -e

128.8(0.00197)3 x 0.074(32.3 -0.074) 0.019 x 16.772
4334 64.4x 10/12
= 3 x (0.00001344)2 . =
Thus, 2 x 0.00427 x 32.3 x o:ci4732 x 10/12
Dpv,p = 55 x 0.00404
µ
= O.G996 + 0.1273 = 0.2269
and
55 x 0.000672 x 0.02 . Now, Equat1" 22.99 must be broken into
v, = 0.074 x 0.00197 = 5·05 ft/s two parts, one for the vertical section of 60 ft and

Now, using Equation 22.95a, one for the 30 ft horizontal section. For the verti-
cal section
0.0473 = (16.77Ee - 5.05)(1 - Ee)
PsVs 2
Ee= 0.306 !!.Pv = --+ (1 - e)p,gL + 4T/DL
1-E
Checkinq with Equation 22.95b,
32. 3 x o:ci4732
(0.306 x 16.77 - 5.05)2 /64.4 x (i0/12) = 0.00404 + 0.00404 x 32.3 x 32.2 x 60
100(0.306-0·47 -1)
+ 0.2269 x 60
(5.13- 5.05)2/53.67 = 100 x 0.745 = 74.5
= 17.83 + 252.11 + 13.61 = 282.55 lb/ft2
0.000119 ,fo 74.5
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
For the horizontal section, the second term on the
right-hand side of Equation 22.99 should be
omitted, since that represents the energy needed to
lift the solids. Thus,
30 30
!::.PH= 17.83 x-+ 13.61 x-= 15.72 lb/ft2
60 60
Finally, at least one bend loss should be included.
!::.Pa= [p,(1 - e) + pe] v2/2g
= (32.3 x 0.00404 + 0.074 x 0.996) 16. 772/64.4
= 0.89 lb/ft2
Then the total pressure drop is
!::.P = (282.55 + 15.72 + 0.89)/144 = 2.08 psi
Thus, the pressure at the blower inlet is
P = 14.70 - 2.08 = 12.62 psia
Fluid-Solid Conveying, Slurry Transport. In
addition to the pneumatic transport described
above, which is widely used for short-distance
movement of solids, the economy of long-distance
moving of pulverized sol ids in slurry form indicates
that such moving will be used with increasing fre-
quency. In view of the expected increase in use of
coal, extensive studies have been made in the trans-
portation characteristics of a slurry of coal in water
and some slurry pipe lines are in operation. For
example, the Mohave Power Station receives
slurried coal through the Black Mesa Pipeline. This
line runs for 273 miles between Arizona and
Nevada. It is 18 in. in diameter and is designed to
move 5.5 million tons/year of coal in a 50 percent
b'i weight slurry (57). Other solids are also moved
in other places. For instance, in Tasmania a
valuable deposit of iron ore that had been con-
sidered inaccessible for conventional transport is
now being moved to dockside by hydraulic trans-
port through a pipeline 6 in. in diameter and 53
miles long.
Generalized correlations upon which designs
can be based are virtually nonexistent, because of
the difficulty in fitting most slurries into conven-
tional fluid mechanics correlations. Many data have
been collected, but unifying principles have been
hard to develop. The heterogeneity of the slurry
seems to preclude the application of continuum
expressions for viscosity and density.
The classic generalization is that of Durand
( 16) and his coworkers who correlated slurry flow
behavior in terms of an increment on the friction
factor expressing the contribution of the solid
phase to the friction factor for the liquid phase if it
651
were flowing alone. This add-on is based upon the
conventional drag coefficient for particles of the
solid, which avoids the necessity of using a vis-
cosity or pseudoviscosity for a slurry. One form of
the equation is
v2
f - fw = K [ - ,.;c,;:,_ ]- 1 .5
fwc gD -
where f and fw are friction factors for slurry and
for liquid alone, and c is the weight fraction of
solids.
C0 = drag coefficient for a free-falling sphere
Co =4/3gd--
(s - 1)
v_2
s = density ratio I?.!.
p
It has been well known that the constants in
the equation must be appropriate to the flow
regime. As is typical in two-phase flow situations,
the mode of flow depends upon the flow rates of
each phase and the total of both. At very low
velocities of the fluid, the solids comprise an essen-
tially stationary bed with very little movement
(regime 0). As the rate of flow of the fluid is
increased, increasing amounts of the solid are
thrown up into the liquid phase in bursts, or slugs.
This has been characterized as "saltation" flow
(regime 1 ). At progressively higher fluid velocities,
the entire portion of the solid is lifted ir.to the
stream and is moving, but the Iarqer particles
congregate toward the bottom of the stream;
hence, this is called "heterogeneous" flow (regime
2). At a sufficiently high velocity, the solids can
become essentially uniformly distributed -
homogeneous flow (regime 3). Turian (57, 58) and
his coworkers examined and ciassified thousands of
published experimental measurements into these
four regimes of flow and fitted for each a set of
correlation constants to generate empirical
equations for pressure drop in the general formula-
tion and variables of Durand's equation (Table
22.5 and Figure 22.39).
v2
[ Dg(s -
r
1 )J
(22.106)
A calculation of the pressure drop for a
flowing slurry must recognize the different flow
mechanisms discussed above; it must identify the
652 APPLICATIONS TO EQUIPMENT DESIGN
Table 22.5. CONSTANTS FOR THE SLURRY FLOW FRICTION FACTOR ·11·
EQUATION (EQUATION 22.106)

Equation Regime
Designation Im) !!. fl s ft !!.

a 0 0.4036 0.7389 0.7717 -0.4054 -1.096

b 1 0.9857 1.018 1.046 -0.4213 -1.354
c 2 0.5513 0.8687 1.200 -0.1677 -0.693 8
d 3 0.8444 0.5024 1.428 0.1516 -0.3531
1000����������������� guide, and the ratios have been used as the identifi-
900 cation number.
BOO The boundaries between adjacent pairs of
Flow with a regimes are generated by an empirical combination
stationary bed

700 (regime 0) of the pressure-drop function's for each adjacent

regime (actually, a "cut-and-fit" process). A plot of
600
principal boundaries is shown in Figure 22.39 for
c the locus of values of regimes number equaling
b 500 unity for the contiguous regimes, applicable to one
§.
-{!; particular system. The constants tabulated in Table
4W 22.6 represent an interpretation of many measure-
·�
ments. These are fitted into Equation 22.107 for a -
rs
given flow system; several sets must be evaluated
i
-o 300
and studied to find the pair of regimes separated
3i
Homogeneous flow by a ratio of unity for the pressure drops expected
(regime 3)
in these two regimes.

v2
T= 22.5 "C Rmn = ' = 1
P1 = 2.977 g/cm3
a' • c"' • fWC) d'
( • Co ; • Dg(s - 1)
C = 5% by volume gc
Pipe LO. == 0.957 in. 4
(22.107)

9 10
Inspection of Equation 22.106 in relation to
Mean slurry velocity, u (ft/sec) Figure 22.39 indicates that the ratio computed in

Figure 22.39. Flow regime diagram for solids-water flow in
t-ln. PVC pipe (57). (By permission of A.I.Ch.E., copyright
©1977.)
existing regime and be applicable to it. This is true
for both design of a new installation and analysis
of performance data from ar. ·exis1:ing svstem. In
general, determination of the existing regime
cannot be made by observation.
Identification of the existing regime is
effected by comparing the pressure drops predicted
by the appropriate functions for each and analyz-
ing the order of changes indicated by regimes in
order of increasing flow velocity ..4t a boundary
between regimes, the pressure drops should be
equal, or the ratio of them should be unity. Since
the differences of friction factors are small num-
bers, they are not satisfactory as an identification
all cases increases with velocity, beooming unity at
a velocity corresponding to the maximum velocity
for that regime and for the system being studied.
ldentifieatlon of the existing regime is based upon
applying to the existing operating conditions sets
of constants appropriate to regimes in increasiog
order of velocity. The existing regime is the first
one in which R < 1.
The transitions occur in zones, rather than the
sharp increments indicated by Figure 22.39. Par-
ticularly for fine particles, the regime boundaries
cross so that transition is expected from stationary
bed flow directly to heteroqeneous, or homo-
geneous flow. The reader is referred to the oriqinal
publication if these transitions are encountered.
Example 22.11. Determine the pressure drop
per mile that would be expected in the Black Mesa
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 653

Table 22.6. CONSTANTS FOR THE SLURRY FLOW REGIME

EQUATION (EQUATION 22.107)
Boundary
Equation Between
Designation Regions a' b' c' d'

a 0-1 31.93 1.083 1.064 -0.06160

b 1-2 2.411 0.2263 -0.2334 -0.3840
c 1-3 1.167 0.5153 -0.3820 -0.5724
d 0-2 0.4608 -0.3225 -1.065 -0.05906
e 0-3 0.3703 0.3183 -0.8837 -0.7496
f 2-3 0.2859 1.075 -0.6700 -0.9375

pipeline described above. As stated, this pipeline is and flow is in the laminar region. For the flow of
18 in. 1.0. 273 miles long. It supplies 5.5 million water alone.
tons/year of coal to the Mohave Power Station as a 1.5 x 5.36 x 62.4
50 weight percent slurry. Consider the median 1.0 x 0.000672- 747,000
particle to be 0.07 mm in diameter with a spher-
icity of 0.8. · fw = 0.0125, using Appendix C-3

Solution. The first step is to determine the r; 0.125

- =-- = 0.003125
type of flow regime using Equation 22.107 with 4gc 4
the constants given in Table 22.6. Each of the
Now the regime boundaries can be determined
groups in Equation 22.107 must first be obtained,
starting with the bulk mean velocity. v2/Dg(s - 1)
-------�--------=77.4
5.5 x 106 x 2000 x 1/0.5 31.93(0.5) 1 ·083(0.003125) 1 ;064(600)-0.0616

x 2.51 x 106 lb/hr of slurry

365 24 (22.107 a)

1.70
p = (0.5 x 1.0+ 0.5 x 1.35) = 1.i8 g/cm3 for the slurry
2.411 (0.5)0,2263(0.003125)- 0.2334(600)- 0.3840 = 2.50
2.51 x 106 x---- (22.107b)
1.18 x 62.4

v 5.36 ft/s 1.70

8 929
3600 x 4" x 1.52 1.167(0.5)0-5153(0.003125)-0.382(600)-0.5724 = ·

(?2.107c)
v2 • (5.36)2
---= =1.70 1.70

Dg(s - 1) 1.5 x 32.2 x ( 1.35 - 1) -----��----- = 0 0031

0.4608(C!.5)-0·3275 (0.003125)--', .oe5(60C)-o.05806 .
To calculate the free-fall drag coefficient C0 _, (22.107 d)
Figure 22.1 can be used along with Equation 22. i9. 1.70
4 25
At NR, = 1, 0.3703(0.5)0·3183(0.003125)-0.8830(600)-0.7496 = ·

C _ 4gD/ p(p, - p) (22.107e)

0 - 3µ2 1.70
75 105 5
0.2859(0.5) 1.0 (0.003125J-0.67 (60Q)•-0.9375 = -
4x32.2x 0.07 x 0.1 )3 x62.42(1.35-1)
. ( (22.107f)
2.54 x 12

