LEARNING UNIT 2: THE FUNDAMENTALS OF MINERALS PROCESSING

Contents

2.1 Introduction....................................................................................................2

2.2 Mineral.......................................................................................................3

2.3 Mineral deposit and mineral ore................................................................4

2.4 Mineral beneficiation..................................................................................7

2.5 Mineral beneficiation approach..................................................................8

2.5.1 Wet grinding......................................................................................11

2.5.2 Particle size distribution (PSD) representation methods...................18

2.5.3 Graphical representation of sizing analyses.....................................20

2.5.4 The attainable region (AR) method...................................................25

2.6 Mineral liberation.....................................................................................29

2.7 Mineralogical analysis.............................................................................31

2.8 Worked examples....................................................................................32

References........................................................................................................36

1
Learning outcomes

After completing this learning unit, you will be able to

 analyse the particle size distribution of a mineral sample correctly

 state the different mineralogical techniques used in the characterisation of

mineral ores

 differentiate between a mineral deposit and a mineral ore

 illustrate the significance of minerals processing as a professional discipline

 discuss the importance of liberation in the beneficiation of ores

 describe the factors that affect the mineral beneficiation process

2.1 Introduction

The first process that most of the ores undergo after they leave any mine (run-of-

mine ore) is minerals processing, sometimes known as ore dressing. It is a process

of ore preparation, milling and beneficiation. Ore dressing is a process of

mechanically separating the grains of ore minerals from the gangue minerals. It is

done in order to produce a concentrate containing most of the ore minerals and a

tailing part (discard) containing the rest of the gangue minerals.

The importance of minerals processing is taken for granted today, but it is interesting

to reflect that less than a century ago, ore concentration was often a fairly crude

operation, involving relatively simple gravity and hand-sorting techniques performed

by the mining engineers. Over the years, however, minerals processing has

developed into an important professional discipline. Without modern-day processing

techniques, the concentration of many ores would be hopelessly uneconomical (Wills

1991).

2
The effective liberation of material of value remains one of the major challenges in

treating modern ores which, due to the increasing complexity and smaller grain sizes

of these ores, requires finer grinding to achieve the necessary degree of liberation

(Veasey & Will 1991). Vizcarra et al (2010) state that an effective means to liberate

minerals from the surrounding phases would be a significant advance in minerals

processing technology. This is because the degree of liberation determines the

efficiency of both the comminution and subsequent units in a beneficiation process.

Regrettably, efforts to address this liberation problem by way of fine and ultrafine

grinding have led to higher energy utilisation, because both the valuable (mineral)

and valueless (gangue) components are indiscriminately milled. Hlabangana et al

(2017) estimate that milling alone accounts for more than 30–50% of the total power

used in the concentration process, but this can rise to as high as 70% for hard or

finely dispersed and intergrown ores. The objective of this learning unit is for you to

develop an understanding of the fundamentals of minerals processing.

2.2 Mineral

A mineral is a natural inorganic compound that has a specific or homogeneous

chemical composition and atomic structure (Wills 2005). For example, the mineral

pentlandite, which is the primary source of nickel metal, has a chemical composition

(Fe.Ni)9.S8. The mineral galena (PbS) is the primary source of lead metal. Another

example of a mineral is chalcopyrite, which is the primary source of copper metal

and has the chemical formula CuFeS2.

3
The term “mineral” is often loosely used to describe anything of economic value

mined below the surface of the earth, e.g. coal. Coal is not a mineral but just a rock,

since it does not have a homogeneous chemical and physical composition. Through

a process known as metamorphosis, which occurs over a period of more than 200

million years, vegetative matter is turned into a sedimentary rock known as coal.

There are different “grades” or “ranks” of coal, depending on the degree of

metamorphosis. Carbon content is usually used as a measure of the rank of coal.

The order of the ranks starting from the lowest is wood, peat, lignite, sub-bituminous,

bituminous and anthracite.

The “grade” of a mineral ore is measured by the metal content of its most valuable

component. This grade may be expressed as a percentage, parts per million or parts

per billion, depending on the quantity of the elemental metal. Since all metals exist in

mineral form due to their degree of reactivity with their environment, particularly with

oxygen, carbon dioxide and sulphur, it is sometimes necessary to convert from metal

content to mineral content or from mineral content to metal content. These

conversions are important in determining the metallurgical assay of an ore. Worked

examples at the end of learning unit 4 will illustrate how you can apply these

conversions.

