1

Particle Size Analysis

1.1 INTRODUCTION

In many powder handling and processing operations particle size and size
distribution play a key role in determining the bulk properties of the powder.
Describing the size distribution of the particles making up a powder is therefore
central in characterizing the powder. In many industrial applications a single
number will be required to characterize the particle size of the powder. This can
only be done accurately and easily with a mono-sized distribution of spheres or
cubes. Real particles with shapes that require more than one dimension to fully
describe them and real powders with particles in a range of sizes, mean that in
practice the identification of single number to adequately describe the size of the
particles is far from straightforward. This chapter deals with how this is done.

1.2 DESCRIBING THE SIZE OF A SINGLE PARTICLE

Regular-shaped particles can be accurately described by giving the shape and a

number of dimensions. Examples are given in Table 1.1.
The description of the shapes of irregular-shaped particles is a branch of
science in itself and will not be covered in detail here. Readers wishing to know
more on this topic are referred to Hawkins (1993). However, it will be clear to the
reader that no single physical dimension can adequately describe the size of
an irregularly shaped particle, just as a single dimension cannot describe the
shape of a cylinder, a cuboid or a cone. Which dimension we do use will in
practice depend on (a) what property or dimension of the particle we are able to
measure and (b) the use to which the dimension is to be put.
If we are using a microscope, perhaps coupled with an image analyser, to view
the particles and measure their size, we are looking at a projection of the shape of
the particles. Some common diameters used in microscope analysis are statistical
diameters such as Martins diameter (length of the line which bisects the particle

Introduction to Particle Technology, 2nd Edition Martin Rhodes

2008 John Wiley & Sons Ltd. ISBN: 978-0-470 -01427-1
2 PARTICLE SIZE ANALYSIS

Table 1.1 Regular-shaped particles

Shape Sphere Cube Cylinder Cuboid Cone
Dimensions Radius Side length Radius and Three side Radius and
height lengths height

image), Ferets diameter (distance between two tangents on opposite sides of the
particle) and shear diameter (particle width obtained using an image shearing
device) and equivalent circle diameters such as the projected area diameter (area
of circle with same area as the projected area of the particle resting in a stable
position). Some of these diameters are described in Figure 1.1. We must

Figure 1.1 Some diameters used in microscopy

DESCRIBING THE SIZE OF A SINGLE PARTICLE 3

Figure 1.2 Comparison of equivalent sphere diameters

remember that the orientation of the particle on the microscope slide will affect
the projected image and consequently the measured equivalent sphere diameter.
If we use a sieve to measure the particle size we come up with an equivalent
sphere diameter, which is the diameter of a sphere passing through the same
sieve aperture. If we use a sedimentation technique to measure particle size then
it is expressed as the diameter of a sphere having the same sedimentation
velocity under the same conditions. Other examples of the properties of
particles measured and the resulting equivalent sphere diameters are given in
Figure 1.2.
Table 1.2 compares values of these different equivalent sphere diameters used to
describe a cuboid of side lengths 1, 3, 5 and a cylinder of diameter 3 and length 1.
The volume equivalent sphere diameter or equivalent volume sphere diameter
is a commonly used equivalent sphere diameter. We will see later in the chapter
that it is used in the Coulter counter size measurements technique. By definition,
the equivalent volume sphere diameter is the diameter of a sphere having the
same volume as the particle. The surface-volume diameter is the one measured
when we use permeametry (see Section 1.8.4) to measure size. The surface-
volume (equivalent sphere) diameter is the diameter of a sphere having the same
surface to volume ratio as the particle. In practice it is important to use the method of

Table 1.2 Comparison of equivalent sphere diameters

Sphere passing Sphere Sphere having Sphere having the
the same sieve having the same surface the same surface to
Shape aperture, xp same volume, xv area, xs volume ratio, xsv
Cuboid 3 3.06 3.83 1.95
Cylinder 3 2.38 2.74 1.80
4 PARTICLE SIZE ANALYSIS

size measurement which directly gives the particle size which is relevant to the situation
or process of interest. (See Worked Example 1.1.)

1.3 DESCRIPTION OF POPULATIONS OF PARTICLES

A population of particles is described by a particle size distribution. Particle size

distributions may be expressed as frequency distribution curves or cumulative
curves. These are illustrated in Figure 1.3. The two are related mathematically in
that the cumulative distribution is the integral of the frequency distribution; i.e. if
the cumulative distribution is denoted as F, then the frequency distribution
dF=dx. For simplicity, dF=dx is often written as fx: The distributions can be by
number, surface, mass or volume (where particle density does not vary with size,
the mass distribution is the same as the volume distribution). Incorporating this
information into the notation, fN x is the frequency distribution by number, fS x
is the frequency distribution by surface, FS is the cumulative distribution by

Figure 1.3 Typical differential and cumulative frequency distributions

CONVERSION BETWEEN DISTRIBUTIONS 5

Figure 1.4 Comparison between distributions

surface and FM is the cumulative distribution by mass. In reality these distribu-

tions are smooth continuous curves. However, size measurement methods often
divide the size spectrum into size ranges or classes and the size distribution
becomes a histogram.
For a given population of particles, the distributions by mass, number and
surface can differ dramatically, as can be seen in Figure 1.4.
A further example of difference between distributions for the same population
is given in Table 1.3 showing size distributions of man-made objects orbiting the
earth (New Scientist, 13 October 1991).
The number distribution tells us that only 0.2% of the objects are greater than
10 cm. However, these larger objects make up 99.96% of the mass of the
population, and the 99.3% of the objects which are less than 1.0 cm in size
make up only 0.01% of the mass distribution. Which distribution we would use is
dependent on the end use of the information.