3 x ( 1.0 x 0.000672) 2 Comparing these results with the configuration

shown in Figure 22.39, it is apparent that the flow
128.8 x 0.012 x ,0-9 x 1362
----3-5_5_x_1_0 6 = 1 ·554 is in the homogeneous region. In the various forms
of Equation 22.107, a large calculated number
Then, using Figure 22.1 for ,J; = 0.8 C0_ = 1200, means a greater velocity than the transition velocity,
654 APPLICATIONS TO EQUIPMENT DESIGN
or perhaps a smaller particle diameter than that reason orifices, venturis, and area meters are hardly
equivalent to the transition, all other things being suitable, and obstructionless meters such as those
equal. that compare the time required for a small pressure
The next step is to calculate the frictional pulse to travel a given distance diagonally across a
loss, this time using Equation 22.106d. tube with the slurry flow and against it are often
used.
f- f
w =
t. )1.428 (C
0.8444Co.5024 ( --!".. )0.151 s Pressure-sensing tubes and sample tubes need
4gc 4gc o: to be continuously flushed to prevent sediment
v2 )-0.3531 buildup, and measurements of density or of flow
must account for the fact that solids can and do
(
Dg(s - 1) settle onto the bottom of the pipe or other con·
tainer.
= 0.8444(0.5)0.5024 ( 0.0125 )1.428
Relation Between Regions of Fluid.Solid Flow
4 x 32.2
As evident from the treatments above, flow
(600)0.1 51 6 (1.70)-0.3531 through packed beds, sedimentation of solids,
fluidization, and fluid-solid conveying are all opera-
f- fw
--- = 0.8444 x 0.71 x 0.00000186 x 2.64 x 0.829 tions depending upon the mechanics of a fluid·
4 x 32.2 solid system. In fact, the flow of droplets and
f - fw = 0.000314 particles through a spray dryer and the behavior of
the two liquid phases in a spray-extraction column
f= 0.0128 depend upon the same mechanics. All these opera-
Finally, the pressure drop can be obtained: tions can be considered in a unified way by
focusing attention on the relative, or slip, velocity
fpv2U of particle and fluid (31, 42).
-t.P=---
2gcD Thus, by definition
0.0128 x 1.18 x 62.4 x 5.242 x 5280
64.4 x 1.5
where
= 1414 lb,/ft2 = 9.82 psi/mile
v,1 = relative, or slip velocity in the direction
of particle movement
The calculational method given above for
vp = velocity of particle
slurry-flow pressure drop is adequate for slurries in vr, = fluid free-stream velocity
which the .solids are noninteractive and relatively
stable. Many slurries, however, form gels on
standing or show apparent viscosities that are A single particle is affected by the fluid
dependent upon fluid motion. The equations given flowing around it which is displaced from in front
of the particle, flows beside it, and passes to the
above will not prove adequate for these materials,
and hence more specialized methods should be rear. The friction and kinetic-energy changes that
resuit determine the velocity of the solid particle
used (38).
Specially designed pumps must usually be relative to the fluid. If a significant portion of the
used for moving slurries, even when the solids total stream consists of solid particles, the area for
content is quite low. These may be of the dia- fluid passage around the particles is more sharply
phragm type for small flow rates, but with larger restricted than it is with a single particle in a free
flow, centrifugal pumps may be needed. These stream. Moreover, there is a significant interaction
pumps are likely to have large, unobstructed flow between particles. By considering only the flow
paths, abrasion-resistant lining, lower impeller path around the particles and the effect it has on
speeds, and larger drive motors. Pump rnanu- the separation eddies behind the partic!es,
facturers will be pleased to assist in pump selection Equation 22.108 can be derived for spherical
for these applications. particles of fixed diameter spaced in a cubic
Slurry-measuring systems must also be care- arrangement (42).
fully chosen. Flow meters should offer as little
restriction as possible in the flow path. For this v,I,< =1-1.209(1-E)213 (22.108)
Vs/,1;=1
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 655
where velocities by eliminating Vp .e . The superficial solid
v,1,, = slip velocity for particles in a fluid- flow rate (vp,) in cubic feet per square foot of
solid stream of porosity E and infinite cross section per hour is
extent
v,1,,.1 = slip velocity for a single particle in a Vp, = Vp,,(1 - E) (22.112)
fluid stream Placing this equation in Equation 22.111 gives

Equation 22.108 is based upon the minimum free

Vps [ v,,, 1 v, ]
cross section at the stream section containing the --=(1-E) -'-+---
particle diameters and hence gives a result that is Vs/,c=1 Vsf,E=l E Vsf,E:1

too low. A similar equation can be derived based (22.113)

upon the overall average free cross section of the
stream (42), which gives a result that is too high. The term v,1 ,lv,1 ,.1 is a function of porosity, as
From the definition of slip velocity for a shown by Equation 22.108. Although Equation
system of identical particles, 22.108 gives too low a result for the slip-velocity
ratio, inserting it into Equation 22.113 will give a
Vp,E = Yst.e + Vrs (22.109) result that is qualitatively correct.

where
v.!!! = (1 - €) [ 1 - 1.209(1 - €)213 +-1 -v-]
vp,, = particle velocity in a fluid-solid system
Vsf,E•1 € Vsl,t:•1
of porosity E and infinite extent
v,1,, = slip velocity of particles in a fluid-solid (22.114)
system of porosity E and infinite extent
v,, = free-stream, or average liquid, velocity Figure 22.40 is a plot of Equation 22.114 (42).
There are three major regions in the diagram. At
and where the direction of particle movement is positive values of fluid-throughput ratio
taken as positive, Equation 22. 109 can be written (v,lv,1 ,. 1 0). cocurrent flow exists. Since this
in terms of the superficial fluid velocity region' is also in the region of positive solid-
v, throughput ratio (vp,lv,1,, •1 .ol. both phases must
Vpe=Vs/e+-
, , € (22.110) flow downward, In region II, which lies in the area
of negative liquid-throughput ratio but positive
and then divided by the single-particle slip velocity solids-throughput ratio, countercurrent flow
obtains. This region is the one within which settlers
1
--=--+---
V Vs/,e
p,E Vs
(22.111) and countercurrent spray dryers and spray extrac-
VsJ,E=l VsJ,E=1 € VsJ,e=1 tors operate. In region 111, for which the solids·
throughput ratio is negative, cocurrent upward
For continuous flow, Equation 22.111 can be flow exists. This is the fluid-solid conveying region.
written in terms of superficial liquid and solid flow For a fluidized bed, the solid throughput must be

-0.15 -0.05 0 0.05 O.IO 0.15 0.20 0.25

Solids throughput ratio ! ups/usl, f .. 1 0 J

Figure 22.40. Theoretical throughput ratio (42). (By permission of A.I.Ch.E.,

copyright© 1955.)
 656 APPLICATIONS TO EQUIPMENT DESIGN
zero, but the bed porosity can vary between about
0.42 for a loosely packed bed of spheres and 1.0.
At porosities between about 0.42 and about 0.37,
the bed is packed and stable. Thus, the packed-bed
region is a somewhat elongated point at € = 0.4,
vp,lv,1,,.1 = 0. The dotted line along the top of
the diagram represents moving-bed operation. The
maximum throughput line connects the maximum
solids-throughput ratio points on the lines of con-
stant liquid-throughput ratio. This line divides the
countercurrent flow region into two sections.
Below the line, increasing the liquid-throughput
ratio at constant solids throughput increases the
porosity. Above th'is line, a similar change de-
creases the porosity. Thus, within this region at
any fixed liquid and solid throughput, two values
of porosity can be obtained such as points A and
B. If a long solids holding time is desired, operation
at point A would be preferred over that at point 8.
In a liquid-extraction column, point A would
correspond to operation with the heavy phase
continuous, and point B would correspond to
operation with the light phase continuous. In the
operation of a gas-solids device, point B would
correspond to a falling-particle operation such as in
countercurrent spray drying, and operation at
point A would correspond to a fluidized bed
moving downward against an upward-flowing gas
stream as is found in some catalytic reactors. Since
the maximum-throughput line represents both the
maximum solids throughput at fixed liquid through-
put and the maximum liquid throughput at fixed-
solids rate, it is also the flooding curve.
Figure 22.40 is a highly idealized presenta-
tion. The equation upon which it is based applies
to infinite systems of equal-sized spheres and uses a
· slip-velocity relation based upon minimum free
cross section. Therefore, it should be used oniy for
a qualitative understanding of the interrelationship
among these operations rather than for quantita-
tive results.
Filtration
Filtration is one of the most common applications
of the flow of fluids through packed beds. As
carried out industrially, it is exactly analogous to
the filtrations carried out in the chemical labora-
tory using a filter paper in a funnel. The object is
still the separation of a solid from the fluid in
which· it is carried. In every case, the separation is
accomplished by forcing the fluid through a porous
membrane. The solid particles are trapped within
the pores of the membrane and build up as a layer
on the surface of this membrane. The fluid, which
may be either gas or liquid, passes through the bed
of solids and through the retaining membrane.
Industrial filtration differs from laboratory
filtration only in the bulk of material handled and
in the necessity that it be handled at low cost.
Thus, to attain a reasonable throughput with a
moderate-sized filter, the pressure drop for flow
may be increased, or the resistance to flow may be
decreased. Most industrial equipment decreases the
flow resistance by making the filtering area as large
as possible without increasing the overall size of
the filter apparatus. The choice of filter equipment
depends largely upon economics, but the economic
advantages will vary depending upon the following:
1. Fluid viscosity, density, and chemical
reactivity.
2. Solid particle size, size distribution, shape,
flocculation tendencies, and deformability.
3. Feed slurry concentration.
4. Amount of material to be handled.
5. Absolute and relative values of liquid and
solid products.
6. Completeness of separation required.
7. Relative costs of labor, capital, and power.
Granular Media Filters. The simplest of
industrial filters is the granular media filter, con-
sisting of one or more layers of particulate solids
supported on a gravel bed on a grating through
which the material to be filtered flows by gravity
or by pressure. Figure 22.41 shows a granular
media filter built for pressure operation. In some
granular media filters, two beds are used. In such a
filter there might be a layer of anthracite
of modest particle size (perhaps 1-3 mm) on top of
a layer of much smaller sand particles (perhaps
0.5 mm). Granular media filters are used primarily
where large flows of very dilute slurry are to be
treated, where neither the solid nor the liquid
product has high unit value, and where the solid
product is not to be recovered. Thus, they are a
mainstay of wastewater purification systems. The
dual media filter allows for longer cycle times
before backwashing is needed, since large particles
or floes may be caught in the more open anthracite
bed. Eventually the flow will drop, or the pressure
drop will become excessive. Then filtration is
stopped and the bed is cleaned by backflushing
with wash water, and perhaps also with air. Back-
washing may be rapid enough to fluidize the
granular bed. Flow rate thro.ugh a granular media
filter or sand filter may be calculated using
Equation 22.86 for the condition immediately
after backwashing when the bed is clean. As solids
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 657

Fll l[R£0

STRAIN[R
STACK

DOU BL[ DISH

UNOCRORA.IN

Figure 22.41. Cutaway view of a vertical granular bed filter built for pressure
operation. (Courtesy The Permutit Company, Inc.)

build up between the sand particles, the porosity 1( x o.5283

decreases and the flow rate drops. 6 x 1728 ,
This gives Dp = = 0.0021 ft
rr x 0.028 2
Example 22.12. An open sand filter uses a 0.9 x 144
3-ft-deep bed of -20 + 28 mesh sand as primary
Applying Equation 22.86 directly requires a trial-
filter bed. The sand particles used have an esti- and-error solution for v,.
As a first approximation,
mated sphericity of 0.9. !f the slurry being filtered Equation 22.76, the Carman-Kozcny equation, will
is essentially water and stands 2 ft deep over the be applied and the answer then adjusted as neces-
top of the sand, determine the maximum flow rate ·sary using Equation :.22.86.
through the bed that occurs immediately after
backwashing. (-t.P)r (1 - d2 µv,
--- ·g = 180-----
L c r3 D/
Solution. The average particle size, as deter- From Appendix B, Figure B-10. the porosity is
mined from the screen openings, is 0.0280 in. estimated at 0.40. Then,
Since the particle area and volume are not known,
the particle diameter (Dp) cannot be precisely 5 x 62.4 0.62 ( 1 x 0.000672) x v,
determined. One method would be to take DP as --- x 32.2 = 180 0.43 x 0.00212
equal to 0.0280. An alternate approach is to v, = 0.0218 ft/s
assume the particles have a volume equal to that of
a sphere of D = 0.0280 in. Then Note that in this solution p t.z glgc has been sub-
stituted for t>P. The Carman-Kozeny equation is
correct only at low NA•. Checking the NA e,
.,, x 0:0282 2
Surface of sphere = ft
144 0.0021 x 0.0218 x 62.4
_ Asphe,e _ rr X 0.0282 rr X 0.0283