2.3 Mineral deposit and mineral ore

Table 2.1 shows an estimate of the relative abundance of metals deposited within

the earth’s crust. If these metals were to be found evenly distributed throughout the

earth’s crust, it would be quite uneconomical to extract them. Fortunately, geological

conditions as well as vegetative matter determine where minerals are located in

nature. The minerals determine the vegetative matter and so vegetative matter can

4
be indicative of the presence of certain minerals. Certain minerals are naturally found

in association with specific types of rocks. There are mainly two types of rock

material, namely sedimentary and igneous. Sedimentary rocks result from stratified

deposition of matter, which is then subjected to pressure and temperature changes

over millions of years. Igneous rocks result from the cooling of magma either within

or outside the earth’s crust.

Due to this phenomenon of mineral-rock association, it is possible to predict as well

as quantify the amount of minerals below the surface of the earth by inference from

geological studies. However, not all deposits can be mined profitably. When a

deposit contains valuable metals that are economically viable to exploit, the mineral

deposit then becomes a mineral ore.

Table 2.1: Relative abundance of metals in the earth’s crust

Element Abundance Amt. in top 3.5 Element Abundance Amt. in top 3.5
(%) km (tons) (%) km (tons)
Oxygen 46.4 - Vanadium 0.014 14
10 −10
15

Silicon 28.2 Chromium 0.01 -

16 18
Aluminium 8.2 10 −10 Nickel 0.0075
Iron 5.6 Zinc 0.007
13 14
Calcium 4.1 - Copper 0.0055 10 −10
Sodium 2.4 - Cobalt 0.0025
Magnesiu 2.3 16
10 −10
18
Lead 0.0013
m
Potassium 2.1 - Uranium 0.00027
Titanium 0.57 15
10 −10
16
Tin 0.0002
11 13
Manganese 0.095 - Tungsten 0.00015 10 −10
−6
Barium 0.043 - Mercury 8 ×10
−6
Strontium 0.038 - Silver 7 ×10
−6
Rare earth 0.023 - Gold ¿ 5 ×10
11
Zirconium 0.017 14
10 −10
16
Platinum ¿ 5 ×10
−6
¿ 10
metals
Source: Wills (2005)

5
A mineral ore is a heterogeneous conglomerate or aggregate made up of a mixture

of extractable value minerals and non-valuable material known as gangue. An ore is

usually classified in reference to the nature of the constituent value minerals.

Sulphide ores contain the value metal in a sulphide compound form, whereas native

ores contain the value metal in elemental form. Oxidised ores contain the value

metal in an oxide, sulphate, silicate or carbonate form. Ores can also be classified

either as simple or complex. A simple ore contains only a single extractable value

mineral, and a complex ore contains more than a single value mineral that is

economically viable to exploit. An ore can also be classified as either metallic or non-

metallic, depending on its mineral composition. However, a complex ore can be both

metallic and non-metallic, depending on which minerals of interest are targeted for

exploitation (Bailey 2003).

A precious group mineral (PGM) ore is a low-grade complex metallic ore usually

composed of platinum, rhodium, rubidium, strontium and palladium in quantities as

low as 1 part per million (ppm). PGM ores are normally associated with base metal

sulphide ores, e.g. chalcocite, chalcopyrite or pentlandite. To improve the economic

feasibility of exploiting PGM ores, minerals processors proposed a path in which the

PGM ore is first processed for base metals, namely copper (Cu) or nickel (Ni), and

the tailings processed for PGMs. These tailings would now have an improved grade

in terms of the PGMs resulting from the reverse concentration effect brought about

by the initial extraction of base metals.

A diamond ore is a simple non-metallic ore that is known to have the lowest grade of

all mined ores in the world. On top of that, diamond ores such as the Kimberlite ore

6
found in both South Africa and Botswana are notorious for not being associated with

any other mineral of economic value. The Debswana ore in Botswana has reported

grades as low as 0.02 ppm but still remains the richest diamond mine in the world.