1.4 CONVERSION BETWEEN DISTRIBUTIONS

Many modern size analysis instruments actually measure a number distribution,

which is rarely needed in practice. These instruments include software to

Table 1.3 Mass and number distributions for man-made objects orbiting the earth

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

101000 7000 0.2 99.96
110 17 500 0.5 0.03
0.11.0 3 500 000 99.3 0.01
Total 3 524 500 100.00 100.00
6 PARTICLE SIZE ANALYSIS

convert the measured distribution into more practical distributions by mass,

surface, etc.
Relating the size distributions by number, fN x, and by surface, fS x for a
population of particles having the same geometric shape but different size:

Fraction of particles in the size range

x to x dx fN xdx

Fraction of the total surface of particles in the size range

x to x dx fS xdx

If N is the total number of particles in the population, the number of particles in

the size range x to x dx NfN xdx and the surface area of these particles
x2 aS NfN xdx, where aS is the factor relating the linear dimension of the
particle to its surface area.
Therefore, the fraction of the total surface area contained on these particles
[fS xdx] is:

x2 aS NfN xdx
S

where S is the total surface area of the population of particles.

For a given population of particles, the total number of particles, N, and the
total surface area, S are constant. Also, assuming particle shape is independent of
size, aS is constant, and so

fS x / x2 fN x or fS x kS x2 fN x 1:1

where

aS N
kS
S

Similarly, for the distribution by volume

fV x kV x3 fN x 1:2

where

aV N
kV
V

where V is the total volume of the population of particles and aV is the factor
relating the linear dimension of the particle to its volume.
DESCRIBING THE POPULATION BY A SINGLE NUMBER 7

And for the distribution by mass

fm x km rp x3 fN x 1:3

where

aV r p N
km
V

assuming particle density rp is independent of size.

The constants kS , kV and km may be found by using the fact that:

1
fxdx 1 1:4
0

Thus, when we convert between distributions it is necessary to make assump-

tions about the constancy of shape and density with size. Since these assumptions
may not be valid, the conversions are likely to be in error. Also, calculation errors
are introduced into the conversions. For example, imagine that we used an
electron microscope to produce a number distribution of size with a measure-
ment error of
2%. Converting the number distribution to a mass distribution we
triple the error involved (i.e. the error becomes
6%). For these reasons,
conversions between distributions are to be avoided wherever possible. This
can be done by choosing the measurement method which gives the required
distribution directly.

1.5 DESCRIBING THE POPULATION BY A SINGLE NUMBER

In most practical applications, we require to describe the particle size of a

population of particles (millions of them) by a single number. There are many
options available; the mode, the median, and several different means including
arithmetic, geometric, quadratic, harmonic, etc. Whichever expression of central
tendency of the particle size of the population we use must reflect the property or
properties of the population of importance to us. We are, in fact, modelling the
real population with an artificial population of mono-sized particles. This section
deals with calculation of the different expressions of central tendency and
selection of the appropriate expression for a particular application.
The mode is the most frequently occurring size in the sample. We note,
however, that for the same sample, different modes would be obtained for
distributions by number, surface and volume. The mode has no practical
significance as a measure of central tendency and so is rarely used in practice.
The median is easily read from the cumulative distribution as the 50% size; the
size which splits the distribution into two equal parts. In a mass distribution, for
example, half of the particles by mass are smaller than the median size. Since the
8 PARTICLE SIZE ANALYSIS

Table 1.4 Definitions of means

gx Mean and notation
x arithmetic mean, x
a
x2 quadratic mean, x
q
x3 cubic mean, x
c
log x geometric mean, x
g
1=x harmonic mean, x
h

median is easily determined, it is often used. However, it has no special

significance as a measure of central tendency of particle size.
Many different means can be defined for a given size distribution; as pointed
out by Svarovsky (1990). However, they can all be described by:
1 1 1
gxdF
gx
0
1 but dF 1 and so gx
gxdF 1:5
0 dF 0 0

where x
is the mean and g is the weighting function, which is different for each
mean definition. Examples are given in Table 1.4.
Equation (1.5) tells us that the mean is the area between the curve and the Fx
axis in a plot of Fx versus the weighting function gx (Figure 1.5). In fact,
graphical determination of the mean is always recommended because the
distribution is more accurately represented as a continuous curve.
Each mean can be shown to conserve two properties of the original population
of particles. For example, the arithmetic mean of the surface distribution
conserves the surface and volume of the original population. This is demon-
strated in Worked Example 1.3. This mean is commonly referred to as the
surface-volume mean or the Sauter mean. The arithmetic mean of the number