A ---- and V = ft3 NAe = 1 x 0.000672 = 4·25

P ,J, - 0.9 x 144 P 6 x 1728
658 APPLICATIONS TO EQUIPMENT DESIGN
From Figure 22.30, it is apparent that at this NR e
the Carman-Kozeny equation is in error by
36.5/35.0 in terms of 1/v,m 2. Thus, the corrected
Vsm is
The volumetric flow is 0.0214 x 60 x 7.48 =
9.6 gal/ft2 min. This applies immediately after
backwashing. The average rate for the entire filter
cycle may be less than half this figure.
The solids to be collected in a granular media
filter may well be smaller than the interstices
between particles in the bed. Still, high-efficiency
solids removal is vital if legal effluent limitations
are to be met. These small particles may still be
collected if they can be assembled in floes or if
they can be captured by adsorption on the bed
surfaces. Because solids in the presence of water
have intrinsic electrical surface charges (for colloids
these are called zeta potentials). and because these
charges are usually negative for both the particles
being filtered and for the granular bed particles, it
may be necessary to alter the zeta potential chemi-
cally if high-efficiency filtration is to occur.
The addition of flocculants, or the altering of
zeta potentials by addition of electrolytes to the
solution, must be tailored to the particular solids
and fluids involved. A summary of known tech-
nology is given in References 24 and 34.
For filtering a gas-solid material, the bag, or
hat filter is often used. This filter consists of large
felt or canvas bags stretched across the openings in
a framework bu i It across the gas-flow passageway.
Several hundred of these bags may be placed in
parallel in this way. The gas passing through the
bags deposits the entrained solids on the inside of
Figure 22.42. Plate-and-frame pair of simple corner-hole nonwashing design with
closed discharge and waffle-grid surface. (Courtesy T. Shriver and Company.)
the bags. Periodically the bags are cleaned by
shaking the rack to which they are fastened. The
household vacuum cleaner operates on this same
principle.
Plate-and-Frame Filter Press. The filter press
has long been the most common filtering device
throughout the chemical industry. Although it is
now being replaced in large installations by con-
tinuous filter devices, it has the advantages of low
first cost, very low maintenance, and extreme flexi-
bility. On the other hand, the need for periodic
manual disassembly represents a labor requirement
that is often excessive.
The filter press is designed to accomplish a
variety of functions, the sequence of which is
controlled manually. During filtration the press, 1)
permits the delivery of feed slurry to the filter
surfaces through its own duct, 2) permits the
forcing of feed slurry against the filter surfaces, 3)
permits filtrate that has passed through the filter
surfaces to exit through its own duct, while it 4)
retains the solids that were originally in the slurry.
During the wash sequence the press 1) permits
delivery of wash water to the filtered solids
through its own duct, 2) permits the forcing of
wash water through the solids retained in the filter,
and 3) permits wash water and impurities to leave
through a separate duct. Filter design can include
four separate ducts as indicated above or can allow
for only two ducts where the contamination of the
liquid products is not important. After the wash
sequence the press is disassembled, and the solids
may be collected manually or merely removed and
discarded. -
The most common filter-press design consists
of alternate plates and frames hung on a rack and
forced tiqhtlv together with a screw- or hydraulic-
closing mechanism. Figcre 22.42 shows a plate-
01
Frame
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
Fixed head Solids collect
Clear - filtrate
outlet
Figure 22.43. Schematic diagram.of filter press in operation. (Courtesy T. Shriver
and Company.)
and-frame pair; Figure 22.43 is a diagram of a filter
press in operation, and Figure 22.44 is a photo-
graph of a typical filter press. To set up this press,
the plates and frames are hung alternately on the
side rails of the press using the side lugs on the
plates and frames. The filter medium is then hung
over the plates extending over both faces of the
plate. The filter medium may be canvas or syn-
thetic cloth, filter paper, or woven wire. Holes are
cut in the cloth to match the channel holes in the
plates and frames. If cloth is used, it may be
necessary to preshrink the medium so that the
hoies will continue to match. When the filter cloths
are aligned with the plates and frames, the press is
closed with a hand screw or in large sizes by
hydraulic- or electric-closing devices. When the
press is closed, the filter medium acts as a gasket,
sealing the plates and frames and forming a con-
tinuous flow channel from the holes in the plates
and frames, as shown in Figure 22.43. Feed slurry
is then pumped to the press under pressure and
flows in the press of Figures 22.42 and 22.43 into
the bottom-corner duct. This duct has outlets into
each of the frames, so the slurry fills the frames in
parallel. The solvent, or filtrate, then flows through
the filter media while the solids build up in a layer
on the frame side of the media. The filtrate flows
between the tilter cloth and the face of the plate to
an outlet duct. As filtration proceeds, the cakes
659
build up on the filter cloths until the cakes being
formed on each face of the frame meet in the
center. When this happens, the flow of filtrate,
which has been decreasing continuously as the
cakes build up, drops off abruptly to a trickle.
Usually filtration is stopped well before this
-
occurs.
In many cases, it is desirable to wash the filter
cake in order to remove the solvent trapped in the
cake or dissolve impurities from the cake. In a filter
using the plates of Figure 22.42, the washing could
be done by feeding wash water into the feed
opening, but if the cake nearly fills the frames, the
wash water may be blocked from passage just like
the feed slurry. A better system is afforded by the
through-washing plate-and-frame press. In this
press, a separate channel is supplied for the wash-
water inlet: in closed-delivery presses, a separate
exit channel is also supplied. Wash water enters the
channel, which has ports opening behind the cloths
at every other plate. The wash water then flows
through the filter cloths, through the cake built up
in an entire frame, through the filter media at the
other side of the frames, and out the discharge
channel. The flow path is shown in Figure 22.45.
In this figure, an open-discharge filter press is
pictured. In the closed-discharge press shown in
Figure 22.43, the outlet streams would be col-
lected into a common duct like the inlet duct.
660 APPLICATIONS TO EQUIPMENT DESIGN

Figure 22.44. A filter press operating in a chemical plant separating neutralizing

salts from 1,2,&hexanetrio\ before distillation. (T. Shriver and Company, Courtesy
Am. Chem. Soc.)
Note that in this press there are two kinds of
plates: those with ducts to admit wash water
behind the filter media alternating with those with-
out such ducts. In closed-delivery presses, the alter-
nate plates often have ducts to permit the with-
Cloth Frame
drawal of wash through a channel separate from
Wast,, the one used to remove filtrate. These various

water
inlet
_r:==,r= plates and frames are coded with buttcns on the
top edge. One button signifies a nonwashing plate,
two buttons a frame, and three buttons a washing
plate. Figure 22.46 shows the plates and frame for
a through-washing open-delivery filter press.
Filter presses can be made of any construc-
tion material desired, such as wood, cast iron,
rubber, and stainless steel. They can be buiit for
slurry pressures up to 1000 psia. They can handle
the filtration of heavy slurries or the "polishing" of
a liquid containinq only a faint haze of precipitate.
Closed Oosed For filtraticns in which only :i small amount of

Figu,e 22.45. Schematic diagram of through-washing
in a plate-and-frame filter press with open delivery.
Note one-button, two-button, three-button coding on
the top edge of the plates and frames. (Courtesy T.
Shriver and Company.)
precipitate is to be handled, recessed plates car. be
used, thus eliminating the need for frames. Filter
presses are built with plates up to 48 in. x 72 in. in
area and with mechanical devices for opening the
press, for moving each plate so that the cake can
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 661

Nonwash plate/ Frame •

Wash otate "

Figure 22.46. Plates and frame for a through-washing open-delivery filter press.
{Courtesy D. R. Sperry and Company.)

drop off, and for closing the press. Thus, a large
filter press could be handled using only one
operator whose job would be to stop the automatic
cycle and remove the cake if the cake stuck in
place when a plate was moved, despite such auto-
mation, filter presses are inevitably cyclical in
operation, and thus are usually used in batch
processes or processes with modest throughputs.
Other Batch Fillers. Many filters, while still
batch filters, do not require the complete dis-
assembly for cleaning that is necessary with
a plate-and-frame filter press. A few of these
are shown in Figures 22.47, 22.48, 22.49, and
22.50. All these filters use varieties of filter leaves.
The filter leaf is a hollow, internally supported plate,
as shown in Figure 22.48, that is permanently

Figl!re 22.47. Cross-sectional schematic diagram of a horizontal plate filter. Patented

scavenqer-piate filters final fiow of each batch. The scavenger valve is opened during
precoating and closed until the end of the cycle. Then the outlet valve is closed,.
scavenger valve opened, and remaining liquid is filtered through the scavenger plate
by introducing air or gas pressure through the intake. (Courtesy Sparkler Mfg.
Company.)
 662 APPLICATIONS TO EQUIPMENT DESIGN

Filter cake

Figure 22.48. Cutaway view of a vertical-leaf filter and sectional diagram showing
filter-leaf construction. (Courtesy Industrial Filter & Pump Mfg. Co.)

covered with filter medium. The slurry to be
filtered fills the space around the leaf and is forced
by pressure on the slurry or vacuum within the leat
to flow through the leaf. Filter cake is built up on
the outside of the leaf and filtrate passes from
within the leaf to the fiitrate-discharge system.
When a cake of the desired thickness is built up on
the leaves, the filter is opened, and the leaves are
either removed for cleaning or cleaned in place
manually or automatically by sluicing away the
solids. Of the filters shown, the horizontal-plate
filter (Figure 22.47) is particularly well adapted for
the final clarifying of solutions containing minute
quantities of solids because of the ease of applying
a filter-aid precoat. Filter aids are open-structured
incompressible solids that may be deposited on the
filter cloths to serve as a high-efficiency filter
medium. They are further discussed below. The
vertical leaf filter (Figure 22.49) is built in very
large sizes and can be opened particularly rapidly
for cleaning. The Sweetland filter (Figure 22.50) is
made in two half-cylinders. The bottom half opens
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
Figure 22.49. Vertical leaf filter. (Courtesy AMETEK, Inc., Process Equipment
Division.)
Figure 22.50. A Sweetland pressure filter
closed for filtration. (Courtesy Dorr-Oliver,
Inc.)
downward by releasing the quick-opening cam
locks to expose vertical disk-shaped leaves that are
cleaned in place.
Like the filter press, any of these leaf filters
can be automated to operate the filtering, cleaning,
and precoating operations without manual atten-
tion. Still, their cyclical operation tends to limit
their use to batch and modest-capacity continuous
operations.
663
Continuous Filters. Modern, high-capacity
processing has made the development of continu-
ous filters mandatory, and several varieties are in
common use. In such filters, the slurry is fed con-
tinuously, and the cake and filtrate are produced
continuously.
The horizontal rotary filter of Figures 22.51
and 22.52 is particularly well adapted to the filter-
ing of quick-draining crystalline solids. Its hori-
zontal surface prevents the solids from falling
off or from being washed off by the wash water,
and an unusually heavy layer of solids can be
tolerated. This filter consists of a circular hori-
zontal table that rotates around a center axis. The
table is made up of a number of hollow pie-piece-
shaped segmer.ts with perforated or woven metal
tops. Each of the sections is covered with a suitable
filter medium and ls connected to a central valve
mechanism that appropriately times the removal of
filtrate and wash liquids and the dewatering of the
cake during each revolution. Each segment
successively receives slurry, is sprayed with wash
liquid in two applications, has its cake dewatered
by pulling air through the cake, and has its cake
scooped off the surface by tne discharge scroll. In
some models the wedge-shaped segment rotates
and the cake is blown off of it as it reaches the end
of the circle.
The rotary-disk vacuum filter is shown in
664 APPLICATIONS TO EQUIPMENT DESIGN

Figure 22.51. Horizontal-rotary vacuum filter showing arrangement of piping for

two stages of wash-and-scroll cake-removal mechanism. One section of filter
medium is removed to illustrate deck support and drainage slope. (Courtesy
Dorr-Oliver, Inc.)
Figure 22.53. This filter gives an especially high a cake on its surfaces while the filtrate passes
filtration rate for a given floor space. The filtering through to a central discharge system. The leaf
medium is again a wedge-shaped leaf covered with then carries the filter cake through the upper halt
a filter medium. Here, the leaves rotate in a vertical of its rotation while air pulled through the cake
plane around a horizontal axis. The slurry to be dries it. The cake is scraped oft the leaves by
filtered fills the tilter basin almost up to the filter doctor knives or is blown oft with compressed air
axis. As the leaf dips through the slurry, it collects fed inside the leaves before they dip again into the

I
l Filter
I
Rotating pa-i essembty

Figure 22.52. Cross-sectional view of a rotary horizontal vacuum filter showing

filtrate-removal system, filter cloth, and discharge scroll. (Courtesy Filtration
Engineers Division, American Machine and Metals, Inc.)
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
Figure 22.53. Large rotary-disk filter, built for coal fines
dewatering, has disks 17 ft in diameter that operate at
about 4 rpm and are 55 percent submerged in the slurry.
(Courtesy AMETEK, Inc.)
slurry. No provision is made for washing the cake.
If a filter cloth becomes worn or torn, the single
segment may be removed and replaced with a new
one relatively quickly.
The rotary-drum vacuum filter with string dis-
charge is shown schematically in Figure 22.54, and a
somewhat different rotary vacuum filter is
pictured in Figure 22.55. With either unit, the
cycle is very much like that of the horizontal
vacuum filter. Filter cake is picked up from a
slurry pool by dipping the drum surface and apply-
ing vacuum. The cake is then carried around the
drum where it is successively washed and
dewatered by the continuous application of
vacuum to the inside of the drum. The string-
discharge svstem leads the cake away from the
drum and over a roll with sharp curvature that
causes the cake to drop off. Figure 22.54 shows
the agitator in the slurry pool that prevents settling
of the slurry, the discharge valve that controls
withdrawal of filtrate and wash liquids and the
dewatering of the cake, the drain lines that apply
vacuum to the drum surface, and the drum surface
construction. For coarse particles that settle
rapidly and form a porous cake, a feed hopper on
top of the drum is more satisfactory than the
dipping hopper shown.
The unit shown in Figure 22.55 differs from
that in Figure 22.54 mainly in the cake-discharge
mechanism. This unit uses a motor-driven doctor
knife that moves toward the filter drum at an
665
extremely slow rate. This permits the application
of an initial, or precoat, layer of filter aid perhaps
1 to 2 in. thick. After this layer is in place the
slurry is fed to the dip tank, and the filter cake
builds up on the precoat layer of filter aid. The
doctor knife removes the filter cake and a very thin
layer of filter aid. As filtration progresses, the
filter-aid layer gets progressively thinner until a
new layer must be applied. The doctor knife moves
so slowly that a precoat layer will last as long as a
week.
Filter Media and Filter Aids. As mentioned
(
above, filter media consisting of cloth, paper, or
woven or porous metal may be used. The criteria
upon which a filter medium is selected must
include ability to remove the solid phase, high
liquid throughput for a given pressure drop,
mechanical strength, and chemical inertness to the
slurry to be filtered and to any wash fluids. Of
course, each of these considerations is tempered by
the economics involved, so that the filter operator
tries to choose a medium that meets the required
filtration standards while contributing to the
lowest possible overall filtration cost.
A variety of filter media have been tested for
pore-size distribution (23). The tests showed that
with a woven-cloth medium both interfiber and
interyarn pores exist with 30-50 percent of the
total pore volume being made up of interfiber
pores. The interfiber pores are the spaces between
the fibers making up a single thread of the cloth,
whereas interyarn pores are the spaces between the
threads woven to form the cloth. These interfiber
pores were less than 10 microns in radius. The
interyarn pores were found to range between 70
and 200 microns in radius. With a felt medium, all
the pores are interfiber, but, because of the
random fiber orientation, they were found to vary
between about 2 and about 180 microns nearly
following a logarithmic normal-distribution func-
tion with a mean between ,0 and 20 microns. The
sintered-metal media tested had a much more uni-
form pore-size distribution with 90 percent of the
pore volume falling between 20 and 50 microns in
radius.
In operation, some filter-cake solids usually
penetrate the filter medium and fill some of the
pores. As filtration continues, these particles are
thought to bridge across the pores and cake begins
to form on the face of the medium. In normal
cases, between 5 and 25 percent of the pore
volume of the filter medium is filled with solids. As
a result, the resistance to flow through the medium
increases sharply. In some cases, the solids fill the
 666 APPLICATIONS TO EQUIPMENT DESIGN
Figure 22..54. Schematic drawing of string-discharge rotary-drum vacuum filter.
(Courtesy AMETEK, Inc.)
filter medium to such an extent that the filtration
rate is seriously reduced.
Filter aids are often used to speed filtration or
to make it possible to collect more completely the
finest particles held in the slurry. The filter aids are
finely divided, hard-structured solids that them-
selves form an open, noncompressible cake. The
most common example is diatomaceous earth,
which consists of the skeletons of very small pre-
historic marine animals. This material is mined from
large surface deposits mainly in California, Oregon,
and Nevada. It is practically pure silica and has a
very comp I icated structure. Applied as a precoat
on the filter cloth, the filter aid acts as the primary
filter medium and permits complete removal of
very fine solid particles from the slurry. Another
method of application is to mix the filter aid into
the slurry. Here, it distribute, throughout the cake,
keeping the cake relatively open for flow and con-
tinuously supplying a large surface for adhesion of
very finely divided solids. This action is partic-
ularly valuable when filtering colloidal solids that
form a very dense and compressible cake, which is
not to be recovered. At present, roughly 500,000
tons of diatornaceous earth are used annually as
filter aid in the United States.
Filtration Calculations=General Relations.
The flow of filtrate through the filter cake
should be describable by any of the general equa-
tions for flow through packed beds, such as
Equation 22.86. Actually, in almost all practical
cases, flow is laminar, and the Carman-Kozeny
 PARTICULATE SOLIDS FLOW AND SEPARA110N THROUGH FLUID MECHANICS 667