This is because diamond deposits are exploited mainly for their gem-quality stones

that have the highest value of all minerals. The lower value industrial quality stones

are obtained as by-products, although they are now facing stiff competition from

synthetic diamonds.

2.4 Mineral beneficiation

Mineral beneficiation is the science and art of extracting desirable metals from their

ores in order to produce a valuable commercial end product

(Redstarmining.indocoal.com [n.d.]). It is a process in which ore particles are first

reduced in size (comminution) and then separated on the basis of their physical

and/or chemical properties (Danha 2013).

Our objective in mineral beneficiation is to split the material being processed into two

or more process streams. The intention is for the minerals of interest to be diverted

into one stream, the concentrates, whereas the other minerals present are diverted

into a second stream, the tailings. Sometimes when the minerals are difficult to

separate, a third stream called the middlings is produced. This has a mineral content

that is intermediate between that of a concentrate and the tailings. A middlings

stream is usually too rich in the mineral of interest to be discarded as tailings.

Mineral beneficiation is done for the following reasons:

7
  To extract a valuable mineral: Most minerals of commercial value naturally

occur associated with gangue, and extraction of these minerals involves

separating them from this gangue.

 To remove undesirable impurities: Deleterious materials, which reduce the

value of the ore, can be removed by the use of mineral beneficiation

techniques.

 To produce a concentrate: The objective here is to enrich the material

(concentrate) by discarding as much of the gangue as possible.

 To separate two or more valuable minerals: It is common for a valuable ore

to contain more than one desirable mineral. In this situation it is necessary not

only to extract the concentrate and discard the gangue, but also to process

the remaining values in the gangue.

2.5 Mineral beneficiation approach

Mineral beneficiation starts when “run-of-mine” ore arrives at the processing plant

from the mine site. It involves the preparation and “liberation” of the value mineral

from gangue minerals (comminution stage) and the separation of these value

minerals from gangue minerals in order to produce a concentrate and tailings

streams, respectively (mineral extraction stage).

8
 Run of Mine

Crushing Mineral
Oversize concentrate
Screening
Mineral extraction
(separation/concentration)
Grinding
Tailings
Oversize
(gangue)
Classification

Comminution stage
tion for liberation of valuables)
Figure
Figure 1. 2.1: Typical
Typical mineral
Mineral beneficiation
Processing flow sheet
Flowsheet

The comminution stage is the most important unit operation in the beneficiation of

ores. The primary objective of carrying out comminution in the minerals processing

industry is not merely to reduce the particle size of the feed material, but also to free

(liberate) the constituent minerals that make up an ore so that valuable minerals can

be separated from the minerals of little value, as illustrated by figure 2.2. However,

this process is generally performed relatively poorly, and at considerable expense in

terms of operational costs (Hlabangana et al 2017).

Ore
Gangue

Valuable Mineral Gangue

Mineral

Figure 2.2: Liberation of valuable minerals by size reduction (Danha 2013)

9
In conventional practice, the comminution of mineral ores is performed by crushing,

followed by grinding. Crushing, which is usually a multiple-stage dry process,

reduces the size of particles to a level at which grinding can be carried out, in order

to liberate the valuable minerals. Grinding is the last stage in the process of

comminution, and is also the most important part of the process of liberating valuable

minerals from gangue. Grinding has changed very little in the past 50 years, and

continues to be carried out in tumbling mills. It is well known that these are extremely

energy-inefficient and have many limitations. The reasons that conventional mills

have been used for so long are that they are relatively simple devices, and are

continuous, high-throughput machines that can be fairly easily controlled to produce

the desired product particle size distribution (Wills 1990).

Laboratory and industrial grinding tests have shown that grinding aids can influence

the comminution process significantly. It is generally known that water is the most

commonly used grinding aid, and that slurry density plays an important role in the

comminution process. The density of slurry determines the distribution of solid

charge in the mill, with migration of particles from the tumbling zone to the mill

periphery as the solid concentration is increased (Tangsathitkulchai & Austin 1989).

The changes in the circulation path of the mill charge in wet grinding lead to

variations in the grinding performance and mill’s power consumption. The amount of

water that has to be added to the mill charge to maximise the breakage process is

important. Tangsathitkulchai and Austin (1989) report a maximum in breakage rates

at a slurry density of about 45 volume per cent solids in water. However, the

optimum slurry density changes in accordance with the specific objectives of a

particular grinding process, and the mineralogical composition and physical

10
properties of the ore. The advantage of wet over dry grinding lies in the manner

water has of bringing and keeping particles into positions advantageous to breakage.