Figure R1.5 Plot of cumulative frequency against weighting function gx. Shaded area is
1
gx
0 gx dF
DESCRIBING THE POPULATION BY A SINGLE NUMBER 9

distribution x
aN conserves the number and length of the original population and
is known as the number-length mean x
NL :
1
x dFN
number-length mean; x
NL x
aN 0 1 1:6
0 dFN

As another example, the quadratic mean of the number distribution x
qN con-
serves the number and surface of the original population and is known as the
number-surface mean xNS :
1
x2 dFN
number-surface mean; x
2NS x
2qN 0 1 1:7
0 dFN

A comparison of the values of the different means and the mode and median for
a given particle size distribution is given in Figure 1.6. This figure highlights two
points: (a) that the values of the different expressions of central tendency can vary
significantly; and (b) that two quite different distributions could have the same

Figure 1.6 Comparison between measures of central tendency. Adapted from Rhodes
(1990). Reproduced by permission
10 PARTICLE SIZE ANALYSIS

arithmetic mean or median, etc. If we select the wrong one for our design
correlation or quality control we may be in serious error.
So how do we decide which mean particle size is the most appropriate one for
a given application? Worked Examples 1.3 and 1.4 indicate how this is done.
For Equation (1.8), which defines the surface-volume mean, please see Worked
Example 1.3.

1.6 EQUIVALENCE OF MEANS

Means of different distributions can be equivalent. For example, as is shown

below, the arithmetic mean of a surface distribution is equivalent (numerically
equal to) the harmonic mean of a volume (or mass) distribution:
1
x dFS
arithmetic mean of a surface distribution; x
aS 0 1 1:9
0 dFS

The harmonic mean xhV of a volume distribution is defined as:


1 1
0 dFV
1 x
1 1:10
x
hV
0 dFV

From Equations (1.1) and (1.2), the relationship between surface and volume
distributions is:
kv
dFv x dFs 1:11
ks
hence


1 1 kv
x dFs 1
1 0 x ks dFs
10 1:12
x
hV 1 kv
0 x dFs
0 x k dFs
s

(assuming ks and kv do not vary with size)

and so
1
x dFs
x
hV 0 1
0 dFs

which, by inspection, can be seen to be equivalent to the arithmetic mean of the

surface distribution x
aS [Equation (1.9)].
Recalling that d Fs x2 kS d FN , we see from Equation (1.9) that
1 3
x dFN


xaS 01
2
0 x dFN

which is the surface-volume mean, x
SV [Equation (1.8) - see Worked Example 1.3].
COMMON METHODS OF DISPLAYING SIZE DISTRIBUTIONS 11

Summarizing, then, the surface-volume mean may be calculated as the

arithmetic mean of the surface distribution or the harmonic mean of the volume
distribution. The practical significance of the equivalence of means is that it
permits useful means to be calculated easily from a single size analysis.
The reader is invited to investigate the equivalence of other means.

1.7 COMMON METHODS OF DISPLAYING SIZE DISTRIBUTIONS

1.7.1 Arithmetic-normal Distribution

In this distribution, shown in Figure 1.7, particle sizes with equal differences
from the arithmetic mean occur with equal frequency. Mode, median and
arithmetic mean coincide. The distribution can be expressed mathematically by:
" #
dF 1 x  x
2
p exp  1:13
dx s 2p 2s2

where s is the standard deviation.

To check for a arithmetic-normal distribution, size analysis data is plotted on
normal probability graph paper. On such graph paper a straight line will result if
the data fits an arithmetic-normal distribution.

1.7.2 Log-normal Distribution

This distribution is more common for naturally occurring particle populations.

An example is shown in Figure 1.8. If plotted as dF=dlog x versus x, rather than
dF=dx versus x, an arithmetic-normal distribution in log x results (Figure 1.9). The
mathematical expression describing this distribution is:
" #
dF 1 z  z
2
p exp  1:14
dz sz 2p 2s2z

Figure 1.7 Arithmetic-normal distribution with an arithmetic mean of 45 and standard

deviation of 12
12 PARTICLE SIZE ANALYSIS

Figure 1.8 Log-normal distribution plotted on linear coordinates

where z log x, z
is the arithmetic mean of log x and sz is the standard deviation
of log x.
To check for a log-normal distribution, size analysis data are plotted on log-
normal probability graph paper. Using such graph paper, a straight line will
result if the data fit a log-normal distribution.

