Chapter 4

Particle Size Analysis

4.1 INTRODUCTION Several equivalent diameters are commonly

encountered. For example, the Stokes’ diameter is
Size analysis of the various products of a concentrator is measured by sedimentation and elutriation techniques,
of importance in determining the quality of grinding, and the projected area diameter is measured microscopi-
in establishing the degree of liberation of the values from cally, and the sieve-aperture diameter is measured by
the gangue at various particle sizes. In the separation stage, means of sieving. The latter refers to the diameter of
size analysis of the products is used to determine the opti- a sphere equal to the width of the aperture through
mum size of the feed to the process for maximum effi- which the particle just passes. If the particles under
ciency and to determine the size range at which any losses test are not true spheres, and they rarely are in prac-
are occurring in the plant, so that they may be reduced. tice, this equivalent diameter refers only to their sec-
It is essential, therefore, that methods of size analysis ond largest dimension.
be accurate and reliable, as important changes in plant Recorded data from any size analysis should, where
operation may be made based on the results of these tests. possible, be accompanied by some remarks which indi-
Since it is often the case that only relatively small cate the approximate shape of the particles. Fractal anal-
amounts of material are used in sizing tests, it is essential ysis can be applied to particle shape. Descriptions such
that the sample is representative of the bulk material and as “granular” or “acicular” are usually quite adequate to
the same care should be taken over sampling for size convey the approximate shape of the particle in
analysis as is for assaying (Chapter 3). question.
Some of these terms are given below:

4.2 PARTICLE SIZE AND SHAPE Acicular Needle-shaped

Angular Sharp-edged or having roughly polyhedral shape
The primary function of precision particle analysis is to
obtain quantitative data about the size and size distribu- Crystalline Freely developed in a fluid medium of geometric
tion of particles in the material (Bernhardt, 1994; Allen, shape
1997). However, exact size of an irregular particle cannot Dendritic Having a branched crystalline shape
be measured. The terms “length,” “breadth,” “thickness,”
Fibrous Regular or irregularly thread-like
or “diameter” have little meaning because so many differ-
ent values of these quantities can be determined. The size Flaky Plate-like
of a spherical particle is uniquely defined by its diameter. Granular Having approximately an equidimensional
For a cube, the length along one edge is characteristic, irregular shape
and for other regular shapes there are equally appropriate Irregular Lacking any symmetry
dimensions.
Modular Having rounded, irregular shape
For irregular particles, it is desirable to quote the size
of a particle in terms of a single quantity, and the expres- Spherical Global shape
sion most often used is the “equivalent diameter.” This
refers to the diameter of a sphere that would behave in There is a wide range of instrumental and other meth-
the same manner as the particle when submitted to some ods of particle size analysis available. A short list of some
specified operation. of the more common methods is given in Table 4.1,
The assigned equivalent diameter usually depends on together with their effective size ranges (these can vary
the method of measurement, hence the particle-sizing greatly depending on the technology used), whether they
technique should, when possible, duplicate the type of can be used wet or dry and whether fractionated samples
process one wishes to control. are available for later analysis.
Wills’ Mineral Processing Technology.
© 2016 Elsevier Ltd. All rights reserved. 91
92 Wills’ Mineral Processing Technology

The process of sieving may be divided into two stages.

TABLE 4.1 Some Methods of Particle Size Analysis The first step is the elimination of particles considerably
smaller than the screen apertures, which should occur
Method Wet/ Fractionated Approx.
fairly rapidly. The second step is the separation of the so-
Dry Sample? Useful Size
called “near-size” particles, which is a gradual process
Range (μm)a
rarely reaching completion. Both stages require the sieve
Test sieving Both Yes 5
100,000 to be manipulated in such a way that all particles have
Laser Both No 0.1
2,500 opportunities to pass through the apertures, and so that
diffraction any particles that blind an aperture may be removed from
Optical Dry No 0.2
50 it. Ideally, each particle should be presented individually
microscopy to an aperture, as is permitted for the largest aperture
Electron Dry No 0.005
100 sizes, but for most sizes this is impractical.
microscopy The effectiveness of a sieving test depends on the
amount of material put on the sieve (the “charge”) and
Elutriation Wet Yes 5
45
(cyclosizer) the type of movement imparted to the sieve.
A comprehensive account of sampling techniques for
Sedimentation Wet Yes 1
40
(gravity)
sieving is given in BS 1017-1 (Anon., 1989a). Basically,
if the charge is too large, the bed of material will be too
Sedimentation Wet Yes 0.05
5 deep to allow each particle a chance to meet an aperture
(centrifuge)
in the most favorable position for sieving in a reasonable
a
A micrometer (micron) (μm) is 1026 m. time. The charge, therefore, is limited by a requirement
for the maximum amount of material retained at the end
of sieving appropriate to the aperture size. On the other
hand, the sample must contain enough particles to be rep-
resentative of the bulk, so a minimum size of sample is
4.3 SIEVE ANALYSIS specified. In some cases, the sample will have to be sub-
Test sieving is the most widely used method for particle divided into a number of charges if the requirements for
size analysis. It covers a very wide range of particle sizes, preventing overloading of the sieves are to be satisfied. In
which is important in industrial applications. So common some cases, air is blown through the sieves to decrease
is test sieving as a method of size analysis that particles testing time and decrease the amount of blinding occur-
finer than about 75 μm are often referred to as being in ring, a technique referred to as air jet sieving.
the “sub-sieve” range, although modern sieving methods
allow sizing to be carried out down to about 5 μm.
Sieve (or screen) analysis is one of the oldest methods
4.3.1 Test Sieves
of size analysis and is accomplished by passing a known Test sieves are designated by the nominal aperture size,
weight of sample material through successively finer which is the nominal central separation of opposite sides
sieves and weighing the amount collected on each sieve to of a square aperture or the nominal diameter of a round
determine the percentage weight in each size fraction. aperture. A variety of sieve aperture ranges are used, the
Sieving is carried out with wet or dry materials and the most popular being the following: the German Standard,
sieves are usually agitated to expose all the particles to the DIN 4188; ASTM standard, E11; the American Tyler
openings. Sieving, when applied to irregularly shaped par- series; the French series, AFNOR; and the British
ticles, is complicated by the fact that a particle with a size Standard, BS 1796.
near that of the nominal aperture of the test sieve may pass Woven-wire sieves were originally designated by a
only when presented in a favorable orientation. As there is mesh number, which referred to the number of wires per
inevitably a variation in the size of sieve apertures due to inch, which is the same as the number of square apertures
irregularity of weaving, prolonged sieving will cause the per square inch. This has the serious disadvantage that the
larger apertures to exert an unduly large effect on the sieve same mesh number on the various standard ranges corre-
analysis. Given time, every particle small enough could sponds to different aperture sizes depending on the thick-
find its way through a very few such holes. The procedure ness of wire used in the woven-wire cloth. Sieves are now
is also complicated in many cases by the presence of designated by aperture size, which gives the user directly
“near-size” particles which cause “blinding,” or obstruc- the information needed.
tion of the sieve apertures, and reduce the effective area of Since some workers and the older literature still refer
the sieving medium. Blinding is most serious with test to sieve sizes in terms of mesh number, Table 4.2 lists
sieves of very small aperture size. mesh numbers for the British Standards series against
 Particle Size Analysis Chapter | 4 93

TABLE 4.2 BSS 1796 Wire-mesh Sieves

Mesh Nominal Mesh Nominal

Number Aperture Size Number Aperture Size
(μm) (μm)
3 5,600 36 425
3.5 4,750 44 355
4 4,000 52 300
5 3,350 60 250
6 2,800 72 212 FIGURE 4.1 Weaves of wire cloth: (a) plain weave, and (b) twilled
weave.
7 2,360 85 180
8 2,000 100 150
10 1,700 120 125
12 1,400 150 106
14 1,180 170 90
16 1,000 200 75
18 850 240 63 FIGURE 4.2 Cross-section of a micro-plate aperture.

