Particle size distributions

Particle size distributions expressed in terms of frequency, f(x), and cumulative distribution,
F(x), distribution curves. The cumulative distribution curve is the integral of the frequency
distribution curve.
These distributions can depict various particle attributes including numbers fN(x), FN(x);
surfaces fs(x), Fs(x) etc.
0.035
0.030

(m )
-1
0.025 The largest group has a
f(x) or dF/dx 
0.020 particle size of 45 microns
Frequency distribution
0.015
0.010
0.005
0.000
0 20 40 60 80 100
particle size, x(m)
1.2
1.0
0.8
0.6 50% of the particles has a
F

Cumulative distribution particle size of 45 microns

0.4
or less
0.2
0.0
0 20 40 60 80 100
particle size, x(m) 1
 Note: for a given particle population the various distributions can drastically differ:

I.
fN(x) (by number)
f(x)

fs(x) (by surface)

fV(x) (by volume)

Particle size x

II. Mass and number distributions for satellites orbiting the Earth

Size (cm) Number of objects %by number %by mass

10 -1000 7000 0.2 99.96
1 – 10 17500 0.5 0.03
0.1 – 1.0 3500000 99.3 0.01
Total 3524500 100 100
2
Converting between distributions:

Relating size distributions by number fN(x) and by surface fs(x) for particle populations with
the same geometric shape and different size

fraction of particles in the size range x to x + dx = fN(x)dx

fraction of the total surface of particles in the size range x to x + dx = fs(x)dx

number of particles in x to x + dx = NfN(x)dx where N = total number of particles

surface area particles in x to x + dx = (x2as)NfN(x)dx where as relates linear dimension x to its

surface area
( x 2 a s )N fN ( x )dx
fraction of total surface area in x to x + dx = S = total particle surface area
S
as N
x2 x2f
considering now that fs(x) is proportional to fN(x) or fs(x) = ks N(x) so ks =
S
avN
3 k
Similarly for distribution by volume: fv(x) = kv x fN(x) with v 
V
V = total particle volume, aV = factor relating linear dimension to its volume

and if the particle density is independent of particle size then for distribution by mass obtain:
a v rp N
fm(x) = km x3 fN(x) where km  and rp = particle density
V 3
Describing the population by a single number
Many different single numbers (means) can be defined for a given size distribution.
These means can all be described by the equation:

 g( x )f ( x )dx  

g( x ) 0
where here  f ( x )dx1 so g( x )  g( x )f ( x )dx
 0 0
 f ( x )dx
0

where x is the mean and g is the weighting function which is different for each mean definition
g(x) mean and notation
x arithmetic mean
x2 quadratic mean
x3 cubic mean
4
Some examples of mean diameters:


number length mean diameter xNL   x fN ( x )dx
0
i. sum of all particle diameters divided by the total number of particles

number surface mean diameter xNS  [  x 2 fN ( x )dx]0.5
0
i. average surface area of a particle given by the sum of all particle surface areas divided
by the number of particles
ii. the diameter of such a particle


number volume mean diameter xNV  [  x 3 fN ( x )dx]0.333
0

i. average volume of a particle given by the sum of all particle volumes divided by the
number of particles
ii. the diameter of such a particle

5
and in discrete form which is frequently more accessible and amenable to calculation

xNL  ( 
Ni xi
number length mean diameter )
 Ni

( i i
2
Nx
number surface mean diameter xNS )0.5
 Ni

( i i
3
Nx
number volume mean diameter xNV )0.333
 Ni

6
Finding appropriate expressions for particle distributions

Gaussian distribution

Just understand this equation, do not need to remember

dF 1 ( x  x )2
 f (x)  exp[  ]
dx (2)0.5 22

where  is the standard deviation

mean
0.035 value
0.030
(m )
-1

0.025
f(x) or dF/dx 

0.020
0.015
0.010
0.005
0.000
0 20 40 60 80 100
particle size, x(m)