1.8 METHODS OF PARTICLE SIZE MEASUREMENT

1.8.1 Sieving

Dry sieving using woven wire sieves is a simple, cheap method of size analysis
suitable for particle sizes greater than 45 mm. Sieving gives a mass distribution and a
size known as the sieve diameter. Since the length of the particle does not hinder its
passage through the sieve apertures (unless the particle is extremely elongated), the
sieve diameter is dependent on the maximum width and maximum thickness of the

Figure 1.9 Log-normal distribution plotted on logarithmic coordinates

METHODS OF PARTICLE SIZE MEASUREMENT 13

particle. The most common modern sieves are in sizes such that the ratio of adjacent
sieve sizes is the fourth root of two (eg. 45, 53, 63, 75, 90, 107 mm). If standard
procedures are followed and care is taken, sieving gives reliable and reproducible
size analysis. Air jet sieving, in which the powder on the sieve is fluidized by a jet or
air, can achieve analysis down to 20 mm. Analysis down to 5 mm can be achieved by
wet sieving, in which the powder sample is suspended in a liquid.

1.8.2 Microscopy

The optical microscope may be used to measure particle sizes down to 5 mm. For
particles smaller than this diffraction causes the edges of the particle to
be blurred and this gives rise to an apparent size. The electron microscope
may be used for size analysis below 5 mm. Coupled with an image analysis
system the optical microscope or electron microscope can readily give number
distributions of size and shape. Such systems calculate various diameters from
the projected image of the particles (e.g. Martins, Ferets, shear, projected area
diameters, etc.). Note that for irregular-shaped particles, the projected area
offered to the viewer can vary significantly depending on the orientation of the
particle. Techniques such as applying adhesive to the microscope slide may be
used to ensure that the particles are randomly orientated.

1.8.3 Sedimentation

In this method, the rate of sedimentation of a sample of particles in a liquid is

followed. The suspension is dilute and so the particles are assumed to fall at their
single particle terminal velocity in the liquid (usually water). Stokes law is
assumed to apply Rep < 0:3 and so the method using water is suitable only for
particles typically less than 50 mm in diameter. The rate of sedimentation of the
particles is followed by plotting the suspension density at a certain vertical
position against time. The suspension density is directly related to the cumulative
undersize and the time is related to the particle diameter via the terminal
velocity. This is demonstrated in the following:
Referring to Figure 1.10, the suspension density is sampled at a vertical
distance, h below the surface of the suspension. The following assumptions are
made:

 The suspension is sufficiently dilute for the particles to settle as individuals (i.e.
not hindered settling see Chapter 3).

 Motion of the particles in the liquid obeys Stokes law (true for particles
typically smaller than 50 mm).

 Particles are assumed to accelerate rapidly to their terminal free fall velocity UT
so that the time for acceleration is negligible.
14 PARTICLE SIZE ANALYSIS

Figure 1.10 Size analysis by sedimentation

Let the original uniform suspension density be C0 . Let the suspension density
at the sampling point be C at time t after the start of settling. At time t all those
particles travelling faster than h=t will have fallen below the sampling point. The
sample at time t will therefore consist only of particles travelling a velocity  h=t.
Thus, if C0 is representative of the suspension density for the whole population,
then C represents the suspension density for all particles which travel at a
velocity  h=t, and so C=C0 is the mass fraction of the original particles which
travel at a velocity  h=t. That is,

C
cumulative mass fraction
C0

All particles travel at their terminal velocity given by Stokes law [Chapter 2,
Equation (2.13)]:

x2 rp  rf g
UT
18 m

Thus, equating UT with h=t, we determine the diameter of the particle travelling
at our cut-off velocity h=t. That is,
" #1=2
18 mh
x 1:15
trp  rf g

Particles smaller than x will travel slower than h=t and will still be in suspension
at the sampling point. Corresponding values of C=C0 and x therefore give us the
cumulative mass distribution. The particle size measured is the Stokes diameter,
i.e. the diameter of a sphere having the same terminal settling velocity in the
Stokes region as the actual particle.
METHODS OF PARTICLE SIZE MEASUREMENT 15

A common form of this method is the Andreason pipette which is capable of

measuring in the range 2100 mm. At size below 2 mm, Brownian motion causes
significant errors. Increasing the body force acting on the particles by centrifu-
ging the suspension permits the effects of Brownian motion to be reduced so that
particle sizes down to 0:01 mm can be measured. Such a device is known as a
pipette centrifuge.
The labour involved in this method may be reduced by using either light
absorption or X-ray absorption to measure the suspension density. The
light absorption method gives rise to a distribution by surface, whereas the
X-ray absorption method gives a mass distribution.

1.8.4 Permeametry

This is a method of size analysis based on fluid flow through a packed bed
(see Chapter 6). The CarmanKozeny equation for laminar flow through a
randomly packed bed of uniformly sized spheres of diameter x is [Equation 6.9]:

p 1  e2 mU
180
H e3 x2

where p is the pressure drop across the bed, e is the packed bed void fraction,
H is the depth of the bed, m is the fluid viscosity and U is the superficial fluid
velocity. In Worked Example 1.3, we will see that, when we are dealing with non-
spherical particles with a distribution of sizes, the appropriate mean diameter for
this equation is the surface-volume diameter x
SV , which may be calculated as the
arithmetic mean of the surface distribution, x
aS :
In this method, the pressure gradient across a packed bed of known voidage is
measured as a function of flow rate. The diameter we calculate from the
CarmanKozeny equation is the arithmetic mean of the surface distribution
(see Worked Example 6.1 in Chapter 6).