22 710 300 53
25 600 350 45 For aperture sizes above about 1 mm, perforated plate
30 500 400 38
sieves are often used, with round or square holes
(Figure 4.3). Square holes are arranged in line with the
center points at the vertices of squares, while round holes
are arranged with the centers at the apices of equilateral
triangles (Anon., 2000b).
nominal aperture size. A fuller comparison of several
standards is given in Napier-Munn et al. (1996).
Wire-cloth screens are woven to produce nominally
4.3.2 Choice of Sieve Sizes
uniform square apertures within required tolerances
(Anon., 2000a). Wire cloth in sieves with a nominal aper- In each of the standard series, the apertures of consecutive
ture of 75 μm and greater are plain woven, while those in sieves bear a constant relationship to each other. It has
cloths with apertures below 63 μm may be twilled long been realized that a useful sieve scale is one in
(Figure 4.1). which the ratio of the aperture widths of adjacent sieves
Standard test sieves are not available with aperture is the square root of 2 (O2 5 1.414). The advantage of
sizes smaller than about 20 μm. Micromesh sieves are such a scale is that the aperture areas double at each
available in aperture sizes from 2 μm to 150 μm, and are sieve, facilitating graphical presentation of results with
made by electroforming nickel in square and circular particle size on a log scale.
mesh. Another popular type is the “micro-plate sieve,” Most modern sieve series are based on a fourth root of
which is fabricated by electroetching a nickel plate. The 2 ratio (4O2 5 1.189) or, on the metric scale, a tenth root
apertures are in the form of truncated cones with the small of 10 (10O10 5 1.259), which makes possible much closer
circle uppermost (Figure 4.2). This reduces blinding (a sizing of particles.
particle passing the upper opening falls away unhindered) For most size analyses it is usually impracticable and
but also reduces the percentage open area, that is, the per- unnecessary to use all the sieves in a particular series. For
centage of the total area of the sieving medium occupied most purposes, alternative sieves, that is, a O2 series, are
by the apertures. quite adequate. However, over certain size pffiffiffi ranges of par-
Micro-sieves are used for wet or dry sieving where ticular interest, or for accurate work the 4 2 series may be
accuracy is required in particle size analysis down to the used. Intermediate sieves should never be chosen at ran-
very fine size range (Finch and Leroux, 1982). The toler- dom, as the data obtained will be difficult to interpret.
ances in these sieves are much better than those for In general, the sieve range should be chosen such that
woven-wire sieves, the aperture being guaranteed to no more than about 5% of the sample is retained on the
within 2 μm of nominal size. coarsest sieve, or passes the finest sieve. (The latter
94 Wills’ Mineral Processing Technology

FIGURE 4.3 Arrangement of square and round holes in perforated plate sieves.

guideline is often difficult to meet with the finer grinds

experienced today.) These limits of course may be low-
ered for more accurate work.

4.3.3 Testing Methods

The general procedures for test sieving are comprehen-
sively covered in BS 1796 (Anon., 1989b). Machine siev-
ing is almost universally used, as hand sieving is tedious,
and its accuracy and precision depends to a large extent
on the operator. FIGURE 4.4 No. 10 USA standard test sieve (Courtesy Haver Tyler
Sieves can be procured in a range of diameters, Corporation).
depending on the particle size and mass of material to be
sieved. A common diameter for laboratory sieves is
200 mm (Figure 4.4).
The sieves chosen for the test are arranged in a stack, or
nest, with the coarsest sieve on the top and the finest at the
bottom. A tight-fitting pan or receiver is placed below the
bottom sieve to receive the final undersize, and a lid is placed
on top of the coarsest sieve to prevent escape of the sample.
The material to be tested is placed in the uppermost,
coarsest sieve, and the nest is then placed in a sieve
shaker, which vibrates the material in a vertical plane
(Figure 4.5), and, on some models, a horizontal plane.
The duration of screening can be controlled by an auto-
matic timer. During the shaking, the undersize material
falls through successive sieves until it is retained on a
sieve having apertures which are slightly smaller than the
diameter of the particles. In this way, the sample is sepa-
rated into size fractions.
After the required time, the nest is taken apart and the
amount of material retained on each sieve is weighed.
Most of the near-mesh particles, which block the open-
ings, can be removed by inverting the sieve and tapping
the frame gently. Failing this, the underside of the gauze FIGURE 4.5 W.S. TYLERs RO-TAPs test sieve shaker (Courtesy
may be brushed gently with a soft brass wire or nylon Haver Tyler Corporation, W.S. TYLERs and RO-TAPs are registered
brush. Blinding becomes more of a problem the finer the trademarks of Haver Tyler Corporation).
 Particle Size Analysis Chapter | 4 95

TABLE 4.3 Results of Typical Sieve Test

(1) (2) (3) (4) (5) (6)

Sieve Size Range (μm) Sieve Fraction Nominal Aperture Size (μm) Cumulative (%)

wt (g) wt (%) Undersize Oversize

1250 0.02 0.1 250 99.9 0.1
2250 to 1180 1.32 2.9 180 97.0 3.0
2180 to 1125 4.23 9.5 125 87.5 12.5
2125 to 190 9.44 21.2 90 66.3 33.7
290 to 163 13.10 29.4 63 36.9 63.1
263 to 145 11.56 26.0 45 10.9 89.1
245 4.87 10.9