2.5.1 Wet grinding

Wet grinding is performed in tumbling mills where the tumbling action of the grinding

medium reduces the size of the ore particles. The grinding medium can be steel

rods, or balls, the ore itself (autogenous grinding) or a combination of the ore and

balls (semi-autogenous grinding). Grinding is an indiscriminate process that is

governed by the laws of probability, and the degree of grinding of an ore particle

depends on the probability of it being caught between the grinding media (Metin,

Güleç & Cingitaş 2015). Wet grinding in a ball mill is influenced by a number of

factors:

 mill power

 mill type

 mode of operation

 rotational speed

 product quality

 mill conditions

 screening

2.5.1.1 Mill power

The power demand (kW) of a conventional ball mill is a function of its known

dimensions and basic operating conditions. The Hogg and Fuerstenau Model (1972),

Harris Power Model (Harris, Schnock & Arbiter 1970), Beeck’s Model (1970) and

Moys Model (Dong & Moys 2003) are some of the power models used in the

comminution industry.

11
2.5.1.2 Mill type

The most common type of mill for grinding is a tumbling ball mill. A ball mill is a

retention mill in which a reservoir of particles is acted on repeatedly. It is simple,

relatively easy to construct, reliable, easy to control and maintain, and has low

energy requirements per ton of product. A ball mill is an all-purpose mill that uses

steel balls as grinding media. It is particularly suitable for fine grinding, for the

following reasons (Wills & Finch 2016):

 Ball mills have greater breakage forces than rod mills of the same diameter.

This is because ball-on-ball contact occurs at a point, whereas rod-on-rod

contact occurs along a line. Thus, for the same diameter and mass of grinding

medium, the area of contact is less with balls than it is with rods. Greater

breakage forces result in a greater degree of size reduction.

 In ball mills, the presence of large particles does not impede the breakage of

small particles in the way it does in rod mills.

 In fine grinding, the number of particles to be ground is enormous. A larger

number of contact points should therefore increase the grinding capacity. In

ball milling, this can be achieved by the use of small balls.

2.5.1.3 Mode of operation

The mode of operation of a tumbling ball mill (figure 2.3) is as follows: The rotation

carries balls and granular material round the mill and as the balls tumble down, they

strike particles caught against other balls. In addition, the general movement of the

balls in the bed will rub particles between them. The three types of size reduction

which can occur are caused by fracture, chipping, abrasion or attrition. The

measured values of the primary progeny fragment distribution are the total of the

12
fragments produced by each mechanism. At low rotational speed the balls have a

relatively gentle tumbling action, but as the speed increases, the tumbling action

converts to a cascading state. At higher rotational speeds, a cataract of balls is

formed.

Figure 2.3: Mode of operation of a ball mill (Danha 2013)

2.5.1.4 Rotational speed

The tumbling action in a mill is a result of the rotation of the mill about its axis. The

rates and modes of particle (powder) breakage vary with the mill speed. At low

rotational speeds, the components of the load remain in contact with one another,

and the tumbling action is called cascading. The dominant size reduction in this

mechanism is attrition. At high speeds, the components of the load are ejected into

free flight before impacting on the mill liners. This tumbling action is called

cataracting and the dominant size reduction is by impact. The net power required to

turn a mill is also a function of rotational speed.

13
2.5.1.5 Product quality

The quality of the product obtained from the grinding circuit is often specified by one

or more of the following measurements: the weight fraction of product passing one or

two preselected sieve sizes, the specific surface area, or the achievement of a

preselected degree of liberation of valuable material from a larger mass of rock. It is

sometimes desirable to give an approximate indication of the extent of size reduction

accomplished or desired. A term often used is reduction ratio, which is the ratio of

the maximum particle size in the feed to that in the product. To calculate the

reduction ratio, it is necessary to characterise the size of the feed and size of the

product by single numbers. This ratio is commonly taken to be the ratio of 80%

passing sizes of feed to product. The product specifications chosen in a given

situation depend on which characteristic is most easily correlated to the desired

performance of the end product.