1.8.5 Electrozone Sensing

Particles are held in supension in a dilute electrolyte which is drawn through a

tiny orifice with a voltage applied across it (Figure 1.11). As particles flow
through the orifice a voltage pulse is recorded.
The amplitude of the pulse can be related to the volume of the particle passing
the orifice. Thus, by electronically counting and classifying the pulses according
to amplitude this technique can give a number distribution of the equivalent
volume sphere diameter. The lower size limit is dictated by the smallest practical
orifice and the upper limit is governed by the need to maintain particles in
suspension. Although liquids more viscous than water may be used to reduce
sedimentation, the practical range of size for this method is 0.31000 mm. Errors
are introduced if more that one particle passes through the orifice at a time and so
dilute suspensions are used to reduce the likelihood of this error.
16 PARTICLE SIZE ANALYSIS

Figure 1.11 Schematic of electrozone sensing apparatus

1.8.6 Laser Diffraction

This method relies on the fact that for light passing through a suspension, the
diffraction angle is inversely proportional to the particle size. An instrument
would consist of a laser as a source of coherent light of known fixed
wavelength (typically 0:63 mm), a suitable detector (usually a slice of photo-
sensitive silicon with a number of discrete detectors, and some means of
passing the sample of particles through the laser light beam (techniques are
available for suspending particles in both liquids and gases are drawing them
through the beam).
To relate diffraction angle with particle size, early instruments used the
Fraunhofer theory, which can give rise to large errors under some circumstances
(e.g. when the refractive indices of the particle material and suspending medium
approach each other). Modern instruments use the Mie theory for interaction of
light with matter. This allows particle sizing in the range 0.12000 mm, provided
that the refractive indices of the particle material and suspending medium are
known.
This method gives a volume distribution and measures a diameter known as
the laser diameter. Particle size analysis by laser diffraction is very common in
industry today. The associated software permits display of a variety of size
distributions and means derived from the original measured distribution.

1.9 SAMPLING

In practice, the size distribution of many tonnes of powder are often assumed
from an analysis performed on just a few grams or milligrams of sample. The
importance of that sample being representative of the bulk powder cannot be
overstated. However, as pointed out in Chapter 11 on mixing and segregation,
most powder handling and processing operations (pouring, belt conveying,
WORKED EXAMPLES 17

handling in bags or drums, motion of the sample bottle, etc.) cause particles
to segregate according to size and to a lesser extent density and shape. This
natural tendency to segregation means that extreme care must be taken in
sampling.
There are two golden rules of sampling:

1. The powder should be in motion when sampled.

2. The whole of the moving stream should be taken for many short time increments.

Since the eventual sample size used in the analysis may be very small, it is often
necessary to split the original sample in order to achieve the desired amount for
analysis. These sampling rules must be applied at every step of sampling and
sample splitting.
Detailed description of the many devices and techniques used for sampling in
different process situations and sample dividing are outside the scope of this
chapter. However, Allen (1990) gives an excellent account, to which the reader is
referred.

1.10 WORKED EXAMPLES

WORKED EXAMPLE 1.1

Calculate the equivalent volume sphere diameter xv and the surface-volume equivalent
sphere diameter xsv of a cuboid particle of side length 1, 2, 4 mm.

Solution

The volume of cuboid 1  2  4 8 mm3

The surface area of the particle 1  2 1  2 1 2 1 2  4 28 mm2

The volume of sphere of diameter xv is px3v =6

Hence, diameter of a sphere having a volume of 8 mm3 ; xv 2:481 mm

The equivalent volume sphere diameter xv of the cuboid particle is therefore xv 2:481 mm
28
The surface to volume ratio of the cuboid particle 3:5 mm2 =mm3
8
The surface to volume ratio for a sphere of diameter xsv is therefore 6=xsv

Hence, the diameter of a sphere having the same surface to volume ratio as the particle
6=3:5 1:714 mm

The surface-volume equivalent sphere diameter of the cuboid, xsv 1:714 mm

18 PARTICLE SIZE ANALYSIS

WORKED EXAMPLE 1.2

Convert the surface distribution described by the following equation to a cumulative

volume distribution:

FS x=452 for x  45 mm
FS 1 for x > 45 mm

Solution

From Equations (1.1)(1.3),

kv
fv x xfs x
ks

Integrating between sizes 0 and x:

x

kv
Fv x xfs xdx
0 ks

Noting that fs x dFs =dx, we see that

d  x 2 2x
fs x
dx 45 452

and our integral becomes

x

kv 2x2
Fv x dx
0 ks 452

Assuming that kv and ks are independent of size,

 x
kv 2x2
Fv x dx
ks 0 452
" #
2 x3 kv

3 452 ks

kv =ks may be found by noting that Fv 45 1; hence

90 kv kv
1 and so 0:0333
3 ks ks

Thus, the formula for the volume distribution is

Fv 1:096  105 x3 for x  45 mm

Fv 1 for x > 45 mm
WORKED EXAMPLES 19

WORKED EXAMPLE 1.3

What mean particle size do we use in calculating the pressure gradient for flow of a fluid
through a packed bed of particles using the CarmanKozeny equation (see Chapter 6)?