Figure 22.55. Continuous-vacuum precoat filter, 5 ft, 3 in. diameter x 8 ft face.

(Courtesy Dorr-Oliver, I nc.J
equation In order to integrate Equation 22.116 to
obtain a relation usable over the entire process,
(-1!.P),gc (1 - €)2 µv, only two variables may appear in the equation. As
--- = 180 --- -- (22.76)
written, the quantities V, e, L, (-1!.P),, S0 and e
L €
3
o/ may all vary. The cake thickness (L) may be related
applies. This equation relates the pressure drop , to the volume of filtrate by a material balance,
through the cake to the flow rate, the cake since the thickness will be proportional to the
porosity and thickness, and the solid-particle volume of feed delivered to the filter.
diameter. Some modification of the equation is

necessary so that the measurable variables of filtra- LA(1-E)p,=w(V+ELA) (22.117)

tion can be introduced into it. where
In its more usual form, it is written in terms
of the specific surface area of the particles by p, = density of the solids in the cake

incorporating Equation 22.73. w weight of solids in the feed slurry per

=
volume of liquid in this slurry
6 6 V = volume of filtrate that has passed through
Dp = -- = - (22.73a)
AP So the filter cake ·
VP The final term of Equation 22.117 (ELA) repre-
where sents the volume of filtrate held in the filter cake,
S0 � specific surface area cf particle, ft2/ft3 It is normally infinitesimal compared to \/, the
of solid volume filtrate that has passed through the bad. Assuming
Thus, this term negligible and ccmbining Equation
2

(-1!.Pl,gc 5(1 - e) µv,S/ 22.116 with Equation 22.117 to eliminate L gives

(22.115)
L €3

1 dV (-1!.Pl,gc
Solving this equation for the velocity of flow gives
A dB wV c,µwV
3 5-µ(1 - E)Sa2
(-Cl.P),gcE 1 (dV) Ap, A
v,= =- - .(22.116)
5LµS0 2 (1 -d 2 A de
(22.118)
where

dV/d8 = filtration rate, that is, the volume of where c, is the specific cake resistance, defined as
filtrate passing' through the bed per
5(1 - E)S 02
unit time c, = (22.119)
A = filtration area
668 APPLICATIONS TO EQUIPMENT DESIGN
A similar equation in terms of L could also be
obtained by eliminating V between Equations
22.116 and 22.117.
Equation 22.118 is the basic filtration equa-
tion in terms of the pressure drop across the filter
cake alone. The collection of all terms involving the
filter cake properties into the specific cake resist-
ance does not infer that the resistance (c,) will be
constant for a given feed slurry, regardless of filter-
ing pressure drop or of filter type or size. The
specific cake resistance may not even remain con-
stant throughout a given filter operation at con-
stant (-t.P)1 because of variations in e and S0. The
void fraction (e) usually varies with variation of the
compacting stress applied to the bed. This stress
will be directly proportional to (-t.P)IL, and,
since L varies throughout the process, e may also
vary. Both e and S0 are sensitive to the degree of
flocculation of the precipitate in the feed. The
flocculation may vary with the turbulence of flow
of ·slurry fed to the press and hence may be a
function of the filter rate. However, in most
constant-pressure filtrations, c, is constant except
in the initial moments of filtration when the flow
rate is very high and the form of the filter cake has
not been fixed. In fact, for many filter cakes, c, is
relatively. insensitive to changes in (-t.P),. Such
cakes are said to be incompressible, although
probably no cake is completely incompressible.
Filtration Calculations-Inclusion of Filter-
Medium Resistance. Equation 22.118 is
expressed in the familiar form of a rate propor-
tional to a driving force divided by a resistance.
Here, both driving force and resistance apply to the
filter cake alone. Practically any (-h.P) measured
will at least include the pressure drop across the
filter medium and will probably include the pres-
sure drop of various flow channels before and after
the actual filtering area. If such an overall pressure
drop is to be used in an equation like Equation
22.118, the resistance term must also include the
flow resistances of the additional parts of the
apparatus. Since these resistances are arranged in
series, Equation 22.118 becomes
dV (-t.P,)gc
(22.120j
A de
Jl(°':V +RM)
where RM has the ur.its (ft)·-·1 and represents
the resistance of filter medium and piping to filtrate
flow. For convenience in analyzing filtration·
performance data, the resistance of filter cloth and
flow channels is usually expressed in terms of an
equivalent volume of filtrate
dV (-t.P,)gc
A de
µ: (V+ Ve)
(22.121)
Here, v. is the volume of filtrate necessary to
build up a fictitious filter cake, the resistance of
which is equal to the resistance of the filter
medium and the piping between the 'pressure taps
used to measure (-t.P,). The filter-medium resist-
ance of significance here is the resistance of the
medium with the pores partially blanked with filter
cake and with the initial layer of filter cake upon
which the bulk of the filter cake will be built in
place. It is considerably greater than the resistance
to water flow of the clean filter cloths.
Filtration Calculations-Integration of Filtra-
tion Equations. The integration of Equation
22.121 can now be performed if (-t.P,) and c, can
be related to the variables of V and e. As men-
tioned above, there are many slurries for which c, is
substantially constant throughout the filter cycle.
In the discussion that immediately follows, only
those slurries that form incompressible cakes will
be considered. Thus, c, = constant.
The variation in (-t.P,) through the filter
cycle depends upon the operating procedure and
the type of filter used. In sand filters, (-t.P,)
varies with the liquid head over the filter bed. In
rotary vacuum filters, it will be constant at some-
what less than 1 atm. In filter presses and pressure-
leaf filters, it can be varied at the will of the
operator within the equipment limits. The operator
may control the pressure at a fixed value through-
out the entire run. This may be the easiest method
of operation if the slurry is fed from a pressur-
ized tank or by hydrostatic pressure from a storage
tank. The operation may be run at a constant feed
rate, as would be done by delivering the feed
through a positive-displacement pump and continu-
ing the run until the filter feed pressure reached a
limit. Often, the operation starts at a constant
throughput and continues in this way until the.
pressure reaches a predetermined level, after which
the pressure is held constant. This has the advant-
age of forming a rather loosely knit initial cake and
forcing a minimum amount of solids into the pores
of the filter medium. It is also a convenient
method of operation, for initially the filter will
accept the entire throughput of the teed pump at
relatively low feed pressure. As the cake builds up,
the feed pressure also increases, but the rate of
delivery to the filter falls off only slightly even
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 669

with a centrifugal feed pump. When an economic-

ally optimum or maximum safe pressure is reached,
the run is continued controlling the feed pressure
at this value.
Filtration Calculations for Incompressible
Cakes. ·Performing the integration of Equation
22.121 for incompressible cakes (ex= constant) and
for operation at constant (-t,P,) gives

v • g A2(-t,P)
Jo !V+ v.ldV= Jo c
µcxw
'de

(22.121 a)
V2 g A2(-t,P)
- + vv.= c ' e (22.122)
2 µexw
or

µexw (V 2
Volume of filtrate collected I VJ
8= -+VV )
9cA2(-'t,P,) 2 e
Figure 22.56. Typical filtration-rate data for
(22.122a) constant-pressure condition.

from which the time necessary to pass any given
volume of filtrate can be calculated.
The solution of Equation 22.122a requires
evaluation of the two constants ex and Ve. The
specific cake resistance (ex) may perhaps be evalu-
ated from the properties of the filter cake if€ and
S0 are known for the particular filtration condi-
tion. However, the volume of filtrate equivalent to
the filter-medium and piping-flow resistances (V.)
must be determined from pilot-filtration data. For
this reason, it is usual practice to evaluate both ex
arid v. from a pilot-filtration run using the actual
slurry to be filtered under conditions as close to
those to be employed in the p!ant as possible. To
permit evaluation of these constants from experi-
mental data, Equation 22.121 is inverted to give
de µexw
dV 9cA2(-t,P,) (V+ V,)
(22.123)
This equation is a straight-line relation between
a balance should be struck between areas of the
bars above the line and areas not in the bars but
below the line. The slope of this line is then
cxµwlgcA 2 (-t,P,)
from which ex may be determined, and the inter-
cept is
[µexwlgcA�(-t,P,l] v•.
The intercept divided by the slope is ve- The data
plotted in Figure 22.56 are typical in that the
.initial measuring periods give highly irregular data.
This irregularity occurs as the cake begins to form
on the filter medium and may be caused partly by
higher flow rates giving more tendency toward
turbulence and partly bv the instabilitv of the
ii lrer cake.
If the test run is made under conditiuns of
constant rate rather than constant pressure, Equa-
tion 22.121 can be rearranged to

d8/dV and V if (-t,P,) is held constant. Thus, µcxw /dV)

-t,P,= -A21- (V+ V.) (22.121b)
from actual constant-pressure filtration data, d8/dV 9c \ de
may be plotted as the ordinate as a function of
Equation 22.121b gives a straight line for constant
Vas the abscissa to give a straight line. Such a plot dV/d8 (that is, for constant filtration rate) when
is shown in Figure 22.56 where measurements have
(-t,P,) is plotted against V. Herc the slope is
been taken over finite time differences rather than
differentials. The bar-graph technique is used to (µexwlgcA2 )/(dV/d8),
indicate that the value (t, VI t,8); represents the and the intercept is
average rate during the interval between V and
V + t, V. Therefore, in plotting the correlating line, [(µexwlgcA2 )/(dV/d8)] v.
 670 APPLICATIONS TO EQUIPMENT DESIGN

Again, the unknown constants can be determined may be graphically solved for DI and v. by plotting

filtrate volume. Once DI and v. are determined,

they can be used directly to predict the path of a t:,0
constant-rate filtration process, but only if the V,I e, t:,0

!:,V
cake is incompressible.
Equations equivalent to Equation 22.121 0 0 1.8 9.0
through 22.123 can be developed in terms of cake 0.2 1.8 2.4 12.0
thickness (L) rather than filtrate volume. The 0.4 4.2 3.3 16.5
resulting equations are convenient when dealing 0.6 7.5 3.7 18-, 5
0.8 11.2 4.2 21.0
with leaf or rotary vacuum filters where it is desir- 1.0 15.4 5.1 25.5
able to set the filtration cycle by the thickness of 1.2 20.5 6.2 31.0
filter cake built up on the surfaces. 1.4 26.7 6.7 33.5
1.6 33.4 7.6 38.0
1.8 41.0 7.8 39.0
2.0 48.8 8.9 44.5
Example 22.13. Ruth and Kempe (49) report 2.2 57.7 9.5 47.5
the results of laboratory filtration tests on a pre- 2.4 67.2 10.1 50.5
cipitate of CaC03 suspended in water. A specially 2.6 77.3 11.4 57.0
designed plate-and-frame press with a single frame 2.8 88.7
was used. The frame had a filtering area of 0.283 ft2
and a thickness bf 1.18 in. Aii tests were con-
ducted at 66°F and with a slurry containing The plotted data are shown in Figure 22.57. The
/o.0723 weight fraction CaC03. The density of the slope of the line drawn through the differenced
dried cake was 100 lb/ft3. Test results for one run data is 18.05 s/12, and the intercept is 5.9 s/1. From
are given below. these values,

t:J' = 40 psi= constant

Ve = 5.9/18.05 = 0.327 I
Volume of Filtrate, I Time, s
Dlµw 18.05
0.2 1.8 ---=
2
14 500 s/(ft3)2
0.4 4.2 (0.0353) '
0.6 7.5
0.8 11.2
1.0 15.4 60
1.2 20.5
1.4 26.7 �5
1.6 33.4
50
1.8 41.0
2.0 4a8 45
';;;-

2.4 67.2
2"
.
;5
2.6 !7.3
Slope:18.05 s/12

2.8 13&.7 30
iE \

Determine the filtrate volume equivalent in resist- i

·g
25 -

20
ance to the filter medium and piping (V.i. the "'
specific cake resistance (DI), the cake porosity (€),
and the cake specific surface (S0).