7
 Log-normal distribution
This distribution is skewed to one side but can be transformed to Gaussian form by plotting as
dF( x )
understand this equation, do not need to remember vs log (x)
d(log x )
dF 1 ( z  z)2
 f (x)  exp[  ] where z = log x and
dx  z ( 2 )0.5
2 2 z is the standard deviation of z
z

f(x) dF( x )
d(log x )

particle size, x (m) log(x)

Rossin-Rammler distribution Just understand this equation, do not need to remember

Describes materials that have been ground

f ( x )nbxn1 exp( bxn ) where n is a characteristic of the material and b is a measure of the
range of particle size
and the cumulative distribution is obtained by  f(x)dx  F(x)  exp(bxn )
8
 Particle size production (reduction and enlargement)
Particle size reduction:

1. Rittingers postulate: energy required for particle reduction directly proportional to the area
of the new surface created.

1 1 dE 1
E  CR [  ] or  CR 2
x 2 x1 dX x
2. Kicks postulate: energy required for particle reduction directly proportional to the ratio in size
of the feed particle to the products particle.

dE 1 x
 CK or E  Ck ln( 1 )
dX x x2
3. Bonds Postulate

Bond empirically proposed an intermediate energy requirement solution of the form

10 10 dE 1
E  CB [ 0.5
 0.5
] or in differential form:  CB 1.5
( x2 ) ( x1) dx ( x)

9
The general form encompassing all three postulates is then

dE 1
 C N with N = 2 for Rittinger, N = 1 for Kick and N = 1.5 for Bond
dx x

typically
Kicks works better for coarse crushing and grinding 10-3 - 10 m
Rittingers works better for ultra fine grinding 10-8 -10-4 m
Bond works better for intermediate particle sizes 10-6 -10-3 m

10
Prediction of product size distribution
Product size distribution is modeled on the basis of two function, Sj and b(i, j).
Sj is the probability of a particle of size j being broken in unit time (or in say a certain number
of mill revolutions)
b(i,j) is the fraction of breakage product from size interval j that is transformed into size interval i
These numbers are found from small scale pilot tests usually.
To illustrate: We have 10 kg of monosized particles in size interval 1. If S1 = 0.6 we would
expect 4 kg of material to remain in size interval 1 after unit time. The size
distribution of the breakage product would be described by the set of b(i,j) values. Thus for
example if b(4,1) = 0.25 we would expect to find 25% by mass from size interval 1 to fall into
size interval 4. Mass b(i,j)
(kg)

2 0.9 0.15
10 kg of monosized 4 kg of unbroken
particles in interval 1 feed particles
3 1.8 0.3
+ 1.5 0.25
4

5 1.2 0.2

6 0.6 0.1
Unbroken product Broken product 11
Feed Product
Thus the rate of change of the mass of particles in size interval i with time:

j i 1
dmi
dt
  [b(i, j)S jm j ]  Simi
j 1

j i 1
 [b(i, j)S jm j ] = mass broken into interval i
j 1

Simi = mass broken out of interval i

and the rate of change of mass fraction of material in size interval i with time

j i 1
dyi
dt
  [b(i, j)S jy j ]  Siyi
j 1
mi
yi is the mass fraction in size interval i=
M

12
 Single particles in a fluid
Particle processing occurs most frequently in a fluid environment. This section focuses on the
forces acting on isolated particles and the steady state velocity that these particles attain
relative to the fluid.

The drag force for very slow relative motion (creeping motion) between a single rigid sphere
and a fluid of infinite extent has two components:

a pressure drag force: Fp = 2xU

a shear stress drag force: Fs = xU

and hence total drag force: FD = 3xU Stokes drag force expression

where:
U relative velocity between particle and fluid
x sphere diameter
 fluid viscosity
xUrf
Region of validity of Stokes expression for different Reynolds numbers Rep 
where: 
x rigid sphere diameter
U relative velocity
rf density of fluid 13
 viscosity of fluid
holds:
exactly for Rep  0.1