Solution

The CarmanKozeny equation for laminar flow through a randomly packed bed of
particles is:

p 1  e2 2
K Sv mU
L e3

where Sv is the specific surface area of the bed of particles (particle surface area per unit
particle volume) and the other terms are defined in Chapter 6. If we assume that the bed
voidage is independent of particle size, then to write the equation in terms of a mean
particle size, we must express the specific surface, Sv , in terms of that mean. The particle
size we use must give the same value of Sv as the original population or particles. Thus
the mean diameter x
must conserve the surface and volume of the population; that is,
the mean must enable us to calculate the total volume from the total surface of the
particles. This mean is the surface-volume mean x
sv


av av 1
x
sv  total surface  total volume eg: for spheres;
as as 6
1 1
kv
and therefore x
sv fs xdx  fv xdx
k
0
1s 0

Total volume of particles; V x3 aV NfN xdx

0
1
Total surface area of particles; S x2 aS NfN xdx
0
1
as 0 x3 av NfN xdx
Hence; x
sv 1 2
av 0 x as NfN xdx

Then, since aV , aS and N are independent of size, x,

1 3 1 3
x fN xdx x dFN
x
sv 01 2 01
0 x f N xdx 2
0 x dFN

This is the definition of the mean which conserves surface and volume, known as the
surface-volume mean, x
SV .

So
1 3
x dFN
x
SV 01 1:8
2
0 x dFN

The correct mean particle diameter is therefore the surface-volume mean as defined
above. (We saw in Section 1.6 that this may be calculated as the arithmetic mean of the
20 PARTICLE SIZE ANALYSIS

surface distribution x
aS , or the harmonic mean of the volume distribution.) Then in the
CarmanKozeny equation we make the following substitution for Sv :

1 ks
Sv


xSV kv
e.g. for spheres, Sv 6=x
SV:

WORKED EXAMPLE 1.4 (AFTER SVAROVSKY, 1990)

A gravity settling device processing a feed with size distribution Fx and operates with
a grade efficiency Gx. Its total efficiency is defined as:
1
ET Gxd FM
0

How is the mean particle size to be determined?

Solution

Assuming plug flow (see Chapter 3), Gx UT A=Q where, A is the settling area, Q is
the volume flow rate of suspension and UT is the single particle terminal velocity for
particle size x, given by (in the Stokes region):

x2 rp  rf g
UT Chapter 2
18 m

hence 1
Agrp  rf
ET x2 d FM
18 mQ 0
1
where 0 x2 d FM is seen to be the definition of the quadratic mean of the distribution by
mass x
qM (see Table 1.4).

This approach may be used to determine the correct mean to use in many
applications.

WORKED EXAMPLE 1.5

A Coulter counter analysis of a cracking catalyst sample gives the following cumulative
volume distribution:

Channel 1 2 3 4 5 6 7 8
% volume differential 0 0.5 1.0 1.6 2.6 3.8 5.7 8.7

Channel 9 10 11 12 13 14 15 16
% volume differential 14.3 22.2 33.8 51.3 72.0 90.9 99.3 100
WORKED EXAMPLES 21

(a) Plot the cumulative volume distribution versus size and determine the median size.

(b) Determine the surface distribution, giving assumptions. Compare with the volume
distribution.

(c) Determine the harmonic mean diameter of the volume distribution.

(d) Determine the arithmetic mean diameter of the surface distribution.

Solution

With the Coulter counter the channel size range differs depending on the tube in use.
We therefore need the additional information that in this case channel 1 covers the size
range 3:17 mm to 4:0 mm, channel 2 covers the range 4:0 mm to 5:04 mm and so on up to
channel 16, which covers the range 101:4 mm to 128 mm. The ratio of adjacent size range
boundaries is always the cube root of 2. For example,
p 4:0 5:04 128
2 ; etc:
3

3:17 4:0 101:4

The resulting lower and upper sizes for the channels are shown in columns 2 and 3 of
Table 1W5.1.

Table 1W5.1 Size distribution data associated with Worked Example 1.5

1 2 3 4 5 6 7 8
Channel Lower Upper Cumulative Fv 1=x Cumulative Fs 9
number size of size of per cent area under Cumulative
range range undersize Fv versus area under
mm mm 1=x Fs versus x