aperture, and brushing, even with a soft hair brush, of The results of a sieving test should always be plotted
sieves finer than about 150 μm aperture tends to distort graphically to assess their full significance (Napier-Munn
the individual meshes. et al., 1996).
Wet sieving can be used on material already in the There are many ways of recording the results of sieve
form of slurry, or it may be necessary for powders which analysis, the most common being that of plotting cumula-
form aggregates when dry-sieved. A full description of tive undersize (or oversize) against particle size, that is,
the techniques is given in BS 1796 (Anon., 1989b). the screen aperture. Although arithmetic graph paper can
Water is the liquid most frequently used in wet sieving, be used, it suffers from the disadvantage that points in the
although for materials which are water-repellent, such as region of the finer aperture sizes become congested. A
coal or some sulfide ores, a wetting agent may be necessary. semi-logarithmic plot avoids this, with a linear ordinate
The test sample may be washed down through a nest for percentage oversize or undersize and a logarithmic
of sieves. At the completion of the test the sieves, abscissa for particle size. Figure 4.6 shows graphically the
together with the retained oversize material, are dried at a results of the sieve test tabulated in Table 4.3.
suitable low temperature and weighed. It is not necessary to plot both cumulative oversize
(or coarser than) and undersize (finer than) curves, as
they are mirror images of each other. A valuable quan-
4.3.4 Presentation of Results tity that can be determined from such curves is the
“median size” of the sample. This refers to the mid-
There are several ways in which the results of a sieve test
point in the size distribution, X50, where 50% of the par-
can be tabulated. The three most convenient methods are
ticles are smaller than this size and 50% are larger, also
shown in Table 4.3 (Anon., 1989b). Table 4.3 shows for
called the 50% passing size.
the numbered columns:
Size analysis is important in assessing the performance
1. The sieve size ranges used in the test. of grinding circuits. The product size is usually quoted
2. The weight of material in each size range. For exam- in terms of one point on the cumulative undersize curve,
ple, 1.32 g of material passed through the 250 μm this often being the 80% passing size, X80. On Figure 4.6
sieve, but was retained on the 180 μm sieve: the mate- the X80 is about 110 μm. In a grinding circuit it is com-
rial therefore is in the size range 2250 to 1180 μm. mon to refer to the 80% passing size of the feed and
3. The weight of material in each size range expressed as product as F80 and P80, respectively, and to T80 as the
a percentage of the total weight. transfer size between grinding units (e.g., SAG mill
4. The nominal aperture sizes of the sieves used in the test. product going to a ball mill). Although the X80 does not
5. The cumulative percentage of material passing show the overall size distribution of the material, it does
through the sieves. For example, 87.5% of the material facilitate routine control of the grinding circuit. For
is less than 125 μm in size. instance, if the target size is 80% 2250 μm (80% finer
6. The cumulative percentage of material retained on the than 250 μm), then for routine control the operator needs
sieves. only screen a fraction of the mill product at one size. If
96 Wills’ Mineral Processing Technology

FIGURE 4.6 Sieve analysis presented as semi-log plot (Table 4.3). FIGURE 4.7 Example of Gates-Gaudin-Schuhmann type plot for the
set of laboratory grind times in Appendix IV.

it is found that, say, 70% of the sample is 2250 μm, plant). An example G-G-S plot is shown in Figure 4.7 for
then the product is too coarse, and control steps to increased grinding time in a laboratory ball mill (data
remedy this can be made. given in Appendix IV). The figure shows a series of
Many curves of cumulative oversize or undersize roughly parallel (or self-similar) linear sections of the size
against particle size are S-shaped, leading to congested distributions. This result, showing a consistent behavior,
plots at the extremities of the graph. More than a dozen is quite common in grinding (and other) studies (but see
methods of plotting in order to proportion the ordinate are Chapter 5 for exceptions) and supports the use of a single
known. The two most common methods, which are often metric to describe the distribution, for example, the 80%
applied to comminution studies where non-uniform size passing size, X80 (Austin et al., 1984).
distributions are obtained, are the Gates-Gaudin- Self-similar distributions such as in Figure 4.7
Schuhmann (Schuhmann, 1940) and the Rosin
Rammler become a unique distribution by plotting against a
(Rosin and Rammler, 1933) methods. Both methods are reduced particle size, typically X/X50 where X50 is the
derived from attempts to represent particle size distribu- 50% passing size or median size. If it is known that data
tion curves by means of equations. This results in scales obtained from the material usually yield a linear plot on
which, relative to a linear scale, are expanded in some log
log axes, then the burden of routine analysis can be
regions and contracted in others. greatly relieved, as relatively few sieves will be needed
In the Gates-Gaudin-Schuhmann (G-G-S) method, to check essential features of the size distribution. Self-
cumulative undersize data are plotted against sieve aper- similarity can be quickly identified, as a plot of, say,
ture on log
log axes. This frequently leads to a linear % 2 75 μm versus % 1150 μm will yield a single curve
trend from which data can be interpolated easily. The lin- if this property holds.
ear trend is fitted to the following: Plotting on a log
log scale considerably expands
the region below 50% in the cumulative undersize
X
logP 5 α log 1 log100 (4.1) curve, especially that below 25%. It does, however,
K
severely contract the region above 50%, and especially
which can be written as: above 75%, which is a disadvantage of the method
 α (Figure 4.8).
X The Rosin
Rammler method is often used for repre-
P 5 100 (4.2)
K senting the results of sieve analyses performed on mate-
rial which has been ground in ball mills. Such products
where P is the cumulative undersize (or passing) in per- have been found to obey the following relationship
cent, X the particle size, K the apparent (theoretical) top (Austin et al., 1984):
size (i.e., when P 5 100%) obtained by extrapolation, and
α a constant sometimes referred to as the distribution 100 2 P 5 100 expðbXÞn (4.3)
coefficient (Austin et al., 1984). This method of plotting
is routinely used in grinding studies (laboratory and where b and n are constants.
 Particle Size Analysis Chapter | 4 97

FIGURE 4.9 Fractional (frequency) distribution and histogram repre-

sentation of screen analysis data.

Allen (1997), it should be used with caution, since taking

logs always apparently reduces scatter, and thus taking
logs twice is not to be recommended.
Although cumulative size curves are used almost
exclusively, the particle size distribution curve itself is
sometimes more informative. Ideally, this is derived by
differentiating the cumulative undersize curve and plot-
ting the gradient of the curve obtained against particle
FIGURE 4.8 Comparison of scales.
size. In practice, the size distribution curve is obtained by
plotting the retained fraction of the sieves against size.
This can be as a histogram or as a frequency (fractional)
curve plotted at the “average” size in between two sieve
This can be rewritten as: sizes. For example, material which passes a 250 μm sieve
  but is retained on a 180 μm sieve, may be regarded as
100
log ln 5 log b 1 n log X (4.4) having an arithmetic mean particle size of 215 μm or,
100 2 P
more appropriately since the sieves are in a geometric
Thus, a plot of ln [100/(100
P)] versus X on log
log sequence, a geometric mean (O(250 3 180)) of 212 μm
axes gives a line of slope n. for the purpose of plotting. If the distribution is repre-
In comparison with the log
log method, the sented on a histogram, then the horizontals on the col-
Rosin
Rammler plot expands the cumulative undersize umns of the histogram join the various adjacent sieves
regions below 25% and above 75% (Figure 4.8) and it used in the test. Unless each size increment is of equal
contracts the 30
60% region. It has been shown, how- width, however, the histogram has little value. Figure 4.9
ever, that this contraction is insufficient to cause adverse shows the size distribution of the material in Table 4.3
effects (Harris, 1971). The method is tedious to plot man- represented on a frequency curve and a histogram.
ually unless charts having the axes divided proportionally Fractional curves and histograms are useful and rapid
to log [ln (100/(100
P))] and log X are used. The data ways of visualizing the relative frequency of occurrence
can, however, be easily plotted in a spreadsheet. of the various sizes present in the material. The only
The Gates-Gaudin-Schuhmann plot is often preferred numerical parameter that can be obtained from these
to the Rosin
Rammler method in mineral processing methods is the “mode” of the distribution, that is, the
applications, the latter being more often used in coal- most commonly occurring size.
preparation studies, for which it was originally developed. For assessment of the metal losses in the tailings of a
The two methods have been assessed by Harris (1971), plant, or for preliminary evaluation of ores, assaying must
who suggests that the Rosin
Rammler is the better be carried out on the various screen fractions. It is impor-
method for mineral processing applications. The tant, therefore, that the bulk sample satisfies the minimum
Rosin
Rammler is useful for monitoring grinding opera- sample weight requirement given by Gy’s equation for
tions for highly skewed distributions, but, as noted by the fundamental error (Chapter 3).
98 Wills’ Mineral Processing Technology