within 1% for Rep  0.3

within 3% for Rep  0.5

within 9% for Rep  1.0

why is that? The Reynolds number is the ratio of inertial to viscous forces. Stoke considered
only the viscous contributions. At high Reynolds numbers (when turbulent flow starts to be a
factor, there is an additional fluid drag force that becomes important as well.
R'
A drag coefficient , CD is defined as CD 
0.5 rf U2
where R’ is the force per unit projected area of the particle
FD
which for a sphere becomes: R' 
x 2
( )
4
and Stokes expression in terms of this drag coefficient becomes:
24
CD 
Rep
14
 The behaviour of the drag coefficient as a function of Re is shown below. Expressions for the
drag coefficient outside the Stokes region of validity cannot be found analytically using the
Navier-Stokes because the inertial forces as discussed become influential, but experiment allows
correlations to be developed for the other regimes.
Creeping flow Inertial flow
log CD
Stokes law Intermediate Newton’s Law Boundary layer separation

~0.3 ~500 ~2 105 log Rep

Region Stokes Intermediate Newton’s Law

Rep range <0.3 0.3<Rep<500
CD [24/Rep] [~24/Rep] + 0.44 ~ 0.44
also [24/Rep][1+0.15(Rep)0.687] 15
 Particles falling under gravity through a fluid:

The forces acting on the particle are:

gravity – buoyancy – drag = acceleration force

Eventually a force balance is achieved (steady state) and the particle attains a terminal velocity.

so:

(The buoyancy force on a body has the same magnitude as the weight of the fluid displaced by
the body: FB = rliquid g Vsolid)

x 3 x 3 x 2 CD 
R'
rpg  rf g  R' 0 which combined with gives:
6 6 4 0.5 rf U2
x 3 2 x
2
(rp  rf )g  CD 0.5rf UT 0 where UT is the single particle terminal velocity
6 4
4 g x [(rp rf )]
which in terms of CD becomes: CD 
3 U2T rf
24 x 2 (rp  rf )g
which in the Stokes regime where CD  gives UT 
Rep 18
16
(note that in the Stokes regime UT is proportional to the square of the particle diameter)
 x(rp  rf )g 0.5
In the Newton’s regime with CD ~ 0.44 UT 1.74[ ]
rf
(note that in the Newton’s regime UT is proportional to the square root of the particle diameter
and is independent of the fluid viscosity)

Generally though when calculating UT for a given particle diameter or the particle diameter for
a given UT,it is not necessarily known which region of operation is relevant:
CD
One way around this is by using the dimensionless groups: CD Rep2 and
Rep
To calculate UT, for a given size x:

4 x3rf (rp  rf )g
CD Rep2  which is independent of UT
3 2

For given particle and fluid properties CD Rep2 is a constant and therefore

log CD vs log Rep is a straight line with a slope of -2.

The intersection with the drag curve gives Rep and hence UT

17
To calculate size x, for a given UT:

CD 4grp rf )
 which is independent of particle size x
Rep 3U3Tr2f
CD
For a given terminal velocity, particle density and fluid properties, is a constant and therefore
Rep
log CD vs log Rep is a straight line with a slope of 1.

The intersection with the drag curve gives Rep and hence x

log CD slope = -2
from CD Re2 = constant

slope = +1
from (CD/Rep) = constant

X UT log Rep 18
 Non spherical particles:
The effect of the shape of non-spherical particles on their drag coefficient is dealt with by
defining a sphericity for these particles.
surface area of sphere of equal volume to theparticle
Sphericity:  
surface area of theparticle
The advantage of this definition is that a sphere has the smallest surface area per unit volume
of any shape. Hence, the value of sphericity will be fractional for all shapes and unity for spheres.
Drag curves for particles of different sphericities
10 4
Drag coefficient, CD

1000

100
 = 0.125
10  = 0.22
8
6
4  = 0.6
2   0.806
1   1.0

0.1
0.001 0.01 0.1 1 10 100 1000 10 4 105 106 19
0.4 0.8 Single particle Reynolds number, Re (Re : equiv. volume d)
0.2 0.6 p p
 Multiple particle systems
When many particles flow in a fluid in close proximity to each other the motion of each particle
is influenced by the presence of the others. Thus the analysis for fluid particle interaction for
a single particle needs to be adapted for multiple particle systems.