1 3.17 4.00 0 0 0.2500 0.0000 0.0000 0.0000

2 4.00 5.04 0.5 0.005 0.1984 0.0011 0.0403 0.1823
3 5.04 6.35 1 0.01 0.1575 0.0020 0.0723 0.3646
4 6.35 8.00 1.6 0.016 0.1250 0.0029 0.1028 0.5834
5 8.00 10.08 2.6 0.026 0.0992 0.0040 0.1432 0.9480
6 10.08 12.70 3.8 0.038 0.0787 0.0050 0.1816 1.3855
7 12.70 16.00 5.7 0.057 0.0625 0.0064 0.2299 2.0782
8 16.00 20.16 8.7 0.087 0.0496 0.0081 0.2904 3.1720
9 20.16 25.40 14.3 0.143 0.0394 0.0106 0.3800 5.2138
10 25.40 32.00 22.2 0.222 0.0313 0.0134 0.4804 8.0942
11 32.00 40.32 33.8 0.338 0.0248 0.0166 0.5973 12.3236
12 40.32 50.80 51.3 0.513 0.0197 0.0205 0.7374 18.7041
13 50.80 64.00 72 0.72 0.0156 0.0242 0.8689 26.2514
14 64.00 80.63 90.9 0.909 0.0124 0.0268 0.9642 33.1424
15 80.63 101.59 99.3 0.993 0.0098 0.0277 0.9978 36.2051
16 101.59 128.00 100 1 0.0078 0.0278 1.0000 36.4603
Note: Based on arithmetic means of size ranges.
22 PARTICLE SIZE ANALYSIS

Figure 1W5.1 Cumulative volume distribution

(a) The cumulative undersize distribution is shown numerically in column 5 of

Table 1W5.1 and graphically in Figure 1W5.1. By inspection, we see that the median
size is 50 mm (b), i.e. 50% by volume of the particles is less than 50 mm.

(b) The surface distribution is related to the volume distribution by the expression:
fv x ks
fs x  from Equations1:1 and 1:2
x kv

Recalling that fx dF=dx and integrating between 0 and x:

x x
ks 1 dFv dFs
dx dx
kv 0 x dx 0 dx

or x x
ks 1
dFv dFs Fs x
kv 0x 0

(assuming particle shape is invariant with size so that ks =kv is constant).

So the surface distribution can be found from the area under a plot of 1=x versus Fv
x1
multiplied by the factor ks =kv (which is found by noting that x0 dFs 1).

Column 7 of Table 1W5.1 shows the area under 1=x versus Fv . The factor ks =kv is
therefore equal to 0.0278. Dividing the values of column 7 by 0.0278 gives the surface
distribution Fs shown in column 8. The surface distribution is shown graphically in
Figure 1W5.2. The shape of the surface distribution is quite different from that of the
volume distribution; the smaller particles make up a high proportion of the total surface.
The median of the surface distribution is around 35 mm, i.e. particles under 35 mm
contribute 50% of the total surface area.

(c) The harmonic mean of the volume distribution is given by:

1

1 1
dFv
x
hV 0 x
WORKED EXAMPLES 23

Figure 1W5.2 Cumulative surface distribution

This can be calculated graphically from a plot of Fv versus 1=x or numerically from the
tabulated data in column 7 of Table 1W5.1. Hence,
1

1 1
dFv 0:0278
x
hV 0 x

and so, x
hV 36 mm.

We recall that the harmonic mean of the volume distribution is equivalent to the
surface-volume mean of the population.

(d) The arithmetic mean of the surface distribution is given by:

1
x
aS x dFs
0

This may be calculated graphically from our plot of Fs versus x (Figure 1W5.2) or
numerically using the data in Table 1W5.1. This area calculation as shown in Column 9
of the table shows the cumulative area under a plot of Fs versus x and so the last figure
in this column is equivalent to the above integral.
Thus:
x
as 36:4 mm

We may recall that the arithmetic mean of the surface distribution is also equivalent to
the surface-volume mean of the population. This value compares well with the value
obtained in (c) above.

WORKED EXAMPLE 1.6

Consider a cuboid particle 5:00  3:00  1:00 mm. Calculate for this particle the follow-
ing diameters:

(a) the volume diameter (the diameter of a sphere having the same volume as the
particle);
24 PARTICLE SIZE ANALYSIS

(b) the surface diameter (the diameter of a sphere having the same surface area as the
particle);

(c) the surface-volume diameter (the diameter of a sphere having the same external
surface to volume ratio as the particle);

(d) the sieve diameter (the width of the minimum aperture through which the particle
will pass);

(e) the projected area diameters (the diameter of a circle having the same area as the
projected area of the particle resting in a stable position).

Solution

(a) Volume of the particle 5  3  1 15 mm3

px3v
Volume of a sphere
6 r
3 15  6
Thus volume diameter, xv 3:06 mm
p
(b) Surface area of the particle 2  5  3 2  1  3 2  1  5 46 mm2

Surface area of sphere px2s

r
46
Therefore, surface diameter, xs 3:83 mm
p
(c) Ratio of surface to volume of the particle 46=15 3:0667
6
For a sphere, surface to volume ratio
xsv
6
Therefore, xsv 1:96 mm
3:0667
(d) The smallest square aperture through which this particle will pass is 3 mm. Hence,
the sieve diameter, xp 3 mm

(e) This particle has three projected areas in stable positions:

area 1 3 mm2 ; area 2 5 mm2 ; area 3 15 mm2

px2
area of circle
4
hence, projected area diameters:

projected area diameter 1 1:95 mm;

projected area diameter 2 2:52 mm;

projected area diameter 3 4:37 mm:

EXERCISES 25

TEST YOURSELF

1.1 Define the following equivalent sphere diameters: equivalent volume diameter, equiva-
lent surface diameter, equivalent surface-volume diameter. Determine the values of each
one for a cuboid of dimensions 2 mm  3 mm  6 mm.