TABLE 4.4 Results of Screen Analysis to Evaluate the Suitability for Treatment by Gravity Concentration

(1) Size Range (μm) (2) Weight (%) (3) Assay (% Sn) Distribution Size (μm) Cumulative Cumulative
(% Sn) oversize (Wt %) distribution (% Sn)
1422 9.7 0.02 0.9 422 9.7 0.9
2422 1 300 4.9 0.05 1.2 300 14.6 2.1
2300 1 210 10.3 0.05 2.5 210 24.9 4.6
2210 1 150 23.2 0.06 6.7 150 48.1 11.3
2150 1 124 16.4 0.12 9.5 124 64.5 20.8
2124 1 75 33.6 0.35 56.5 75 98.1 77.3
275 1.9 2.50 22.7
100.0 0.21 100.0

Table 4.4 shows the results of a screen analysis per- Conversion factors between methods will vary with
formed on an alluvial tin deposit for preliminary evaluation sample characteristics and conditions, and with size where
of its suitability for treatment by gravity concentration. the distributions are not self-similar. For spheres, many
Columns 1, 2, and 3 show the results of the sieve test and methods will give essentially the same result (Napier-
assays, which are evaluated in the other columns. It can be Munn, 1985), but for irregular particles this is not so.
seen that the calculated overall assay for the material is Some approximate factors for a given characteristic size
0.21% Sn, but that the bulk of the tin is present within the (e.g., X80) are given below (Austin and Shah, 1983;
finer fractions. The results show that, for instance, if the Napier-Munn, 1985; Anon., 1989b)—these should be
material was initially screened at 210 μm and the coarse used with caution:
fraction discarded, then the bulk required for further proces-
sing would be reduced by 24.9%, with a loss of only 4.6% Conversion Multiplying Factor
of the Sn. This may be acceptable if mineralogical analysis Sieve size to Stokes’ diameter 0.94
shows that the tin is finely disseminated in this coarse frac- (sedimentation, elutriation)
tion, which would necessitate extensive grinding to give
Sieve size to projected area diameter 1.4
reasonable liberation. Heavy liquid analysis (Chapter 11) on (microscopy)
the 2210 μm fraction would determine the theoretical (i.e.,
Sieve size to laser diffraction 1.5
liberation-limited, Chapter 1) grades and recoveries possi-
ble, but the screen analysis results also show that much of Square mesh sieves to round hole 1.2
the tin (22.7%) is present in the 275 μm fraction, which sieves
constitutes only 1.9% of the total bulk of the material. This
indicates that there may be difficulty in processing this 4.4.1 Stokes’ Equivalent Diameter
material, as gravity separation techniques are not very effi-
cient at such fine sizes (Chapters 1 and 10). In sedimentation techniques, the material to be sized is
dispersed in a fluid and allowed to settle under carefully
controlled conditions; in elutriation techniques, samples
4.4 SUB-SIEVE TECHNIQUES are sized by allowing the dispersed material to settle
Sieving is rarely carried out on a routine basis below against a rising fluid velocity. Both techniques separate
38 μm; below this size the operation is referred to as sub- the particles on the basis of resistance to motion in a
sieving. The most widely used methods are sedimentation, fluid. This resistance to motion determines the terminal
elutriation, microscopy, and laser diffraction, although velocity which the particle attains as it is allowed to fall
other techniques are available. in a fluid under the influence of gravity.
There are many concepts in use for designating parti- For particles within the sub-sieve range, the terminal
cle size within the sub-sieve range, and it is important to velocity is given by the equation derived by Stokes (1891):
be aware of them, particularly when combining size dis-
tributions determined by different methods. It is prefera-
d 2 gðρs 2 ρl Þ
ble to cover the range of a single distribution with a v5 (4.5)
single method, but this is not always possible. 18 η
 Particle Size Analysis Chapter | 4 99

where v is the terminal velocity of the particle (m s21),

d the particle diameter (m), g the acceleration due to
gravity (m s22), ρs the particle density (kg m23), ρf the
fluid density (kg m23), and η the fluid viscosity
(N s m22); (η 5 0.001 N s m22 for water at 20 C).
Stokes’ law is derived for spherical particles; non-
spherical particles will also attain a terminal velocity,
but this velocity will be influenced by the shape of the
particles. Nevertheless, this velocity can be substituted
in the Stokes’ equation to give a value of d, which can
be used to characterize the particle. This value of d is
referred to as the “Stokes’ equivalent spherical diame-
ter” (or simply “Stokes’ diameter” or “sedimentation
diameter”). FIGURE 4.10 Beaker decantation.
Stokes’ law is only valid in the region of laminar
flow (Chapter 9), which sets an upper size limit to the
particles that can be tested by sedimentation and elutria- The time required for a 10 μm particle to settle from
tion methods in a given liquid. The limit is determined by the water level to the bottom of the syphon tube, distance
the particle Reynolds number, a dimensionless quantity h, is calculated (t 5 h/v). The pulp is gently stirred to
defined by: disperse the particles through the whole volume of water
and then it is allowed to stand for the calculated time.
ν dρf The water above the end of the tube is syphoned off and
Re 5 (4.6)
η all particles in this water are assumed to be smaller than
10 μm diameter (Figure 4.10). However, a fraction of the
The Reynolds number should not exceed 0.2 if the 210 μm material, which commenced settling from vari-
error in using Stokes’ law is not to exceed 5% (Anon., ous levels below the water level, will also be present in
2001a). In general, Stokes’ law will hold for all particles the material below the syphon level. In order to recover
below 40 μm dispersed in water; particles above this size these particles, the pulp remaining must be diluted with
should be removed by sieving beforehand. The lower water to the original level, and the procedure repeated
limit may be taken as 1 μm, below which the settling until the decant liquor is essentially clear. In theory, this
times are too long, and also the effects of Brownian requires an infinite number of decantations, but in prac-
motion and unintentional disturbances, such as those tice at least five treatments are usually sufficient, depend-
caused by convection currents, are far more likely to pro- ing on the accuracy required. The settled material can be
duce serious errors. treated in a similar manner at larger separating sizes, that
is, at shorter decanting times, until a target number of size
fractions is obtained.
The method is simple and cheap, and has an advantage
4.4.2 Sedimentation Methods
over many other sub-sieve techniques in that it produces a
Sedimentation methods are based on the measurement true fractional size analysis, that is, reasonable quantities
of the rate of settling of the powder particles uniformly of material in specific size ranges are collected, which
dispersed in a fluid and the principle is well illustrated can be analyzed chemically and mineralogically.
by the common laboratory method of “beaker This method is, however, extremely tedious, as long
decantation.” settling times are required for very fine particles, and sep-
The material under test is uniformly dispersed in low arate tests must be performed for each particle size. For
concentration in water contained in a beaker or similar instance, a 25 μm particle of quartz has a settling velocity
parallel-sided vessel. A wetting agent may need to be of 0.056 cm s21, and therefore takes about 31/2 min to set-
added to ensure complete dispersion of the particles. A tle 12 cm, a typical immersion depth for the syphon tube.
syphon tube is immersed into the water to a depth of h Five separate tests to ensure a reasonably clear decant
below the water level, corresponding to about 90% of the therefore require a total settling time of about 18 min. A
liquid depth L. 5 μm particle, however, has a settling velocity of
The terminal velocity v is calculated from Stokes’ law 0.0022 cm s21, and therefore takes about 11/2 h to settle
for the various sizes of particle in the material, say 35, 12 cm. The total time for evaluation of such material is
25, 15, and 10 μm. For a distribution, it is usual to fix ds thus about 8 h. A complete analysis may therefore take an
for the particles that are most abundant in the sample. operator several days.
100 Wills’ Mineral Processing Technology