Need to adapt our basic variables to the presence of large numbers of particles in the fluid:


e  effective viscosity
f ()

rave  rf  (1  )rp average suspension density

 = voidage or volume fraction occupied by fluid

f(ε) = viscosity function

For operation in the Stokes regime:

24 24  e x vUrf
CD   Rep 
Rep Urel rave x 
R'
where CD  and Urel= relative velocity of particle to fluid
0.5 rave Urel
2

and R’ is the force per unit projected area of the particle 20

Under terminal velocity conditions (zero acceleration) for a particle falling under gravity in
a suspension the force balance:

24 24  e
drag force = weight-upthrust CD  
Rep Urel rave x
becomes:

x 2 x 3 x2 g
( ) 0.5 rave UrelCD  (rp  rave )(
2
)g or Urel  (rp  rave )
4 6 18e

substituting in for rave and e obtain for the terminal velocity for a particle in a suspension:


x 2g e 
Urel T (rp  rf ) f ( ) f ()
18
rave  rf  (1  )rp
Comparing single particle expression to above:
x 2 (rp  rf )g
UrelT=UTf() also known as the hindered settling velocity UT 
18
Some more expressions required, developed below with the following assumptions
fluid and particles incompressible
fluid and particle volumetric flowrates are constant
21
 Qf
Ufs  superficial fluid velocity
A

Qp
Ups  superficial particle velocity
A

A = vessel cross-sectional area

given isotropic conditions:

A f  A flow area occupied by the fluid

Ap (1  )A flow area occupied by the particles

so preserving continuity:

Qf Ufs A Uf A for the fluid

Qp Ups A  Up A (1  ) for the particles

Where Uf and Up are now the actual fluid and particle velocities respectively:
Ufs Ups
Uf  and Up 
 (1  ) 22
 Filtration
Filtration is the removal of suspended particles from a fluid. It is a very important industrial
process as it is often a key stage in product recovery following reaction, precipitation and
crystallization stages, but preceding thermal drying and packaging

We focus on cake filtration here. This involves the building of a “cake” of particles on a porous
surface (filter medium) which normally takes the form of a woven fabric. This filtration process
can be analyzed in terms of the flow of fluid through a packed bed of particles, the depth of
which increases with time. The voidage may also change with time.
We first consider the voidage constant case – incompressible cake

We first ignore the filter medium and focus on the cake itself.

The pressure drop versus liquid flow relationship can be described by Ergun’s equation:

p  U (1  )2 rf U2 (1  )
 150 2  1.75
H x sv  3
x sv 3
Understand this equation, do not need to remember
Most filtering processes involve laminar flow so we only need consider the first term.

For a particle fluid mixture with fixed particle properties a cake resistance is defined as:
150 (1  )2 ( p)
rc  so Ergun’s equation reduces to:  rc  U
x 2sv 3 H 23
 1 dV
U superficial filtrate velocity at time t
A dt
V is the filtrate (liquid) volume passed in time (t)

A is the cross-sectional area of the filter

dV
is the instantaneous volumetric flowrate
dt

HA
 Each unit volume of filtrate is assumed to deposit a certain mass of particles
V which form a certain volume of cake , which is the volume of cake formed by
the passage of unit volume of filtrate
H is the height of the cake

dV A 2 ( p)
so 
dt rc   V
Two notes:
1. Constant rate filtration
dV
If is constant then p V
dt
24
2. Constant pressure drop filtration
If (-p) is constant then:

dV 1

dt V
and integrating:
rc   t rc  
t c V with C1 =  V (for cake)
V 1 2 A 2 ( p) V 2 A 2 ( p)

Now to include the filter medium in the calculation as well:

Total resistance to flow = resistance due to the cake + resistance due to the filter medium

or in terms of pressure drop:

[total pressure drop] = [pressure drop across medium] + [pressure drop across cake]

(-p) = (-pm) + (-pc)

assuming now that the filter medium behaves like a packed bed of depth Hm and resistance rm:
1 dV
(-p)= (rm  Hm  rc  Hc ) and expressing rm Hm in terms of an
A dt equivalent cake thickness Heq: rm Hm = rc Heq 25
  Veq
which combining with Heq  Veq = volume of filtrate that must pass to create a
A cake of thickness Heq.