Solution. The filter data given are differenced

as shown tabularly below, so that the relatior.
0.4 0.8 12 1.6 2.0 2.4 2.8

Fiitrate volume IV), liters

t:,6 µDlw
(22:f"23a) Figure 22.57. · Graphical determination of filtration con·
!J.V gcA2(-M')V+ V.) stants for Example 22.13.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 671

Solving for c,µw and for c, yields To calculate the slurry density,
c,µw = 14,500 x 32.2 x (40 x 144) x (0.283 x 2)2 1 1
- = - x 0.9277 + - x 0.0723
1 .
p 62.4 183
= 8.60 x 108 lb/s ft3 .,.
= 0.01485 + 0.00040
0.0723
w = -- = 4.86 lb CaC03/ft3 of H20 1
0.9277 p = --- = 65.5 lb/ft3
0.01525
62.4

27, 100
8.60 x 108 Volume of slurry= --- x 7.48 = 3100 gal
c,= =2.41 x 1011 ft/lb 65.5
1.1 x 0.000672 x 4.86
The required filtration time can be found by
The cake porosity can be directly calculated solving Equation 22.122a, since the constants "'
from the measurement of dry-cake density. and Ve have already been determined.

Density of solid CaC03 = 2.93 x 62.4 =

e= µc,w (v22 + v.v ) (22.122al
183 lb/ft3 = p, 2( -
gcA -!if,)
100 In this equation, Vis the volume of filtrate, and it
E = 1 - - = 1 - 0.547 = 0.453
183 can be determined from the material balance based
upon the cake.
The specific cake surface (S0) can now be deter-
mined from the values of c,, E, and p, calculated LA(1 -E)p,=w(V+ELA) (22.117)
above. 19.60(1 - 0.453)183 = 4.86(V + 19.60 x 0.453)
5(1 - E)Sa2 V= 395.0 ft3

c,=
Inserting this value into Equation 22.122a gives
2.41x1011x183xo.4532 8.60 x 108
2 12 e
So = O = 3.35 x 10 = --�=c----
5 x .547 32.2 x 1882 x 40 x 144
S0 = 1.83 x 106 ft2/ft3 of solids 3952 )
x ( -- + 0.327 x 0.0353 x 395

Example 22.14. A 30 by 30 in. plate-and-

frame filter press with twenty frames 2.50. in. thick
is to be used to. filter the CaC03 slurry used in the
test of Example 22.13. The effective filtering area
per frame is 9.4 ft2, and Ve may be assumed to be
the same as that found in the test run. If filtration
is carried out at constant pressure with (-!iP) =
40 psi, determine the·volume ot slurry that will be
handled until the frames are full, and the time
required for this filtration.
Solution. The required slurry volume can be
directly calculated from the volume of frames and
the slurry concentration.
9.40 2.50
Volume of frames= x x 20 = 19.60 ft
-- --
2 12
Solids in cake= 19.60 x 100 = 1960 lb
' 1960
Weight of slurry fed= --- = 27, 100 lb
-<, 0.0723
e = 0.131 x (78,000 + 4.5) = 10,200 sec = 2.84 hr
Filtration Calculations for Compressible
Cakes. As mentioned before, most chemical pre-
cipitates form compressible filter cakes, in which
higher compressive forces deform the solid
particles, break up flocculent aggregates, and force
the particles closer together. Empirically, it has
been found that at moderate pressures the relation
"="a+ si-s», )' (22.124)
where
O'o ::::: specific cake resistance at zero com-
pressive pressure, a constant
s = cake compressibility factor, a con-
3 stant over moderate pressure ranges
b = a constant
-!:,Pc= pressure drop acrcss the cake
holds for most of these precipitates. The values of
c,0, b, and s can be determined from a series of
constant-pressure filtration tests .
672 APPLICATIONS TO EQUIPMENT DESIGN
A simpler equation has been even more gener- permeability" experiments. The slurry to be tested
ally used, is enclosed in a cylinder with a porous bottom, and
a filter cake is built up on this porous surface by
(22.124a) letting the filtrate drain through it. A piston is then
where 010' and s' are empirical constants with put into the cylinder and slowly forced down onto
significance equivalent to 010 and s of Equation the filter cake. The piston is then loaded with an
22.124. This equation is obviously wrong at low increasing series of weights. At each piston loading,
values of -b.Pc. for .at -b.Pc = 0 it predicts a the porosity is determined by noting the piston
specific cake resistance of zero. It is convenient, position. Filtrate is fed to the filter cake, and the
however, for the constants are readily determined, value of 01 determined by solving Equation 22.126
and it can be used over restricted pressure ranges. assuming 0< to be constant at any given loading.
A more fundamental approach to compres- Then, S0 can be calculated from the values of 0<
sible filter cakes was made by Grace (22). He noted and e already determined. Figure 22.58 shows
that the compressive force on a particle in a filter typical compression-permeability data of Grace
cake depends upon its position in the cake, varying (22) giving e, 50, and°'• as functions of /'J.P. Grace
from a maximum at the cloth to a minimum at the showed that the results of such experiments can be
cake surface. This gradation even arises for the used under industrial filtration conditions.
simple case of pressure filtration on a flat vertical
surface, since the hydraulic pressure applied to the
Example 22.15. Determine the average
surface of the cake equals the sum of the
specific cake resistance (0<) that applies for the
mechanical compressive pressure and the hydraulic
pressure at any point within the cake. In this case, filtration of Type B ZnS slurry at a ·(-b.Pc I of
the compressive pressure (P,) varies from zero at 70 psia. Compression permeability data for this
slurry are given in Figure 22.58.
the filter-cake-slurry interface to P1 -P2 at the
filter medium, where P1 is the pressure on the cake Solution. The specific cake resistance at a
surface and P2 is the pressure at the face of the fixed compressive pressure (0<p) is shown as a func-
filter medium. This constancy of hydraulic plus tion of compressive pressure on line 5 of Figure
compressive pressures can be shown by making a '22.58a. The data are replotted in arithmetic co-
simple force balance at a point within the filter ordinates on Figure 22.59 with 1/a,, as a function
cake. For more complicated situations, a similar of P.
relation exists between bed depth and compressive
force. If the relations among e, S0, and P can be
determined, Equation 22.121 can be integrated to
P, psia .CX,,, ft/lb .l., 10-12
fY.p
give the filtration rate as a function of filtrate
volume. This integration requires the writing of 1 0.85x 1012 1.19
Equation 22.121 as a differential function of P. 10 6.0 x ,o12 0.17
20 10.5 x 1012 0.095
dV -·gcAP, P,-P, e3 30 16 x 1012 0.063
A d6 = 5wµV- Jo (1 - e)S 2 dP 50 29 x 1012 0.034
0 70 45 x 1012 0.022

(22.125i
Alternativelv, Equation 22.125 can be simplified The specific cake resistance applicable to the
by inserting the specific cake resistance (0<•) applic- entire filter cake can be obtained from Equation
able at a point in the cake. 22.126. From this equation
dV -gcA f P, -P, dP
--=--
A de wVµ 0
-
°'P
(22.126) 1 dV -gcA
-----
f P,-P, dP
----·--
-gcA (b.P)
A de wVp. o °'P wVµ a:
which Is again restricted to pressure filtration on a or
flat vertical surface. P1 and P2 here refer to the
pressures at the filter-cake-slurry interface and at b.P
01=-----
the filter-medium surface, respectively.
The relations between P and°'•, e, and S0 can
(-P, dP
be experimentally determined by. "compression- . 0 °',,
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 673

I
I
-c
£
���
I\
.. 0
.. ... --j- +n �§ I I i\ I
'" II J!:1 I I ' s')I
D
<,
.i ,s I
:; 1013
5,, 111 I 1 11 § 106 I

i� � �
�
v
,.,
c
.. 11/ !I 4 ,�,

a
e w12 ,,'" i IGv '1ili-1r I [I: 1-, ii . .
!'
B .. 4
I
3
I I
I ''
v
o � ' tt, Ill II
'",..,._
I
; 1011
l-..iJ°"'k,, I' Ii I Ii I 2
c ";{.,",.I
&

.' I ,,' I II

1010 '
·I I 1,I I
I IO 100 l.000 10,000 0.3 0-5 0.7 0.9
Compressive pressure (P, ), psra

(a) (bl

1.0
I
o.s
-- ..,._-; ��
M
,-.§...
-0....
--. -..,
,,.., -....
�-
0.8
--,...., 0...
�
I "� '2 -,

z, 0.7 � r-« >--- ,..

·a;
e0 2
a
i 0.6 - �� ' I
u

I
4 -+- -i-u
J_ l
�
I
0.5
t-1 11

I
I
Ii I
I
I

I I
' I

04 � I
I I 11 I
-

0.3
100
j 11 11
1.000
I i i
!O

Compressive pressure {Pr), psla

(c)

Figure 22.5e. Compression perrneabilitv data (23). (3) Specific cake resistance.
(b) Porosttv as a function of specific surface. (c) Cake porosity. {By µermission of
Chem. Eng. Prog., copvriqht © 1953_)
1-Superlite CaC0 3 (flocculated). pH= 9.8 4-R-110 grade Ti02, pH = 3.5
2-Superlite CaC0 3, pH = 10.3 . 5-Zns, Type B, pH = 9.1
3-R-110 grade Ti02 (flocculated), pH = 7.8 6-ZnS, Type A, pH= 9.1
674 APPLICATIONS TO EQUIPMENT DESIGN
may be seen by examining Figure 22.45. There-
0.24 fore, the wash rate in this unit is one-quarter the
final filter rate if the filtrate is very nearly water
's: 0.22 and the pressure drop for washing is the same as
the final filtering pressure drop.
In the dewatering part of a filtration cycle, air
is drawn through the filter cake, pulling the filtrate
or wash water remaining in the pores of the cake
out ahead of it and partially through-drying the
cake. In part of this operation, two-phase flow
occurs. Methods have been developed for calculat-
ing the (-t.P,) versus flow rate relation in this
period (6, 8, 9, 10). They are particularly valuable
0 0.08 in vacuum-filter design because this period of oper-
ation places the greatest load on the vacuum
i 0.06
system. Such calculations are beyond the scope of
� 0.04 this book.

0.02

I
Centrifugal Filtration
0
0 10 20 30 40 50 60 70 80 A filter operation can be carried out using a centri-
Compressive pressure IP), psia
fugal force rather than the pressure force used in
Figure 2259. Compression-permeability data for ZnS the equipment described above. Filters using
Type B suspension presented as 1/o.p as a function of P. centrifugal force are usually used in the filtration
of coarse granular or crystalline solids and are
The indicated solution is carried out from Figure available for batch or continuous operation.
22.59 by graphical integration. In this integration Batch centrifugal filters most commonly con-
the numerical result is greatly influenced by the
sist of a basket with perforated sides rotated
value of ex,, at P = 0. Therefore, the data have been
around a vertical axis, as shown in Figure 22.60.
replotted with ex,, as a function of P. This curve can
An electric motor, located either above or below
be more readily extrapolated and gives Ol.p = 0.25 x
the basket, turns it at rates usually below
1012 at P = 0. Then, carrying out the graphical
integration 4000 rpm. Basket diameters may be as great as
48 in. The slurry is fed into the center of the
P,-P, dP rotating basket and is forced against the basket
J -= 15.51 x 10-12 sides by centrifugal force. There, the liquid passes