TABLE 4.5 Number of Decantations Required for

Required Efficiency of Removal of fine Particles

Relative Particle 0.95 0.9 0.8 0.5 1

Size (d1/d)
Number of 90% 25 12 6 2 1
decantations Efficiency
95% 33 16 8 3 1
Efficiency
99% 50 25 12 4 2
Efficiency

Another problem is the large quantity of water, which

dilutes the undersize material, due to repeated
decantation.
In the system shown in Figure 4.10, after time t, all
particles larger than size d have fallen to a depth below
the level h. All particles of a size d1, where d1 , d, will
have fallen below a level h1 below the water level, where
h1 , h. The efficiency of removal of particles of size d1
into the decant is thus:
h 2 h1
FIGURE 4.11 Andreasen pipette.
L
since at time t 5 0 the particles were uniformly distributed
over the whole volume of liquid, corresponding to depth sizes of particle expressed relative to d, the separating
L, and the fraction removed into the decant is the volume size, where the value of a 5 0.9. It can be shown
above the syphon level, h
h1. (Heywood, 1953) that the value of a has relatively little
Now, since t 5 h/v, and v ~ d2, effect; therefore there is nothing to be gained by attempt-
h h1 ing to remove the suspension adjacent to the settled parti-
5 2 cles, thus risking disturbance and re-entrainment.
d2 d1
Table 4.5 shows that a large number of decantations
Therefore, the efficiency of removal of particles of are necessary for effective removal of particles close to
size d1 the separation size, but that relatively small particles are
quickly eliminated. For most purposes, unless very nar-
h 2 hðd1 =dÞ2 h½1 2 ðd1 =dÞ2   
5 5 5 a 1 2 ðd1 =dÞ2 5 E row size ranges are required, no more than about twelve
L L decantations are necessary for the test.
where a 5 h/L. A much quicker and less-tedious method of sedimenta-
If a second decantation step is performed, the amount tion analysis is the Andreasen pipette technique (Anon.,
of minus d1 material in the dispersed suspension is 1
E, 2001b).
and the efficiency of removal of minus d1 particles after The apparatus (Figure 4.11) consists of a half-liter
two decantations is thus: graduated cylindrical flask and a pipette connected to a
10 mL reservoir by means of a two-way stop-cock. The
E 1 ð1 2 EÞE 5 2E 2 E2 5 1 2 ½12E2 tip of the pipette is in the plane of the zero mark when
In general, for n decantation steps the efficiency of the ground glass stopper is properly seated.
removal of particles of size d1, at a separation size of d, is: A 3 to 5% suspension of the sample, dispersed in the
sedimentation fluid, usually water, is added to the flask.
1 2 ½12En The pipette is introduced and the suspension agitated by
(4.7)
or efficiency 5 1 2 f1 2 a ½1 2 ðd1 =dÞ2 g inversion. The suspension is then allowed to settle, and at
Table 4.5 shows the number of decantation steps given intervals of time, samples are withdrawn by apply-
required for different efficiencies of removal of various ing suction to the top of the reservoir, manipulating the
 Particle Size Analysis Chapter | 4 101

two-way cock so that the sample is drawn up as far as the

calibration mark on the tube above the 10 mL reservoir.
The cock is then reversed, allowing the sample to drain
into the collecting dish. After each sample is taken, the
new liquid level is noted.
The samples are then dried and weighed, and the
weights compared with the weight of material in the same
volume of the original suspension.
There is a definite particle size, d, corresponding to
each settling distance h and time t, and this represents the
size of the largest particle that can still be present in the
sample. These particle sizes are calculated from Stokes’
law for the various sampling times. The weight of solids
collected, w, compared with the corresponding original
weight, wo (i.e., w/wo) then represents the fraction of the
original material having a particle size smaller than d,
which can be plotted on the size-analysis graph.
The Andreasen pipette method is quicker than beaker
decantation, as samples are taken off successively
throughout the test for increasingly finer particle sizes.
For example, although 5 μm particles of quartz will take FIGURE 4.12 Simple elutriator.
about 21/2 h to settle 20 cm, once this sample is collected,
all the coarser particle-size samples will have been taken,
and so the complete analysis, in terms of settling times, is
only as long as the settling time for the finest particles. All elutriators consist of one or more “sorting col-
The disadvantage of the method is that the samples umns” (Figure 4.12) in which the fluid is rising at a con-
taken are each representative of the particles smaller than stant velocity. Feed particles introduced into the sorting
a particular size, which is not as valuable, for mineralogi- column will be separated into two fractions, according to
cal and chemical analysis, as samples of various size their terminal velocities, calculated from Stokes’ law.
ranges, as are produced by beaker decantation. Those particles having a terminal velocity less than
Sedimentation techniques tend to be tedious, due to that of the velocity of the fluid will report to the overflow,
the long settling times required for fine particles and the while those particles having a greater terminal velocity
time required to dry and weigh the samples. The main dif- than the fluid velocity will sink to the underflow.
ficulty, however, lies in completely dispersing the mate- Elutriation is carried out until there are no visible signs of
rial within the suspending liquid, such that no further classification taking place or the rate of change in
agglomeration of particles occurs. Combinations of weights of the products is negligible.
suitable suspending liquids and dispersing agents for vari- An elutriator based on air is the Haultain infrasizer,
ous materials are given in BS ISO 13317
1 (Anon., which comprises six sorting columns (”cones”) plus a
2001a). dust collector (Price, 1962). More commonly, water is the
Various other techniques have been developed that fluid. This involves the use of much water and dilution of
attempt to speed up testing. Examples of these methods, the undersize fraction, but it can be shown that this is not
which are comprehensively reviewed by Allen (1997), as serious as in beaker decantation. Consider a sorting
include: the photo-sedimentometer, which combines grav- column of depth h, sorting material at a separating size of
itational settling with photo-electric measurement; and the d. If the upward velocity of water flow is v, then (as noted
sedimentation balance, in which the weight of material above) by Stokes’ law, v ~ d2.
settling out onto a balance pan is recorded against time to Particles smaller than the separating size d will move
produce a cumulative sedimentation size analysis. upwards in the water flow at a velocity dependent on their
size. Thus, particles of size d1, where d1 , d, will move
upwards in the sorting column at a velocity v1, where
4.4.3 Elutriation Techniques v1 ~ ðd2 2 d12 Þ.
Elutriation is a process of sizing particles by means of an The time required for a complete volume change in
upward current of fluid, usually water or air. The process the sorting column is h/v, and the time required for parti-
is the reverse of gravity sedimentation, and Stokes’ law cles of size d1 to move from the bottom to the top of the
applies. sorting column is h/v1. Therefore, the number of volume
102 Wills’ Mineral Processing Technology