1.2 List three types of distribution that might be used in expressing the range of particle
sizes contained in a given sample.

1.3 If we measure a number distribution and wish to convert it to a surface distribution,

what assumptions have to be made?

1.4 Write down the mathematical expression defining (a) the quadratic mean and (b) the
harmonic mean.

1.5 For a give particle size distribution, the mode, the arithmetic mean, the harmonic
mean and the quadratic mean all have quite different numberical values. How do we
decide which mean is appropriate for describing the powders behaviour in a given
process?

1.6 What are the golden rules of sampling?

1.7 When using the sedimentation method for determination of particle size distribution,
what assumptions are made?

1.8 In the electrozone sensing method of size analysis, (a) what equivalent sphere particle
diameter is measured and (b) what type of distribution is reported?

EXERCISES

1.1 For a regular cuboid particle of dimensions 1:00  2:00  6:00 mm, calculate the
following diameters:

(a) the equivalent volume sphere diameter;

(b) the equivalent surface sphere diameter;

(c) the surface-volume diameter (the diameter of a sphere having the same external
surface to volume ratio as the particle);

(d) the sieve diameter (the width of the minimum aperture through which the particle will
pass);

(e) the projected area diameters (the diameter of a circle having the same area as the
projected area of the particle resting in a stable position).

[Answer: (a) 2.84 mm; (b) 3.57 mm; (c) 1.80 mm; (d) 2.00 mm; (e) 2.76 mm, 1.60 mm and
3.91 mm.]
26 PARTICLE SIZE ANALYSIS

1.2 Repeat Exercise 1.1 for a regular cylinder of diameter 0.100 mm and length 1.00 mm.

[Answer: (a) 0.247 mm; (b) 0.324 mm; (c) 0.142 mm; (d) 0.10 mm; (e) 0.10 mm (unlikely to
be stable in this position) and 0.357 mm.]

1.3 Repeat Exercise 1.1 for a disc-shaped particle of diameter 2.00 mm and length 0.500 mm.

[Answer: (a) 1.44 mm; (b) 1.73 mm; (c) 1.00 mm; (d) 2.00 mm; (e) 2.00 mm and 1.13 mm
(unlikely to be stable in this position).]

1.4 1.28 g of a powder of particle density 2500 kg m3 are charged into the cell of an
apparatus for measurement of particle size and specific surface area by permeametry. The
cylindrical cell has a diameter
 of 1.14 cm and the powder forms a bed of depth 1 cm. Dry
air of density 1:2 kg m3 and viscosity 18:4  106 Pa s flows at a rate of 36 cm3 min
through the powder (in a direction parallel to the axis of the cylindrical cell) and
producing a pressure difference of 100 mm of water across the bed. Determine the
surface-volume mean diameter and the specific surface of the powder sample.
(Answer: 20 mm; 120 m2 =kg:)

1.5 1.1 g of a powder of particle density 1800 kg=m3 are charged into the cell of an
apparatus for measurement of particle size and specific surface area by permeametry. The
cylindrical cell has a diameter of 1.14 cm and the powder forms a bed of depth l cm. Dry
air of density 1:2 kg=m3 and viscosity 18:4  106 Pa s flows through the powder (in a
direction parallel to the axis of the cylindrical cell). The measured variation in pressure
difference across the bed with changing air flow rate is given below:

Air flow cm3 =min 20 30 40 50 60

Pressure difference across the bed (mm of water) 56 82 112 136 167

Determine the surface-volume mean diameter and the specific surface of the powder
sample.
(Answer: 33 mm; 100 m2 =kg:)

1.6 Estimate the (a) arithmetic mean, (b) quadratic mean, (c) cubic mean, (d) geometric
mean and (e) harmonic mean of the following distribution.

Size 2 2.8 4 5.6 8 11.2 16 22.4 32 44.8 64 89.6

cumulative 0.1 0.5 2.7 9.6 23 47.9 73.8 89.8 97.1 99.2 99.8 100
%
undersize

[Answer: (a) 13.6; (b) 16.1; (c) 19.3; (d) 11.5; (e) 9.8.]
1.7 The following volume distribution was derived from a sieve analysis

Size mm 3745 4553 5363 6375 7590 90106 106126 126150 150180 180212
Volume % 0.4 3.1 11 21.8 27.3 22 10.1 3.9 0.4 0
in range
EXERCISES 27

(a) Estimate the arithmetic mean of the volume distribution.

From the volume distribution derive the number distribution and the surface distribu-
tion, giving assumptions made.
Estimate:

(b) the mode of the surface distribution;

(c) the harmonic mean of the surface distribution.

Show that the arithmetic mean of the surface distribution conserves the surface to volume
ratio of the population of particles.

[Answer: (a) 86 mm; (b) 70 mm; (c) 76 mm.]