O Ol.p through the filter medium, which is placed around

and
- 70 - 12
Ci- 15.51 x 10-: 2 -- 4.51 x 10
Filtration Calculations-Washing and Dewater-
ing. After filtering is completed, the washing and
dewatering operations are done with the filter cake
in place. Since the cake thickness is unchanged, the
wash rate usually varies directly with the pressure
drop. If the wash water fol lows the same path as
the slurry and is fed at the same pressure, the wash
rate will equal the final filtration rate, as given by
Equation 22.121. But in the through-washing filter
press, any given volume of wash water passes
through only half as many paraltel filter-cloth and
filter-cake surfaces and twice as thick a layer of
filter cake as does the final filtrate. This flow path
the inside surface of the basket sides, and is caught
in a shielding vessel, called a curb, within which the
basket rotates. The solid phase builds up a filter
ft/lb cake against the filter medium. When this cake is
thick enough to retard the filtration to an un-
economical rate or to endanger the balance of the
centrifuge, the machine is slowed, and the cake is
scraped into a bottom discharge or is scooped out
of the centrifuge. In an underdriven centrifuge, the
entire cake and filter medium may be removed and
a clean filter cloth inserted.
In the automatically discharging batch centri-
fugal filter, unloading occurs automatically while
the centrifuge is rotating, but the filtration cycle is
still a batch one. This machine is shown schematic-
ally in Figure 22.61. The constant rotation speed
of this machine permits lower power requirements
for a given amount of filtrate collected as well as
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
Leh haH- R,ghl ha1f-
flat bottom Self-d1schargmg
basket basket
Liquor
discharge
Bottom discharge
Figure 22 60. A composite cross-sectional view
of a suspended-basket centrifuge showing a flat-
bottomed and a cone-bottomed discharge on
the left and right sides, respectively. [ From
Chem. & Met. Eng., 50, No. 7, p. 119
(July 1943) by permission of the publisher,
copyright© 1943.)
lower labor requirements than would be obtained
in a batch centrifugal.
The slurry is fed to the unit through duct A as
the drum rotates, until the desired cake is built up
in the bowl. Slurry feed then is stopped, and wash
water is fed onto the cake through tube G. After
washing, the cake is spun dry. When this final spin
is completed, the peeler knife (H) is moved up into
the cake by a hydraulic mechanism peeling it off
into ·the chute (/). If desired, the filtrate and wash
liquid can be collected separately rather than corn-
bined in the housing (D).
A continuous centrifugai filter is shown in
Figure 22.62. In this filter, solids handling is held
to a minimum, which permits filtration of fragile
675
solids with the least possible breakage. As with
other centrifugal filters, this unit is best suited for
handling coarse-granular or coarse-crvstalline solids
in nonviscous liquids. Units are available with
capacities up to 25 tons/hr of solids when handling
this type of material.
In operation, the feed slurry is fed through
tube Oto the feed funnel (F) of Figure 22.62. The
funnel is attached to the pusher (0) and rotates at
the same speed as the centrifuge drum. In the
funnel, the feed is accelerated to drum speed and
fed into the back end of the filter drum where the
filtrate is forced through the filter screen (C), and
the solids form a cake on the screen. Inter-
mittently, the cake is pushed toward the discharge
end of the centrifuge by the pusher, which then
retreats again, leaving an open region for the build-
up of new cake. In this way, the cake moves across
the face of the filter screen until it is pushed off
the end into the solids-collector housing (G). As it
moves, the cake passes through a wash region
where wash liquid passes through it into the filtrate
housing (A). The filtrate and wash liquid are kept
separate by partitions in the wet collector.
Other continuous centrifugal filters are built,
such as the solid-bowl centrifugal pictured in
Figure 22.15. They would have a bowl wall consist-
ing of a screen through which the filtrate and wash
water would drain.
Centrifugal-Filtration Calculations. The mech-
anism of centrifugal filtration is identical to that
of pressure filtration, and the same equations
must apply to both cases if the terms within the
equation are properly evaluated. Now, however,
the driving force is the centrifugal force acting on
the fluid, rather than fluid pressure itself. Simple
substitution in the pressure-filtration equation
(Equation 22.121) of this centrifugal force seems
to be indicated. In engineering units,
Fcoc = ma = mrw2 (22.127)
where
a = acceleration
r = radius, ft
w = rate of revolution, radians/s = 2rrN x 60
N = rate of revolution, rpm
Fe = centrifugal force, lb,
m = mass, lb
9c = conversi.on factor, 32.17 ft 2/ s, -
lb
lb,
Thus, the term mrw2 lgc is the equivalent of -1:,P/L
when a centrifugal field rather than a pressure field
is present.
676 APPLICATIONS TO EQUIPMENT DESIGN

Figure 2261. Schematic diagram of an automatically discharging batch centrifugal

filter. (Courtesy Baker-Perkins, Inc.)
A-teed duct, 8-centrifuge OOwl, C-filter medium, D-liquor housing, £-filter
cake, F-filtrate, G-wash pipe, H-peeler knife, /-discharge chute, J-=hydraulic
cylinder

Careful analysis of centrifugal filtration shows
other differences from pressure filtration (22).
Primarily, the centrifugal-force field varies with the
depth in the filter cake, increasing as r increases.
The centrifugal force not only creates a hydraulic
pressure at the cake surface but also acts on the
filtrate as it flows through the cake and on the
solids in the cake, supplementing the hydraulic-
pressure head. Finally, in the typical basket filter,
the cake forms on the inside vertical walls of the

Figure 22.62. Scbernatic dieqr am of continuous :::entrifuga! tilter (Courtesy

Baker Perkins, Inc.)
A-wet housing, C-filter screen, D-cake pusher, £-wash-water nozzle,
F-feed funnel, G-cake discharge chute, H-filtrate discharge nozzle, /-:-wash·
water discharge nozzle, J-machine base, K-centrifuge drum, M-pusher
motor, N-piston rod, 0-feed pipe, P-wash-water pipe, R-access door
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 677
filter. The filter area decreases as the cake thickens, Comparing Equation 22.130 with Equation
and the kinetic energy of the flowing filtrate also 22.117, which was a similar material balance for
changes. These differences are illustrated in Figure the filtration through a flat cake, it is seen that the
22.63, which shows the physical situation and the filtrate volume held in a differential thickness of
nomenclature to be used in describing it. the filter cake has been neglected here. Finally, a
A differential pressure balance about a very filtrate balance gives
thin layer of filter cakes gives dV
- = Znhrv (22.131)
-dP = -dP9 - dPk - dPr (22.128) d8
where or, upon differentiation,
-dP = total effective pressure drop dv - (dV) _.:!! (22.132)

-dP9 = hydraulic pressure gradient developed -- d8 2r.hr2

by liquid in the centrifugal field flow- where
ing through the cake v = linear radial velocity of filtrate flowing
-dPk = hydraulic pressure gradient as a result
through the differential cake thickness dr
of kinetic-energy changes in passing

through the fluid The various terms of Equation 22.128 can

-dP1 = hydraulic pressure gradient as a result now be expanded in terms of the particular nomen-
of frictional drag clature and physical arrangement of centrifugal fil-
tration as diagrammed in Figure 22.63. The
For this differential cake, the area through which individual terms will then be recombined to give a
flow passes is general equation for centrifugal filtration. The
A= 2rrrh (22.129) integration of this equation will, of course, depend

upon the properties of the particular material and

A solids-material balance for this differential cake
the geometry of the particular equipment involved.
thickness results in
the integration will be performed for the simple

wdV= p,(1 - Ep)2rrhrdr (22.130) case of centrifugation in a cylindrical basket of a

slurry that forms a noncompressible cake.
where

The pressure gradient resulting from kinetic-

w = weight of solids per unit volume of energy variations must be
original slurry
dV = increase in filtrate volume in a dif- dPk vdv
-=--
ferential time span d8
as shown in Chapter 20. Combining this with equa-
tions 22.131 and 22.132 gives
2
-p (dV\ dr
(22.133)
dPk = gc d8-} (2rrh)2,.3
where
p = density of the filtrate

The pressure gradie�t resulting from hydraulic

head would be
dP 9 g
-=-dz
p gc
if the head occurs in a gravitational field as shown
by Equation 20.5. In the centrifugal field existing
here, g must be replaced by rw2, and dz must be
replaced by dr. As a result,
pw2rdr
dP = -- (22.134)
9 gc
Figure 22.63. Nomenclature and physical
arrangement, centrifugal filtration. for the centrifunal filtration.
(;78 APPLICATIONS TO EQUIPMENT DESIGN
Finally, the pressure drop resulting from form The pressure at the cake surface (P2) is
drag and skin friction can be found by appropriate pw 2(r/-r 12)
alteration of Equation 22.118. For pressure filtra- P2-P1 (22.139)
tion,
where

-ar, = c,µw2 (dV) dV (22.135) r1 = radius to surface of liquid over the filter
9cA de
cake
Substituting Equations 22.129 and 22.130 into
Generally Equation 22.137 can be integrated
Equation 22.135 gives
only if the relations among °'P, Ep, and r can
-dP,= p,(1 - Ep). exµ (dV)dr (22.136) be determined. For a few simplified cases, the
2rrhgc de r integra- tion can be done directly (22). The most
useful of the integrations is for a cake that can be
Replacing Equations 22.133, 22.134, and 22.135 considered incompressible. Then, putting
into Equation 22.128 and integrating results in Equations 22.138 and 22.139 in Equation 22.137
9c 2 2 and integrating
P, -pw ,, p (dV) Jr,<!!
with P4 = P1 = 0 gives
dP--- rdr+--- -
- IP, - J,, (2rrh)2gc de r, ,3
dV) µRM' _ pw2(r/ - r,2)

p,-
µ (dV)f'' OCp(1-Ep)-
dr (
+- - de 2rrr3hgc 2gc
2rrhgc de ,, r
. (22.137)
where
P2, P3 = pressures at the filter-cake surface
and at the filter-cloth surface,
respectively
r2, r s = radii to the filter-cake surface and
to the filter-cloth surface, respect- which rearranges to
ively
p, p, = density of filtrate and of solids, 2 2 p(dV/de)2 ( .L _ !_)
respectively dV 1thpw (ri- r, ) 4rrh rl r,2
Ep = porosity at any point in- the bed
ac,, = specific cake resistance at any
point in the bed de
h = height of cylindrical surface on (22.141)
which the cake is being built
(dV/dO) = filtration rate The second term in the numerator is that coming
w = rate of rotation, radians/sec from the changes in fluid kinetic energy. It is
almost always very small. Removing it and writing
This equation is the basic rate equation for centri- w = rrN/30 gives
fugal filtration, but, as with Equation 22.118, it
only gives the pressure drop across the bed itself. dV
the pressure at the tilter-cloth surface (P ) can be
3 de r3 RM']
related to the normally measured downstream pres- (30)2µ ac_o,(1 - c) In-+ -
[
sure (P4) in terms of a filter-medium resistance as r2 rs

was done for pressure filtration. (22.142)