changes required to remove all particles of size d1 from

the sorting column is:

h=v1 d2 1
5 5 2 5
h=v d 2 d12
1 2 ðd1 =dÞ2

and the number of volume changes required for various

values of d1/d are:

d1/d 0.95 0.9 0.8 0.5 0.1

Number of volume 10.3 5.3 2.8 1.3 1.0
changes required

Comparing these figures with those in Table 4.5, it

can be seen that the number of volume changes required
is far less with elutriation than it is with decantation. It is
also possible to achieve complete separation by elutria-
tion, whereas this can only be achieved in beaker decanta-
tion by an infinite number of volume changes.
Elutriation thus appears more attractive than decantation,
and has certain practical advantages in that the volume
changes need no operator attention. It suffers from the disad-
vantage, however, that the fluid velocity is not constant
FIGURE 4.13 Cyclosizer (Courtesy MARC Technologies Pty Ltd.).
across the sorting column, being a minimum at the walls of
the column, and a maximum at the center. The separation
size is calculated from the mean volume flow, so that some
coarse particles are misplaced in the overflow, and some
fines are misplaced into the underflow. The fractions thus
have a considerable overlap in particle size and are not
sharply separated. Although the decantation method never
attains 100% efficiency of separation, the lack of sharpness
of the division into fractions is much less than that due to
velocity variation in elutriation (Heywood, 1953).
Elutriation is limited at the coarsest end by the validity of
Stokes’ law, but most materials in the sub-sieve range exhibit
laminar flow. At the fine end of the scale, separations
become impracticable below about 10 μm, as the material
tends to agglomerate, or extremely long separating times are
required. Separating times can be considerably decreased by
utilizing centrifugal forces. One of the most widely used
methods of sub-sieve sizing in mineral processing laborato-
ries is the Warman cyclosizer (Finch and Leroux, 1982),
which is extensively used for routine testing and plant control
in the size range 8
50 μm for materials of specific gravity
similar to quartz (s. g. 2.7), and down to 4 μm for particles of
high specific gravity, such as galena (s. g. 7.5).
The cyclosizer unit consists of five cyclones (see
FIGURE 4.14 Flow pattern inside a cyclosizer cyclone unit.
Chapter 9 for a full description of the principle of the
hydrocyclone), arranged in series such that the overflow
of one unit is the feed to the next unit (Figure 4.13). Water is pumped through the units at a controlled rate,
The individual units are inverted in relation to conven- and a weighed sample of solids is introduced ahead of the
tional cyclone arrangements, and at the apex of each, a cyclones.
chamber is situated so that the discharge is effectively The tangential entry into the cyclones induces the liq-
closed (Figure 4.14). uid to spin, resulting in a portion of the liquid, together
 Particle Size Analysis Chapter | 4 103

with the faster-settling particles, reporting to the apex The optical microscope method is applicable to parti-
opening, while the remainder of the liquid, together with cles in the size range 0.8
150 μm, and down to 0.001 μm
the slower settling particles, is discharged through the using electron microscopy.
vortex outlet and into the next cyclone in the series. Basically, all microscopy methods are carried out on
There is a successive decrease in the inlet area and vortex extremely small laboratory samples, which must be col-
outlet diameter of each cyclone in the direction of the lected with great care in order to be representative of the
flow, resulting in a corresponding increase in inlet veloc- process stream under study.
ity and an increase in the centrifugal forces within the In manual optical microscopy, the dispersed particles
cyclone. This results in a successive decrease in the limit- are viewed by transmission, and the areas of the magni-
ing particle-separation size of the cyclones. fied images are compared with the areas of circles of
The cyclosizer is manufactured to have definite limit- known sizes inscribed on a graticule.
ing separation sizes at standard values of the operating The relative numbers of particles are determined in
variables, viz. water flow rate, water temperature, particle each of a series of size classes. These represent the size
density, and elutriation time. To correct for practical oper- distribution by number from which it is possible to calcu-
ation at other levels of these variables, a set of correction late the distribution by volume and, if all the particles
graphs is provided. have the same density, the distribution by weight.
Complete elutriation normally takes place after about Manual analysis of microscope slides is tedious and
20 min, at which time the sized fractions are collected by error prone; semi-automatic and automatic systems have
discharging the contents of each apex chamber into sepa- been developed which speed up analyses and reduce the
rate beakers. The samples are dried, and weighed to deter- tedium of manual methods (Allen, 1997).
mine the mass fractions of each particle size. The development of quantitative image analysis has
made possible the rapid sizing of fine particles. Image
analyzers accept samples in a variety of forms—photo-
4.4.4 Microscopic Sizing and Image Analysis graphs, electron micrographs, and direct viewing—and
are often integrated in system software. Figure 4.15 shows
Microscopy can be used as an absolute method of particle the grayscale electron backscatter image of a group of
size analysis, since it is the only method in which individ- mineral particles obtained with a scanning electron micro-
ual mineral particles are observed and measured (Anon., scope; grains of chalcopyrite (Ch), quartz (Qtz), and epi-
1993; Allen, 1997). The image of a particle seen in a dote (Epd) are identified in the image. On the right are
microscope is two-dimensional and from this image an plotted the size distributions of the “grains” of the mineral
estimate of particle size must be made. Microscopic siz- chalcopyrite (i.e., the pieces of chalcopyrite identified by
ing involves comparing the projected area of a particle the instrument, whether liberated or not) and the “parti-
with the areas of reference circles, or graticules, of known cles” in which the chalcopyrite is present. The plots are
sizes, and it is essential for meaningful results that the based on the analysis of several hundred thousand parti-
mean projected areas of the particles are representative of cles in the original sample, and are delivered automati-
the particle size. This requires a random orientation in cally by the system software. Image analysis of this kind
three dimensions of the particle on the microscope slide, is available in many forms for the calculation of many
which is unlikely in most cases. quantities (such as size, surface area, boundary lengths)

FIGURE 4.15 Using image analysis to estimate the size distributions of particles and mineral grains (Courtesy JKMRC and JKTech Pty Ltd).
104 Wills’ Mineral Processing Technology

FIGURE 4.16 Principle of the Coulter Counter.