1 dV ( p)A
gives:  Which for the most frequent case of constant p gives
A dt rc  ( V  Veq ) upon integration

t rc   rc  
 V  Veq for cake + filter medium
V 2 A ( p)
2
A ( p)
2

Washing the cake

The solid particles separated by filtration normally must be washed to remove filtrate (liquid)
from the pores. The easy parts can be moved by simple displacement by clean solvent
whereas the harder to reach filtrate, is removed by diffusion in the wash solvent.

Fraction filtrate Sketch of filtrate removal during washing of the filter cake
In exiting wash Displacement
solution 100 only How the filtrate
concentration in
Diffusion+ Displacement the wash solvent
leaving the cake
typically varies
with volume of
wash solvent
0 passed
volume of wash solution 26
Volume of pores = AL
 Centrifugal separation and cyclones

a. Fluids used to transport powders frequently need to be purified again at the end of the transport

b. In the combustion of solid fuels, fine particles of fuel ash become suspended in the combustion
gases and must be removed before the gases can be discharged to the environment

What the most effective separation method is largely depends on the particle size.

Filters
100
Grade efficiency G(x)%

Electrostatic
precipitators

Cyclones
50

Wet separators
0

0.1 1.0 10 100

Particle size, x (m)

Cyclones typically primary separation devices for relatively coarse particles

27
 Gas cyclone:

Clean gas outlet

Inlet for
particle laden
gas

Vortex

Outlet for solids (reverse flow cyclone separator)

Particles in the gas are subjected to centrifugal forces which move them radially outwards,
against the inward flow of gas. Both gas and the particles (near the wall move downwards)
But the direction of flow of the gas reverses itself near the bottom and the gas moves up
whereas the solid particles move out through the bottom outlet.

Qualitatively, the rotational flow gives rise to a radial pressure gradient. This pressure gradient
together with the frictional pressure losses at the gas inlet and outlet and losses due to flow
direction changes give rise to the total static pressure drop. This static p outlet-inlet is typically
proportional to the square of the gas flow rate through the cyclone. This leads to the
definition of the resistance coefficient (Euler number). 28
 p
Euler number = Eu  Which is the ratio of the pressure to the inertial forces acting on
rf  2 a fluid element.
2
rf = gas density
4q
 (characteristic velocity)
D2
q = gas flow rate

D = cyclone diameter

Efficiency of separation

M = Mf + Mc
M = solids mass flowrate into cyclone

Mf = solids mass flowrate leaving with the gas (fine product)

Mc = solids mass flowrate discharged from the solids orifice (coarse product)

29
 In terms of component balance for each particle size x (assuming no breakage or growth):

dF dFf dFc
M = Mf + Mc
dx dx dx
dF dFf dFc Differential frequency size distributions by mass (mass fraction of size x), for
, ,
dx dx dx the feed, the fine product and coarse product

F, Ff , Fc Cumulative frequency size distribution by mass (mass fraction less than size x)
for the feed, fine product and coarse product.
Mc
ET = Total efficiency of separation of particles from gas
M
G(x)= Mass of solids of size x in coarse product Grade efficiency, efficiency with which
Mass of solids of size x in feed cyclone collects particles of a certain size
or
dFc dF
Mc ( ) ( c)
G( x )  dx
dF and G( x ) ET dx which relates total to grade efficiency
M( ) dF
( )
dx dx
dF dFc dFf which relates the size distributions of the feed, coarse and fine
and = ET + (1-ET)
dx dx dx product with the total efficiency
and analogously for the cumulative form F = ETFc + (1-ET)Ff 30
 Theoretical analysis for the gas cyclone separator

Considering a reverse flow cyclone separator of radius R

Forces on particle following a circular path are drag, buoyancy and centrifugal force
U2q
Uq a   r2
Ur Uq r
R a = particle acceleration towards
FB Fc circle centre
 = particle angular velocity
FD Ur
UR
particle on equilibrium
orbit of radius, r