(dV) Equation 22.142 can be compared with the

\de µRM' standard pressure-filtration equation, Equation
P3-P4= (22.138) 22.120 by considering the total hydraulic driving
2rrr3hgc
where force 'to replace -t!.P, in the pressure-filtration
P4 pressure at low-pressure side of tilter
= Equation. By incegration of Equation 22.134 for
cloth the filter cake and the liquid above it,
RM'= resistance of thefilter medium, analo-
gous to the medium resistance given in pw2 r 2 - r 2
AP= 3 1 (22.143)
Equation 22.120 -
9
9c 2
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS
Also, by a mass balance on the solids in the cake,
wV
p,(1-E)=h(2
tt r3 - '2
Combining Equations 22.143 and 22.144 with
Equation 22.142 gives
dV
d8
in which the area terms may be designated
dV
d8
Thus, for centrifugal filtration, the A 2 term in
Equation 22.120 becomes Am • A1m, as defined
by Equation 22.145, but Ac (the area of the filter
cloth) is correct as a replacement for A when
dealing with the resistance of the filter medium.
Example 22.16. The CaC03 slurry described
in Examples 22.13 and 22.14 is to be filtered in a
basket centrifugal filter of 24-in. inside basket
diameter and 10-in. basket height rotating at
1200 rpm. Assuming that the cake is incompres-
sible and that the filter medium resistance is
negligible, what filtration rate can be expected
when the cake is 1 in. thick and the liquid surface
corresponds to the filter-cake surface?
Solution. If the cake is incompressible, the
values of a and E calculated in Example 22.13 can
be used here. The rate (dV/d8) can then be calcu-
lated directly from either Equation 22.141,
22.142, or 22.145. Equation 22.142 wili be used
since it is more immediately applicable. Thus,
dV
,,
3
12l
x (1200)2 x 62.4 x 0[ 12-- (111
2)
d8
900 x 1.0 x G.000672[ 2.41 x 1011
x 183(1 - 0.453) In 12]
dV
- = 0.000300 ft3/s or 0.135 gal/min
d8 . tions. The economic balance can seldom be applied
Cyclic Operation of Batch Filters. The oper-
ation of a batch filter necessarily consists of the
679
steps of filtering, washing and dewatering, and
cleaning and reassembly. The free volume within
2) (22.144) the filter available for cake storage poses the
ultimate limit on filtering time during a single
cycle. A more practical restriction is often found in
the desire of the filter operator to get
the maximum amount of material filtered in an
operat- ing day. This will occur when the curve of
filtrate volume per day as a function of filtering time
per cycle reaches a maximum. In other words,
(22.147)
where
(22.146) Ne= number of complete operating cycles
per day
V = volume of filtrate per cycle
e,= time of filtering in each cycle, hr
The number of cycles per day is ·
24 24
Nc=-=----- (22.148)
8c e, + 8w + ed
where
Be =time for a complete cycle, hr
8w =time for washing and dewatering, hr
ed =time for disassembly, cleaning, and reas-
sembly, hr
The re!ations developed earlier allow e, to be
expressed as a function of V whether pressure or -
centrifugal filtration is considered, and with
compressible or incompressible cakes. In many
cases, the wash time (8w I can also be related to the
filtrate volume or filtering time. This, however, is
at the discretion of the operator. The disassembly,
cleaning, and reassembly time (8d I depends upon
the type of filter, the properties of the material
being filtered, and the labor available. It will
2
usually be a constant, indeper.dent of V, for any
] given installation.
Thus, Equation 22.148 can be reduced to a
function relating, Ne to either 81 or V. Formal
differentiation will then give either d(NV)/dO, or
d(NV)/dV, which can be set equal to zero to get
11 the desired optimum.
This solution is an example of the formal use
of an economic balance to control plant opera-
with such mathematical precision, but it must
always form the yardstick by which process alter-
nates are chosen and operations are guided.
680 APPLICATIONS TO EQUIPMENT DESIGN
PARTICLE SEPARATIONS NOT INVOLVING FLUID
MECHANICS
Screening is an operation for separating solid
materials on the basis of size alone. No fluid
mechanics are involved in this process. A mixture
of solid particles of various sizes is fed onto a
surface provided with suitable openings. Some
material passes through the openings, and some
remains behind; the resulting portions are now
more uniform in size than the original.
The screening equipment may be in the form
of stationary or moving bars, perforated steel plate,
or woven-wire cloth. The screens may be inclined at
an angle so the solids can flow downward by
gravity flow, or the screens may be set in some
type of vibratory motion.
The effectiveness and analysis of the screening
operation depend upon a knowledge of particulate
solids, as discussed in Appendix B.
Electrical Precipitation
Gases containing very small particles can be
cleaned electrically by passing the gas between two
high-voltage electrodes that ionize the gas. The gas
ions attach themselves to the solid particles and
build up a charge on the particle. The charged
particle migrates according to an EMF gradient and
is attracted to collecting electrodes-the negatively
charged particles to the positive electrode and the
positively charged particles to the negative
electrode.
There are two general classes of electrical pre-
cipitators:
1. Single-stage, in which ionization and
collection are combined.
2. Two-stage, in which ionization is followed
by collection, with each operation occurring in a
different part of the apparatus.
Most industrial gases are readily ionized.
Electrical precipitators are operated at the highest
practical voltage, governed by arcing, since this
voltage produces the most rapid migration of
particles. Voltages up to 100,000 are used. Collec-
tion efficiencies are very good. With low gas
velocities, nearly 100 percent efficiency, on a mass
basis, can be obtained, but the economic limit
is about 99 percent.
Mn::"<>ti� Seoaration
Materials that have different magnetic ·attract-
ability may be separated by passing them through a
magnetic field. Most often magnetic separation
I
techniques are used to remove iron, steel, or
magnetic iron oxide from materials low in
magnetic attractability. However, the methods
have been refined so that they can be used to
separate materials that are only slightly reactive to
a magnetic field. One common device is simply an
electromagnet suspended over a conveyor belt.
This might be used to remove tramp iron from the
feed to a crusher. Periodically the magnet is
unloaded into a bin. Another device employs a
magnetized pulley at the end of a conveyor belt.
As the material is conveyed over this pulley,
Magnetically inert material drops off the belt in a
normal manner. Magnetic material is held on the
belt, however, and finally drops off the belt as it
leaves the field of the pulley. Other magnetic separ-
ators can be used to remove magnetic material
from a slurry. As resource recovery from sol id
wastes grows in importance, magnetic separation
will be the major method used to remove iron from
'the waste stream.
REFERENCES
1. Ambler, C. M., Chem. Eng. Progr., 48, p. 150 (19521.
2. Coal Min. and Proc.,8, No. 2, p, 37 (1971).
3. Aude, T. C., et al., Chem. Eng., 78, No. 15, p. 74
(1971).
4. Bhattacharya, D., and D. C. Pei,- Ind. Eng. Chem.
Fund., 73, p. 199 (1974).
5. Bhattacharya, A., and A. N. Roy, Ind. Eng. Chem., 47,
p. 268 (1955)..
6. Brown, G. G., Unit Operations, John Wiley & Sons,
New York (1950), Chapter 18,
7. Brown, G. G., Unit Operations, John Wiley & Sons,
New York (1950), p, 119.
8. Brownell, L. E., and H. E. Crosier, Chem. Eng., 56, No.
10, p. 124 (1949).
9. Brownell, L. E., and G. B. Gudz, Chem. Eng., 56, No.
9,p.112 (1949).
10. Brownell, L. E., and D. L. Katz, Chem. Eng. Progr.,43,
pp. 537. 703 (1947).
11. Burke, S. P., and W. B. Plummer, Ind. Eng. Chem., 20,
p. 1196 ( 1928).
12. Carleton, A. J., and D. C-H. Cheng, Chem. Eng. 84,
No.9, p. 95 (1957).
13. Carman. P. C .. J. Soc. Chem. Jnd., London, 57, 225T
(1938).
14. Coe, H. S., and G. H. Clevenqer, Tran, A.I.M.M.E.. 55
p. 35G (19161.
15. Comings. E.W., Ind. Eng. Chem., 32, p. 663 (1940).
16. Durand, R., "Basic Relationships of the Transportation
of Solids in Pipes - Experimental Research," Proc.
Minn. Intl. Hydraulics Conv., p. 89-103 (1953).
17. Ergun, S., Chem. Eng. Progr., 48. p. 89 (1952).
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 681
18. Ergun, S., and A. A. Orning, Ind. Eng. Chem., 41, 51. Shepherd, C. B., and C. E. Lapple, Ind. Eng. Chem.,
p. 1179 (1949). 31, p. 972 (1939).
19. First, M. W., ASME paper 49-A-127 (1949). 52. Steinour, H. H., Ind. Eng._ Chem., 36, pp. 618, 840,
20. Flood, J.E., Chem. Eng., No. 6, p. 217 (1955). 901 (1944).
21. Gasterstadt, H., VD/ Forschungsarbeiten, No. 265 53. Stermerding, S., Chem. Eng. Sci., 17, p. 599 (1962)
(1924). 54. Stokes, G. G., Math. and Phys. Papers (1901); Trans.
22. Grace, H.P., Chem. Eng. Progr., 49, pp. 303, 367, 427 Cambridge Phil. Soc.,9, Part II, p. 51 (1851).
(1953). 55. Talmadge, W. L., and E. B. Fitch, Ind. Eng. Chem., 47,
23. Grace, H.P., A.I.Ch.£. Journal, 2, p, 307 (1956). p. 38 (1955).
24. Grutsch, J. F., and R. C. Mallat, CEP, 73, No. 4, p. 57 56. Ter linden, A. J., Proc. Inst. Mech. Engrs., London,
(1977). 760, p. 233 (1949).
25. Gunness, R. C., Chem. Eng. Progr., 49, p. 113 (1953). 57. Turian, R. M., and T. F. Yuan, A.I.Ch.£. Journal, 23,
26. Hull, W. 0., H. Keel, J. Kenney, Jr., B. W. Gamson, p. 232 (1977).
Ind. Eng. Chem., 45, p, 256 (1953). 58. Turian, R. M., T. F. Yuan, and G. Mauri, A.I.Ch.£.
27. Knowlton, T. M., and 0. M. Bachovchin, Proc. Intl. Journal, 17, p. 809 (1971).
Conf. on F/uidization, Asilomar Eng. Fndn. {June 59. Vogt, E.G., and R.R. White, Ind. Eng. Chem., 40,
1975). p. 1731 (1948).
28. Kozeny, J., Sitzber. Akad. Wiss. Wien, Math-naturvv. 60. Waddel, H., J. Franklin Inst., 217, p. 459 (1934).
Kl. Abt. Ila, 136, pp. 271-306 (1927). 61. Wilhelm, H. H., and M. Kwauk, Chem. Eng. Progr., 44,
29. Kunii, 0., and 0. Levenspiel, Fluidization Engineering, p, 201 (1948).
John Wiley & Sons, New York (1969). 62. Yang, W. C., A. I.Ch.£. Journal, 21, p. 1013 ( 1975).
30. Kynch, G. J., Trans. Faraday Soc., 48, p. 166 (1952). 63. Yoon, S. M., and 0. Kunii, Ind. Eng. Chem. Proc. Des.
31. Lapidus, L., and J.C. Elgin, A.I.Ch.£. Journal, 3, p. 63 & Dev., 9, p. 559 (1970)."
(1957). 64. Zenz, F. A., and N. A. Weil, A.I.Ch.£. Journal, 4,
32. Lapple, C. E., in Perry (ed.). Chemical Engineer's p.472 (1958).
Handbook, 5th ed., McGraw-Hill, New York (1973),
p, 20· 79-20-83.
33. Lapple, C. E., and C. B. Shepherd, Ind. Eng. Chem., PROBLEMS
32, p. 605 (1940).
34. Lash, L., Chem. Eng. Prag., No. 4, p. 67 (1977).
35. Leva, M., F/uidization, McGraw-Hill, New York 22.1. Prepare a plot showing the terminal-velocity-
(1959). particle-diameter relationship for a spherical particle having
36. Leva. M., and M. Grummer, Chem. Eng. Progr., 43, a specific gravity of 2.0 settling in water at 30°C. Use
pp. 549, 633, 713 (1947). several diameters between 1 micron and 1000 microns.
37. Leva, M., M. Grummer, M. Weintraub, M. Pollchick, 22.2. Repeat Problem 22.1 for the particle settling in
Chem. Eng. Progr., 44, pp. 511, 619 (1948). air at 70°F.
38. Leung, L. S., and R. J. Wiles, Ind. Eng. Chem. Proc. 22.3. Calculate the terminal radial velocity of a
Des. & Dev., 75, p. 552 (1976). soluble coffee particle 60 microns in diameter in air at
39. Leung, L. S., R. J. Wiles, D. J. Nicklin, Tr. Inst. Chem.
250°C entering a cyclone 0.5 m in diameter. The tangential
Eng., T271 (1969).
velocity of the coffee particle is 200 cm/sec, and the
40. Matheson, G. L., W. A. Herbst, P. H. Holt, Ind. Eng.
specific qr avitv of the solid is 1.05.
Chern.,41, p. 1099 (,949).
41. Matsen, J. M., Powder Technal., 7, p. 93 (1973). 22.4. Square mica plates, fr- in. thick and 0.01 in.2 in
42. Mertes, T. S., and H. 8. Rhodes. Chem. Eng. Progr.,51, area are falling randomly through oil with density of
pp. 429, 517 (1955). 55 1b/tt3 and with vecosrtv of 15 centiooises. The specific
43. Morcom, A. R., Treos. tnst. Chern. Engrs. (London}, gravity of the mica is 3.0. What will be the settling velocity?
24, p. 30 (1946). Suppose the area of the mica plate is 1.0 in.2, what would
44. Morse, R. D.,lnd. Eng. Chem.,41,p.1117 (1949). be its settling velocity? ·......
45. Oman, A. 0., and K. M. Watson, National Petrol. News, 22.5. Glass beads 50 micron"?\ln diameter settle in
36, p. R795 (1944). water at 25°C. Specific gravity of qlass'is 2.6.
46. Perry, R.H., and C. H. Chilton (eds.), Chemical (a) rr unhindered settling occurs, what is the
Engineers Handbook, 5th ed., McGraw-Hill, New York maximum velocity attained?
(1973). (bl ft settling occurs such that the mass ratio of
47. ·Roberts, E. J., Trans. A.I.MM.£., 184, p. 61 (1949). water to glass is 2,what is the maximum velocitv?
48. Romero, J. B., and L. N. Johanson, CEP Symp. Series,
22.6. A laboratory viscosimeter consists of a steel
58, No. 38, p. 28 (1962). ,
ball and uniform-diameter glass cylinder. The cylinder is
49. Ruth, B. F., and L. L. Kempe, Trans. A.I.Ch.£., 33,
pp. 34-83 (1937). filled with the test fluid, and the time for the ball to fall a
50. Sittig, M., Petrol. Refiner, 31, No. 9, p. 91 (1952). known distance is recorded. The ball is 0.25 in. in diameter,
682 APPLICATIONS TO EQUIPMENT DESIGN
and the index marks are 8 in. apart. The viscosity of corn
syrup having a density of 1.3 g/cm3 is desired. The
measured time interval is 7.32 sec. What is the viscosity of
the syrup? Specific gravity of the steel ball is 7 .9.
22.7. A mixture of galena and silica is to be
elutriated by a water stream flowing at a velocity of 0.1 m/s
and a temperature of 25°C. The solid feed analyzes 30
weight percent galena, and a screen analysis is indicated in
the following table.
Particle diameter,
microns 20 30 40 50 60 70 80 90 100
Weight percentage
of undersize 33 53 67 77 83 88 91 93 94.5
If the size distribution given applies for both components in
the feed, what fraction of the galena fed is in the overhead
and bottom products, and what is the weight fraction (dry
basis) of galena in these products?
Galena specific gravity= 7.5
Silica specific gravity = 2.65
22.8. Solve Problem 22.7 if carbon tetrachloride is
used as the elutriating agent.
22.9. A gravity settling tank is to be used to clean
wastewater from an oil refinery. The waste stream contains
1 percent oil by volume (specific gravity of the oil is 0.87)
as small drops ranging in size between 10 and 500 microns.
The tank is rectangular and measures 3 m wide and 2 m
deep. Provision is made at the discharge end for the clean
water to be continuously removed from the bottom of the
tank. Periodic skimming of the liquid surface at the dis·
charge end removes the accumulated oil. If 6.3 m3/s-of
wastewater is to be processed, how long must the settling
tank be?
22.10. A mixture of coal and sand in particle sizes
smaller than 20 mesh is to be completely separated by
screening and then elutriating each of the cuts from the
screeninq operation with water as the elutriating fluid.
Recommend ct screen size such that the oversize cut can be
completely separated into coal and sand fractions by water
elutriation. What water velocity will be required? The
specific gravity for sand and coal is 2.65 and 1.35,
respectively.
22.11. Quartz arid pyrites (FeS2l are separated by
continuous hydraulic classifications. The feed to the
classifier ranges in size between 10 microns and 300
microns. Three fractions are: obtained: a pure quartz
product, a pure ovrites product, and a mixture cf quartz
and pyrites. The specific qrevttv of quartz is 2.65, and that
of pvrtt=s is � 1. What is the size ranae of the two materials
in the mixed fraction for each of the following cases:
(a) The bottoms product to contain the maximum
amount of pure pyrites.
(b) The overhead product to contain the maximum
amount of quartz.
22. 12. The mixed feed of Problem 22.11 is to be
separated into two pure fractions of pyrites and quartz by
hindered-settling process. What is the minimum density of
the heavy medillm that will give this separation?
22.13. A tubular-bowl centrifuge is to be used to
separate water from a fish oil. This centrifuge has a bowl
4 in. in diameter by 30 in. long and rotates at 15,000 rpm.
The fish oil has a density of 0.94 g/cm3 and a viscosity of
50 centipoises at 25°C. The radii of the inner and outer
overflow dams are 1.246 in. and 1.250 in., respectively.
Determine the critical diameter of droplets of oil suspended
in water and of droplets of water suspended in oil if the
feed rate is 300 gal/hr of a suspension containing 20 weight
percent fish oil.
2214. Determine the L value for the centrifuge of
Problem 22.13 for oil removal from the water. Note that
with a liquid-liquid separation, L is no longer independent
of the system used.
22.15. A cylindrical-bowl internal-screw centrifugal
is to be used to separate MgS04 ·6H20 crystals from the
mother liquor which comes as product from a vacuum
crystallizer. The centrifugal has a bowl 14 in. in diameter
and 23 in. long and carries liquid to a depth of 3 in. It
rotates at 3000 rpm. If no crystals smaller than 5 microns
in diameter exist in the slurry, what feed rate to the
centrifugal will result in complete removal of the solids?
Assume that the internal screw does not lift any of the solid
particles back into the liquid or disturb their fall through
the liquid. The solution density is 1.21 g/cm3, its viscosity
is 1.5 centipoises. The crystals have a density of
1.66 g/c3.
22.16. A conical solid-bowl centrifugal is to be used
to dewater the liquid product from a classifier in closed
circuit with a fine grinder in the preparation of cement rock
before feeding the cement kiln. The centrifugal is to remove
particles of a size greater than 10 microns in diameter. It
has a 40-in. maximum diameter, a 60-ir.. length, and a 10
deqree ar.gle or. the cone wa!I. Maxir.ium liquid depth is
8 in. Rotation is possible, up to a maximum of 1200 rpm.
Oetarmine the revcrutlons per minute to be used if the feed
rate to the centrifugal is 300 gal/min.
22.17. On the basis of the ability to separate minute
quantities of cottonseed oil at 30°C from a water phase
down to droplet diameters of 2 microns, what would be the
permissible throughput rate vvhen using a No. 2 disk centri-
fuge with specifications equal to those of the second unit of
this type listed in Table 22.1. Could this centrifuge be
operated to remove both oii end water droplets greater then
2 microns in diameter from the-other phase when feeding a
50 volume percent oil and water mix?
22.18. Determine outlet-dam heights and throughput
that would permit the tubular-bowl centrifuge of Problem
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 683
22.13 to separate droplets down to a critical diameter of 1 underflow concentration of 10,000 mg/I and a feed concen-
micron from both oil and water phases. !ration of 2500 mg/I.
22.19. A 3 percent by weight calcium carbonate
Time, min 0 3 5 10 15 20 30
slurry was subjected to a batch-sedimentation test. The
Height of interface,
density of the solids in the slurry was 2.63 g/ml, and that of
mm 1000 690 480 300 220 180 130
the liquid was 1.0 g/ml. The test results are indicated
below. 22.25. An adsorber bed of "molecular sieves" con·