FIGURE 4.17 The laser diffraction instrument principle (from Napier-Munn et al., 1996; Courtesy JKMRC, The University of Queensland).

for most imaging methods, for example, optical, electron. count representing the number of particles larger than
(See also Chapter 17.) some determinable volume proportional to the appropriate
threshold setting. By taking a series of counts at various
4.4.5 Electrical Impedance Method amplification and threshold settings, data are directly
The Beckman Coulter Counter makes use of current obtained for determining number frequency against vol-
changes in an electrical circuit produced by the presence ume, which can be used to determine the size distribution.
of a particle. The measuring system of a Coulter Counter As the instrument measures particle volume, the
is shown in Figure 4.16. equivalent diameter is calculated from that of a sphere of
The particles, suspended in a known volume of electri- the same volume, which can be a more relevant size mea-
cally conductive liquid, flow through a small aperture sure than some alternatives. The instrument is applicable
having an immersed electrode on either side; the particle in the range 0.4
1,200 μm.
concentration is such that the particles traverse the aper-
ture substantially one at a time.
4.4.6 Laser Diffraction Instruments
Each particle passage displaces electrolyte within the
aperture, momentarily changing the resistance between In recent years, several instruments based on the diffraction
the electrodes and producing a voltage pulse of magnitude of laser light by fine particles have become available,
proportional to particle volume. The resultant series of including the Malvern Master-Sizer and the Microtrac. The
pulses is electronically amplified, scaled, and counted. principle is illustrated in Figure 4.17. Laser light is passed
The amplified pulses are fed to a threshold circuit hav- through a dilute suspension of the particles which circulate
ing an adjustable screen-out voltage level, and those through an optical cell. The light is scattered by the parti-
pulses which reach or exceed this level are counted, this cles, which is detected by a solid state detector which
 Particle Size Analysis Chapter | 4 105

measures light intensity over a range of angles. A theory of

light scattering is used to calculate the particle size distribu-
tion from the light distribution pattern, finer particles induc-
ing more scatter than coarser particles. The early
instruments used Fraunhofer theory, which is suitable for
coarse particles in the approximate range 1
2,000 μm (the
upper limit being imposed mainly by mechanical con-
straints). More recently Mie theory has been used to extend
the capability down to 0.1 μm and below. Some modern
instruments offer both options to cover a wide size range.
Laser diffraction instruments are fast, easy to use, and
give reproducible results. However, light scattering theory
does not give a definition of size that is compatible with
other methods, such as sieving. In most mineral proces-
sing applications, for example, laser diffraction size distri-
butions tend to appear coarser than those of other
methods. Austin and Shah (1983) have suggested a proce-
dure for inter-conversion of laser diffraction and sieve-
size distributions, and a simple conversion can be devel-
oped by regression for materials of consistent characteris-
tics. In addition, the results can depend on the relative
refractive indices of the solid particles and liquid medium
(usually, though not necessarily, water), and even particle
shape. Most instruments claim to compensate for these
FIGURE 4.18 The Thermo GammaMetrics PSM-400MPX on-line par-
effects, or offer calibration inputs to the user. ticle size analyzer (Courtesy Thermo Electron Corporation).
For these reasons, laser diffraction size analyzers should
be used with caution. For routine high volume analyses in a
fixed environment in which only changes in size distribu- at target throughput maximizes efficient use of the
tion need to be detected, they probably have no peer. For installed power.
comparison across several environments or materials, or Continuous measurement of particle size in slurries
with data obtained by other methods, care is needed in inter- has been available since 1971, the PSM (particle size
preting the data. These instruments do not, of course, pro- monitor) system produced then by Armco Autometrics
vide fractionated samples for subsequent analysis. (subsequently by Svedala and now by Thermo Gamma-
Metrics) having been installed in a number of mineral
processing plants (Hathaway and Guthnals, 1976).
4.5 ON-LINE PARTICLE SIZE ANALYSIS
The PSM system uses ultrasound to determine particle
The major advantage of on-line analysis systems is that size. This system consists of three sections: the air elimi-
by definition they do not require sampling by operations nator, the sensor section, and the electronics section. The
personnel, eliminating one source of error. They also pro- air eliminator draws a sample from the process stream
vide much quicker results, in the case of some systems and removes entrained air bubbles (which otherwise act
continuous measurement, which in turn decreases produc- as particles in the measurement). The de-aerated pulp
tion delays and improves efficiency. For process control then passes between the sensors. Measurement depends
on-line measurement is essential. on the varying absorption of ultrasonic waves in suspen-
sions of different particle sizes. Since solids concentration
also affects the absorption, two pairs of transmitters and
4.5.1 Slurry Systems
receivers, operating at different frequencies, are employed
While used sometimes on final concentrates, such as Fe to measure particle size and solids concentration of the
concentrates, to determine the Blaine number (average pulp, the processing of this information being performed
particle size deduced from surface area), and on tailings by the electronics. The Thermo GammaMetrics PSM-
for control of paste thickeners, for example, the prime 400MPX (Figure 4.18) handles slurries up to 60% w/w
application is on cyclone overflow for grinding circuit solids and outputs five size fractions simultaneously.
control (Kongas and Saloheimo, 2009). Control of the Other measurement principles are now in commercial
grinding circuit to produce the target particle size distri- form for slurries. Direct mechanical measurement of par-
bution for flotation (or other mineral separation process) ticle size between a moving and fixed ceramic tip, and
106 Wills’ Mineral Processing Technology

FIGURE 4.20 Online visual imaging system (Split Engineering’s Split-

Online system) (camera is at top center) (Courtesy Split Engineering).