UqR
Uin
Because the net flow of gas is inwards the drag force on the particle FD is inward
Assuming Stokes Law then FD = 3xUr
where x = particle diameter,  = gas viscosity, Ur = radial inward particle velocity
x 3 U2q
FB  rf buoyancy force where rf = density of fluid, U2q = tangential particle velocity
6 r at radius r
x 3 U2q
FC  rp centrifugal force where rp = density of particle 31
6 r
 At the particles equilibrium orbit:

 x 3
U 2
q x 3 U2q + 3xU
FC = FD + FB or rp = rf r
6 r 6 r
18  r
which can be rewritten as: x 
2
Ur
(rp  rf ) U2q

Now based on experiments in a cyclone, Uq r0.5 ~constant

thus: Uq r0.5 = UqR R0.5

and if we assume a uniform gas flow rate q towards the centre of the cyclone:

r
then: q = 2rLUr = 2RLUR hence UR = Ur
R
18  UR
and x2  r where r is the radius of the equilibrium orbit for a particle
(rp  rf ) UqR
2 of diameter x

if we assume that all particles with equilibrium orbit of R will be collected then

18  UR
2
xcrit  R is an expression of the critical particle diameter for separation
(rp  rf ) U2qR

32
 Mixing
Achieving good mixing of particulate solids, frequently of different size and density is important
in many of the process industries. For example in pharmaceutical production where the active
ingredient is present at only ~0.5% by mass overall.
Mechanisms of segregation
Particle of diameter x and density rp whose drag is governed by Stokes law is projected
horizontally with a velocity U into a fluid of viscosity  and density rf
1. Trajectory segregation
2
the retarding force on a particle moving in a fluid FD CD 0.5 rf U2 ( x )
4
CD = drag coefficient
deceleration of the particle = a = retarding force
mass of particle
24 xUrf
in Stokes’ law region, CD  Rep 
Rep f

18 U mp
thus deceleration = where rp 
rp x 2 x 3
6
from U(t)2 = U(0)2 + 2as (equation of motion) we obtain for an initial velocity U that the
particle traverses a distance Urp x
2
before coming to rest. Thus a particle of diameter
36 
2x will travel four times as far before coming to rest. This effect can cause segregation, say
33
when particles fall off a conveyor belt
Characterization of mixing
End-use determines quality required

Detergent – each scoop should have the same amount of active ingredient

Pharmaceutical tablet – each tablet should have the same amount of material and the same
amount of active ingredient

Quality of mixing depends on degree of scrutiny

TV image looks homogeneous from far but is made of tiny red blue and green spots when
looked under high magnification

Sampling
To determine the quality of a mixture it is generally necessary to take samples
The sampling of mixtures and the analysis of mixture quality require the application of
statistical methods

34
 Sampling
Mean composition
1N
y   yi N samples of composition y1 to yN
N i1
Standard deviation and variance (statistical measures of spread or variability - a measure
of the extent to which numbers are spread about their mean)
Variance estimated through:
N
 ( yi  )2
S2  i1 if the true composition  is known
N
N
 ( yi  y )2
S2  i1 if the true composition  is unknown
N 1

the standard deviation  is the square root of the variance

Theoretical limits of variance for a two component system where p and (1-p) are the
proportions of the two components in a mixture with n particles in each sample
02 p(1  p) upper limit (completely segregated)

p(1  p) 35
R2  lower limit (randomly mixed)
n
 Mixing indices measure of the degree of mixing

02  2 mixing achieved 1 completely random mixing Lacey index

02  R2 mixing possible 0 completely segregated


approaches unity for completely random mixtures Poole index
R

Standard error

Standard deviation of the variance

Tests for precision and variance - assignment of a degree of confidence to our mean mixture
and variance values obtained from sampling
1. sample composition
Based on N samples of mixture composition with mean y and standard deviation S, the true
mixture composition  may be stated with precision:
tS
 y  t found in Standard statistical tables
(N)0.5
36
2. Variance
when more than 50 samples are taken (n>50)

 2S2  {t E(S2 )} best estimate of the true variance 2

2
E(S2 ) S2 ( )0.5 standard error of the variance
N

when less than 50 samples are taken (n<50) upper and lower limit true variances are defined

S2 (N  1)
L2  lower limit
x 2a
Just understand these equations, do not need to remember

S2 (N  1)
U2  upper limit
x12 a

a, t, x from Standard statistical tables

37
 Movement and storage of particulate materials

Storage and flow of powders - Hoppers

Mass flow Core flow

Moving solids
All solids in motion
during discharge

Stationary
solids

Overall
In perfect mass flow all the powder in the silo (hopper) is in motion. These containers are typically
steep and the flow profile is plug flow until the conical section is reached.