Time, min 0 20 40 60 80 100 120 140 sis ts of randomly packed extruded cylinders � in. in
Height of liquid-solid diameter by 1\ in. long held on a sintered metal plate.
interface, rn 176 100 74 57 42 34 26 22 Oxygen at -200°F, 100 psia is to be passed through the
bed at a superficial velocity of 1 h/s in order to remove
Determine the area and depth of a continuous thickener
impurities such as light hydrocarbons and inert gases. What
that is to handle 400,000 kg of solids per day. Initial solids
pressure drop can be expected through a bed 10 ft long?
concentration in the feed is 3 percent by weight and the
The viscosity of gaseous oxygen under these conditions is
final concentration is to be 40 percent solids.
0.0125 centipoise.
22.20. Wastewater from a de-inking plant is to be 22.26. The adsorbent bed of Problem 22.25 is
clarified by continuous sedimentation. Feed to the supported by a sintered metal plate ! in. thick having
thickener is one million gal/day containing 1.20 percent by passages averaging 50 microns in diameter. Estimate the
weight solids. The underflow from the unit analyzes pressure drop through this plate if its porosity is 0.30.
8 percent solids. Specify the depth and diameter of 22.27. A sand filter consists of uniform spherical
the thickener. particles of -20 + 28 mesh size. After back flushing with
A single batch-settling test on the feed material gave water, the sand bed has settled to a stable depth of 6 ft and
the following information: is flooded with water to a depth of 3 ft above the top of
Specific gravity of solids;, 2.00 the sand. If the drain valves are opened, how long will it
Specific gravity of solution= 1.00 take the bed to drain until the water level is even with the
Concentration of solids in test= 0.12 percent top of the bed?
22.28. Boiler feed water is deionized by pumping it
Time, min O 5 10 20 40 60 180 240
downward through a bed of ion-exchange resin. The resin
Height of liquid-solid
has spherical particles sized to -8 + 10 mesh. The deionizer
interface, cm 31 21 10 3.2 2.2 2.1 2.0 1.96 1.94
is 4 ft in diameter by 6 ft long, and an operating flow rate
22.21. 40 tons/hr of a metallurgical pulp are to be of 20 gal/min/tt3 of bed is used. What pressure drop exists
thickened from 186 g/1 to 1200 g/1 by sedimentation. through the bed?
Batch-settling data on the feed material gave the following 22.29. The ion-exchange resin of Problem 22.28 has
information: a particle density of 80 lb/ft3. It is regenerated by back·
Time, hr O 0.10 0.25 0.50 0.75 1.0 2.00 4.0 flushing with 10 percent brine solution {p =. 1.07 g/cm 3) at
Pulp height, ft 3.0 2.0 1.4 0.9 0.68 0.48 0.25 0.1 a rate such that the bed is expanded 100 percent. What rate
of brine flow is required?
Specific gravity of solid= 4.44
22.30. In the fluidized roasting of zinc sulfide, flota-
Concentration = 186 ,!iii. tion concentrate of the sulfide in particle sizes averaging 50
What depth and diameter thickener is rec:;uired to accornp- microns is roasted at 16C0°F ind about 4 psig. Pubiished
lish the specified assignment? information [Andersen, I. T., and R. Balduc, Chem. Eng.
22.22. Determine the area needed for a thickener to Progr., 49, 527 (1953)] qives a 1-h/sec velocity through the
produce an underflow concentration of 15,000 mg/I from a bed and a fluidized bed density cf 90 lb/ft3. Using a solid
feed containing 2000 mg/I of suspended solids. Flow to the density of 4.00 9/cm3 for the concentrate and assuming the
thickener is to be 1 x 106 gal/day with a sludpe flow of particles are spherical, locate this data point on the plot
showing pressure drop through fixed beds (Figure 22.30).
0.4 x 106 gal/day. Settling tests were performed using a 5-ft
22.31. Estimate the gas flow necessarv to begin
cylinder.
f\uidization in the roasters of Problem 22.27 ar,d the gas
Time, min O 5 15 30 45 60
flow necessary to cause conveying of the bed.
Height of interface, ft 5 3.9 1.9 1.1 0.95 0.8
22.32. In the manufacture of synthetic detergents,
22.23 Repeat Problem 22.22. for underflow rates of the spray dried bead is pneumatically conveyed from the
0.6 x 106 and 0.8 x 106 gal/day. bottom of the spray drier to cyclone separators at the top
22.24. Settling tests were obtained for an activated of the manufacturing building. In a plant producing
sludge. Determine the unit area and overflow rate for an 3000 lb/hr of product the conveyor is a 10-in. 1.0. tube
684 APPLICATIONS TO EQUIPMENT DESIGN
running vertically for 60 ft and horizontally for 30 ft. The
detergent has a mean particle size of 600 microns and a
packaged bulk density of 20 lb/tt3. The blower pulling air
through this conveyor has a rating of 500 std. tt3/min.
What is the pressure at the blower inlet if atmospheric
pressure exists at the bottom of the spray drier? Ignore
pressure drop through the cyclone and assume the conveyor
operates at a constant 80° F.
22.33. Superphosphate fertilizer is to be conveyed to
an open storage bin by a pneumatic conveying system. The
production rate is 10 tons/hr of product with an average
particle ,size of 200 microns and a particle density of
150 lb/ft3. The conveyor is to discharge at a point 70 ft
above the pickup location and 500 ft away from it. Find
tube diameter, blower capacity, and tube pressure drop
such that the product will be satisfactorily conveyed when
the blower is located at the feed end of the system.
22.34. Coal is pumped in water suspension from
Cadiz, Ohio, to Cleveland, a distance of 108 miles in a
10-in. I.D. welded pipeline. A 45 weight percent slurry of
the t-in. coal particles is pumped at 3! miles/hr through the
pipeline by high-pressure, positive-displacement pumps
located at three pumping stations along the pipeline.
Estimate the pump pressure developed and the horsepower
required by each pumping station.
22.35. Coffeebeans are moved from the wharf to
roasters by pneumatic conveying through an 0.25 m duct.
The duct must convey the beans horizontally 100 m and
20 m vertically at a rate of 2000 lb/hr. Specify a blower
(capacity and discharge pressure) that can be used if it is
located at the feed end of this system. State any assump-
tions needed.
22.36. The following data were collected in the
laboratory on the filtration of a CaC03 sludge through a
washing plate-and-frame filter press. After assembling the
press, the data were taken by noting the time required for
the collection of succeeding 5-lb batches of filtrate.
Filtrate collected, lb O 5 10 15 20 25 30
Elapsed-time interval,
min O 0. 70 0.52 0.65 0. 75 1.00 1.08
Temperature 20°C
Area of filter surface 1.87 ft2
Feed concentration, 5 percent CaC03 in water
Pressure at filter inlet, 6 psi9
Determine u: and Ve for the filtration of thiS slurry in this
filter press.
22.37. A rotarv-disk filter is ta be used to collect the
solids obtained as flotation-cell product in the faed-
preparation system of a cement plant. The filter rotates at
20 rph with half its surface submerged and an internal
pressure of 35 kN/m2 . Laboratory experiments run on the
same slurry at a constant-pressure differential of 70 kN/m2
using a plate-and-frame press with 0.35 m2 effective surface
gave the following data:
Time, hr 0. 15 0.30 0.45 0.60
Filtrate volume, m3 3.53 5.74 7.78 9.44
What area is required for the disk filter if 85,000 kg/day of
solids are to be filtered from a slurry containing 10 weight
percent solids?
22.38. A wet-process cement plant grinds cement
rock to -200 mesh, the final grinding being in a water
carrier. The resulting slurry is thickened and then filtered
with the filter cake being sent to the kilns. The thickened
slurry is 30 weight percent solids, with the solids being of
40-rnicron average diameter. It is planned to filter this
slurry through rotary-disk vacuum filters. These filters have
disks 4 ft in diameter which are submerged in the slurry up
to their hubs and rotate at 1 rpm. The pressure inside the
disks is 3-psia, and it is estimated that there will be a 1-psia
pressure drop through the filter cloth on the disk face.
Laboratory tests with this slurry show that the density of
the dried cake is 115 tb/tt3 at any filtration pressure below
40 psig. The solid density of the cement rock is 180 lb/tt3.
If the filter plant must deliver 50 tons/day of dry solids as
filter cake, how many filter disks are required?
22.39. It is desired to increase the capacity of
a rotary-vacuum drum filter. Give, as accurately as possible,
the percent capacity (weight of product per unit time)
increase that results from:
(al Doubling the rate of rotation.
(b) Doubling the submergence, where submergence is
the distance from the slurry pond surface to the bottom of
the drum.
(cl Doubling the concentration of solids in the feed.
(d) Doubling the pressure drop.
State qualitatively what effect each of these changes might
have on the product quality and on the cost of filtration.
22.40. An inofganic dye is to be filtered in a plate-
and-frame filter press having 100 ft2 effective area. The 3
percent solution will be pumped directly to the filter by a
centrifugal pump with the follow:ng head-capacity
characteristics:
Pressure developed, psi 15 20 25 30 35
Flow rate, gal/min 4946433827 0
Laboratory constant-pressure filtration tests with the same
slurry at (-.1P) = 20 psi give straight lines when
2
A (.18/..6.V) is planed against V. The siopes of the lines are
O.C5 min ft4 gal3, whereas the intercepts at V = 0 are
0.1 min ft4 /gal where the filter medium and piping are as
similar to the- plant units as can be obtained. Tests at
varying (-.1P) values indicate that the filter cake is
incompressible. If the plant runs are to be stopped when
the pump pressure reaches 30 psig, how long will the batch
filtrations take?