poorly operating units to be identified and changed while

allowing the cyclone battery to remain in operation.
Figure 4.19 shows an installation of both CiDRA systems
(PST, OSM) on the overflow pipe from a cyclone.
FIGURE 4.19 CiDRA’s CYCLONEtracSM systems in operation: the
PST (at top) and OSM (Courtesy CiDRA Minerals Processing).
4.5.2 On-belt Systems
Image analysis is used in sizing rocks on conveyor belts.
laser diffraction systems are described by Kongas and Systems available include those supplied by Split
Saloheimo (2009). Two recent additions are the Engineering, WipFrag, Metso, and Portage Technologies
CYCLONEtrac systems from CiDRA Minerals (see also Chapter 6). A fixed camera with appropriate
Processing (Maron et al., 2014), and the OPUS ultrasonic lighting captures images of the particles on the belt. Then,
extinction system from Sympatec (Smith et al., 2010). software segments the images, carries out appropriate cor-
CiDRA’s CYCLONEtrac PST (particle size tracking) rections, and calculates the particle size distribution. A
system comprises a hardened probe that penetrates into system installed on a conveyor is shown in Figure 4.20.
the cyclone overflow pipe to contact the stream and effec- Figure 4.21 shows the original camera views and the seg-
tively “listens” to the impacts of individual particles. The mented images for a crusher feed and product, together
output is % above (or below) a given size and has been with the calculated size distributions. (Another example is
shown to compare well with sieve sizing (Maron et al., given in Chapter 6.) Common problems with imaging sys-
2014). The OPUS ultrasonic extinction system (USE) tems are the inability to “see” particles under the top
transmits ultrasonic waves through a slurry that interact layer, and the difficulty of detecting fines, for which cor-
with the suspended particles. The detected signal is con- rection algorithms can be used. The advantage of imaging
verted into a particle size distribution, the number of fre- systems is that they are low impact, as it possible to
quencies used giving the number of size classes acquire size information without sampling or interacting
measured. Applications on ores can cover a size range with the particles being measured. These systems are use-
from 1 to 1,000 μm (Smith et al., 2010). ful in detecting size changes in crusher circuits, and are
In addition to particles size, recent developments have increasingly used in measuring the feed to SAG mills for
included sensors to detect malfunctioning cyclones. use in mill control (Chapter 6), and in mining applications
Westendorf et al. (2015) describe the use of sensors (from to assess blast fragmentation.
Portage Technologies) on cyclone overflow and underflow There are other on-line systems available or under test.
piping. CiDRA’s CYCLONEtrac OSM (oversize monitor) For example, CSIRO has developed a version of the ultra-
is attached to the outside of the cyclone overflow pipe and sonic attenuation principle, in which velocity spectrometry
detects the acoustic signal as oversize particles (“rocks”) and gamma-ray transmission are incorporated to produce a
hit the pipe (Cirulis and Russell, 2011). The systems are more robust measurement in the range 0.1
1,000 μm
readily installed on individual cyclones thus permitting (Coghill et al., 2002). Most new developments in on-line
 Particle Size Analysis Chapter | 4 107

FIGURE 4.21 The original and segmented images of crusher feed and product, with the estimated size distributions (Split Engineering’s Split-
Online system) (Courtesy Split Engineering).

analysis are modifications and improvements of previous Finch, J.A., Leroux, M., 1982. Fine sizing by cyclosizer and micro-sieve.
systems, or adaptations of off-line systems for on-line CIM Bull. 75 (839), 235
240.
usage. Harris, C.C., 1971. Graphical presentation of size distribution data: An
assessment of current practice. Trans. Inst. Min. Metall., Sec. C. 80,
C133
C139.
REFERENCES Hathaway, R.E., Guthnals, D.L., 1976. The continuous measurement of
particle size in fine slurry processes. CIM Bull. 766 (Feb.), 64
71.
Allen, T., 1997. fifth ed. Particle Size Measurement, vol. 1. Chapman
Heywood, H., 1953. Fundamental principles of sub-sieve particle size
and Hall, London. UK.
measurement. Recent Developments in Mineral Dressing: a
Anon, 1989a. British Standard 1017-1: 1989. Sampling of coal and coke.
Symposium Arranged by the Institution. Institute of Mining and
Methods for sampling of coal.
Metallurgy (Great Britain), Imperial College of Science &
Anon, 1989b. British Standard 1796-1: 1989. ISO 2591-1: 1988. Test
Technology, London, England. pp.31
58.
sieving. Methods using test sieves of woven wire cloth and perfo-
Kongas, M., Saloheimo, K., 2009. Online slurry particle size analyzers.
rated metal plate.
In: Malhotra, D., et al., (Eds.), Recent Advances in Mineral
Anon, 1993. British Standard 3406-4: 1993. Methods for the determina-
Processing Plant Design. SME, Littleton, CO, USA, pp. 419
432.
tion of particle size distribution. Guide to microscope and image
Maron, R., et al., 2014. Low maintenance, real-time tracking of particle
analysis methods.
size in individual hydrocyclones for process optimization. Proc. 11th
Anon, 2000a. British Standard 410-1: 2000. ISO 3310-1: 2000. Test
Encuentro International de Mantenedores de Plantas Mineras
sieves. Technical requirements and testing. Test sieves of metal wire
(MAPLA) Conf., Paper: BIO535, Santiago, Chile. 1
11.
cloth.
Napier-Munn, T.J., 1985. The determination of the size distribution of
Anon, 2000b. British Standard 410-2: 2000. ISO 3310-2: 2000. Test
ferrosilicon powders. Powder Tech. 42 (3), 273
276.
sieves. Technical requirements and testing. Test sieves of perforated
Napier-Munn, T.J., et al., 1996. Mineral Comminution Circuits: Their
metal plate.
Operation and Optimisation (Appendix 3). Julius Kruttschnitt
Anon, 2001a. British Standard ISO 13317-1: 2001. Determination of par-
Mineral Research Centre (JKMRC), The University of Queensland,
ticle size distribution by gravitational liquid sedimentation methods.
Brisbane, Australia.
General principles and guidelines.
Price, E.W., 1962. Stokes’ law and the Haultain infrasizer unit. Ind. Eng.
Anon, 2001b. British Standard ISO 13317-2: 2001. Determination of par-
Chem. Process Des. Dev. 1 (1), 79
80.
ticle size distribution by gravitational liquid sedimentation methods.
Rosin, P., Rammler, E., 1933. The laws governing the fineness of pow-
Fixed pipette method.
dered coal. J. Institute Fuel. 7, 29
36.
Austin, L.G., Shah, I., 1983. A method for inter-conversion of microtrac
Schuhmann, R., Jr., 1940. Principles of comminution, I-Size distribution and
and sieve size distributions. Powder Tech. 35 (2), 271
278.
surface calculations. Technical Publication. No. 1189, AIME, 1
11.
Austin, L.G., et al., 1984. Process Engineering of Size Reduction: Ball
Smith, A., et al., 2010. Ultrasonic extinction for full concentration, real
Milling. AIME, New York, NY, USA.
time particle size analysis in the mining industry. Proc. 2nd
Bernhardt, C., 1994. Particle Size Analysis: Classification and
International Congress on Automation in the Mining Industry
Sedimentation Techniques, vol. 5. Chapman & Hall, London, UK.
(Automining), Santiago Chile. 1
10.
Cirulis, D., Russell, J., 2011. Cyclone monitoring system improves
Stokes, S.G.G., 1891. Mathematical and Physical Papers, vol. 3.
operations at KUC’s Copperton concentrator. Eng. Min. J (E&MJ).
Cambridge University Press Warehouse, London, UK.
212 (10), 44
49.
Westendorf, M., et al., 2015. Managing cyclones: A valuable asset, the
Coghill, P.J., et al., 2002. On-line measurement of particle size in min-
Copper Mountain case study. Mining Eng. 67 (6), 26
41.
eral slurries. Miner. Eng. 15 (1-2), 83
90.