In core flow the powder flows towards the outlet in a channel formed within the powder itself.38
Particulars
Advantages of Mass flow over Core flow

- powder motion uniform

- steady state easily approximated
- bulk density uniform
- low stresses, avoiding high powder compaction
- no stagnant regions, thus low risk of material degradation
- narrow range in particle residence times
- much lower particle size segregation

Disadvantages

- Friction between moving solids and walls resulting in wall erosion and/or solids contamination

What governs the flow patterns?

- slope angle
- particle particle friction
- particle wall friction

Thus, there is no such thing as a mass flow hopper; one that gives mass flow for one powder
may not for another

39
 Powder stresses within a cylindrical container

H = thickness of slice
D2
v H = depth at which slice is situated
4
v0 D = bin diameter
v = stress acting on the top surface of the slice
D2
 v =downward force acting on the top
4 surface of slice
D2 (   )
v H
4 v  v = upward force acting on the
bottom surface of slice
wh wh
H h = stress exerted on the wall by the powder
h h
w = wall friction factor
v + v
 D H w h = upward friction force on the slice
D D2
rBgH = gravitational force on slice
4
rB = bulk density of powder
(and does not vary with depth)
40
 Net force:

D2 (   ) D2 D 2
4 v v +  D H w h = v + rBgH
4 4

Dv + 4whH = DrBgH

Assuming that horizontal stress is linearly proportional to the vertical stress: h = kv

and looking at the limit as the slice thickness H goes to 0 obtain:

d v 4 k
 ( w ) v  r B g
dH D
this is the same as:
4μ k
{( w )}H 4μ k
d[e D σ ] {( w )}H
v  ρ g[e D ]
dH B
which integrates to:
Just understand this equation, do not need to remember
4 w k 4 w k
{( )}H DrB g {( )}H
v e D  e D  cons tan t
4 w k 41
 For a stress v0 acting at the free surface (H=0) of the powder:

4 w k 4 w k
DrB g  {( )}H  {( )}H
v  [1  e D ] v0 e
D
4 w k
Just understand this equation, do not need to remember
If there is no force acting on the free surface, v0 = 0

4 w k
DrB g  {( )}H
v  [1  e D ]
4 w k

for a very short H obtain:

v  rBHg (based on e-z 1-z for small z)

which is equivalent to the static pressure in a fluid of density rB at depth H

For large H
DrB g
v  So the vertical stress developed is independent of depth of powder.
4 w k

42
Variation in stress with depth of powder for the case v0 = 0

v This is contrary to intuition perhaps, the force

rBHg exerted by a bed of powder becomes independent
of depth if the bed is deep enough. That means that
most of the weight of the powder is supported by the
walls of the bin. In practice, the stress becomes
independent of depth beyond a depth of ~4D

DrB g
4 w k
H

Mass flowrates
The rate of discharge of powder from an orifice at the base of a bin is found to be independent
of the depth of powder unless the bin is nearly empty. This means that the observation for a
static powder that the pressure exerted by the powder is independent of depth for large depths
is also true for a dynamic system.
For flow through an orifice in the flat-based cylinder, experiment shows that:
Mass flow rate, Mp  (B-a)2.5 B = circular orifice diameter, a = correction factor dependent
on particle size.
For solids discharge from conical apertures in flat-based cylinders Mp = 0.58rBg0.5(B-kx)2.543
 Basic relationship governing flow of gas and particles in a pipe
Gas and particle velocities
volume flow of gas Q
Superficial gas velocity, Ufs   f
cross sec tional area of pipe A

volume flow of solids Qp

Superficial solids velocity, Ups  
cross sec tional area of pipe A
s= superficial,
p= particle
f = fluid
Fraction of pipe cross-sectional area available for flow of gas proportional to volume fraction
occupied by gas ( void fraction)
Fraction of pipe cross-sectional area available for flow of solids proportional to volume fraction
occupied by solids (1-)
Qf U
Actual gas velocity, Uf  or Uf  fs
A 
Qp U
Actual particle velocity, Up  or Up  ps
A(1  ) 1 

relative velocity between particle and fluid Urel  Uf  Up also known as Uslip
44
Note: frequently assume that in vertical dilute phase flow Uslip = single particle terminal velocity UT
Continuity

Mp = particle mass flowrate fed into a transport pipe

Mf = gas mass flowrate fed into a transport pipe

Mp = AUp(1-)rp

Mf = AUfrf

Mp Up (1  )rp
solids loading 
Mf Uf  rf

Pressure drop

To obtain an expression for the total pressure drop along a pipe need to write down a
momentum equation for a section of pipe.

We consider a section of pipe of cross-sectional area A and length dL inclined to the horizontal
at an angle q (which contains the two most common cases - vertical and horizontal pipes),
and carrying a suspension of voidage .
45
 Up + dUp
Uf + dUf
dL

Area A

Up
Exit pressure
Uf
p + dp
q

Inlet pressure p

Momentum balance
[net force acting on pipe content] = [rate of increase in momentum of contents]

[pressure force] – [gas/wall friction force] – [solids/wall friction force] – [gravitational force]

= [rate of increase in momentum of the gas] + [rate of increase in momentum of the solids]
[-Adp] – [ FfwAdL] – [FpwAdL] – [(A[1-]rpdL)gsinq + (ArfdL)gsinq]

= [rfAUfdUf] + [rpA(1-)UpdUp]
46
Ffw and Fpw are respectively, the gas-to-wall and solids-to-wall friction forces per unit volume of pipe
Fluidization Regimes

Understand the equations

Geldart Groupings

Group A Prior to the initiation of a bubbling bed phase, beds from these particles will
expand by a factor of 2 to 3 at incipient fluidization, due to a decreased bulk density. Most
powder-catalyzed beds utilize this group.
Group B Bubbling typically forms directly at incipient fluidization.
Group C This group contains extremely fine and consequently the most cohesive particles.

Group D Fluidization of this

group requires very high
fluid energies and is
typically associated with
high levels of abrasion.
Drying grains and peas,
roasting coffee beans,
gasifying coals, and some
roasting metal ores are such
solids, and they are usually
processed in shallow beds or
in the spouting mode.
 Separation techniques (concentration)
Minerals processing is for separating valuable minerals from
gangue. It is based on the difference of minerals in their size,
density, magnetic susceptibility, surface wettability and
electrical conductivity.

Separation techniques used in minerals engineering:

1. Size separation
2. Gravity separation
3. Magnetic separation
4. Froth flotation
5. Electrostatic separation
6. Optical separation
7. Leaching
 Wettability
• Wetting is the ability of a liquid to maintain contact with a solid
surface, resulting from intermolecular interactions when the two are
brought together.
• Wetting (wettability) is measured by contact angle.
• Contact angle is the angle, conventionally measured through the
liquid, where a liquid/vapor interface meets a solid surface.
• Young's Equation:

• Contact angle less than 90°: hydrophilic surface. Contact angle

greater than 90o: hydrophobic surface.
 Zeta Potential (ζ)
When charged surfaces are contacted with
polyelectrolyte solution, the charge is
balanced by counter ions. The layer of the
counter ion adsorbed on the surface is
called Stern layer and the remainder of the
counter ions are distributed in the diffuse
layer. In case of negatively charged
surface, the electric potential on the
surface is the highest and it decreases
rapidly until the Stern Plane due to the
adsorbed positive ions. Outside the Stern
Plane, the magnitude of potential
decreases slowly.

the zeta potential is the electric potential in the interfacial double layer (DL) at the location of
the slipping plane relative to a point in the bulk fluid away from the interface.