2.5.1.6 Mill conditions

Changing mill conditions is the same as changing the temperature in a chemical

reactor. Below we briefly discuss some of the mill conditions that need to be

specified to address the question of grinding efficiency.

a) Critical speed

The critical speed of the mill is defined as the rotational speed at which the balls

have stopped tumbling and have just started to centrifuge on the mill case.

Critical speed, rpm = 42.2/√ D−d (2.1)

14
 Where

D = Internal mill diameter

d = Maximum ball diameter (Chimwani 2014)

The maximum power drawn by a mill occurs at different fractions of critical speed

from one mill to another, depending on the mill diameter, ratio of ball to mill diameter

and the ball and powder filling conditions. This maximum is usually found in the

range of 70–80% of critical speed. The rotational speed of the mill is normally

specified by the fraction of the critical speed ( ∅ ¿¿ c )¿ (Deniz 2004).

b) Ball filling

The rate of breakage depends on how much of the mill volume is filled with balls.

The fractional ball filling, J, is usually expressed as the fraction of the mill filled by the

ball bed at rest. It is conventional to define constant formal bed porosity for all

calculations. Using a formal bed porosity of 0.4:

J=Mass of balls / ( Ball density × Mill volume ) ×1.0 /0.6 (2.2)

c) Ball diameter

The power required by the mill varies with the ball diameter, and an incorrect choice

of ball diameter results in direct inefficiency. A charge of balls that are too small

causes a slumping in the charge, a low power draw, plus excessive wear rates of the

balls. A different ball diameter allows the same mass of grinding media in the mill,

but will achieve a different extent of grinding and power draw. The optimum choice of

15
ball size depends on the feed size distribution, the desired product size and the

balance between energy and steel costs.

d) Mill filling

It has been established that for a given ball loading, it is undesirable to either

underfill or overfill the mill with powder. In the case of a low filling, much of the

energy of the tumbling balls is taken up in steel-to-steel contact, whereas at high

filling the powder cushions the breakage action. Mill filling by feed is expressed as

the fraction of mill volume filled by the feed bed:

f c =Mass of feed / ( Feed density × Mill volume ) ×1.0 /0.6

(2.3)

To relate the particle loading to the ball loading, the formal bulk volume of powder is

compared with the formal porosity of the ball bed:

U =f c /0.4 J (2.4)

Where

U = 0.6 to 1.1 = A good feed/ball loading ratio to give efficient ball loading in the mill

U = Fraction of the spaces between the balls at rest which is filled with feed

(Hlabangana et al 2016)

e) Slurry density

16
Slurry density is defined by the fraction by weight of solids in the mixture. The

rheological properties of a slurry are better defined by the volume fraction of solids,

c:

c=( w s / ρs ) ×1/ { ( w s / ρs ) + [ ( 1−ws ) / ρl ] } (2.5)

Where

w s = Weight fraction of solid

ρ s∧ρl = Densities of the solid and the liquid

The solids to liquids ratio in a wet grinding process has a strong influence on the

operation of the mill because it affects the viscosity of the slurry, and hence the rate

at which it flows through the mill. This in turn affects the period during which the

solids are resident in the mill, and thus the extent to which they are milled. The solids

to water ratio also affects the grinding action in the mill. A slurry that is too dilute

leads to excessive contact between the components of the medium, which gives

high rates of wear. Conversely, too high a solids concentration results in a

cushioning of the medium, which reduces the grinding capacity. The slurry density

therefore has to be within an optimum range, usually quoted at 45% by volume.

2.5.1.7 Screening

Screening of material is used extensively in the minerals industry, and is generally

carried out on relatively coarse material since the efficiency decreases as the

fineness increases. Fine screens are expensive, very fragile and tend to become

blocked with retained material. Screening is therefore usually limited to material

above about 38 µm in size (Austin & Rogers 1985), whereas finer sizing is normally

17
done by classification or laser diffraction. Screening can be performed wet or dry, but

even in dry screening, wet washing is required. The main purposes of screening in

the minerals processing industry (Gupta 2003) are

 to prepare a closely sized end product, because the final product specification

usually provides information that is important to subsequent processes

 to prevent oversize material from passing to the next stage in crushing and

grinding operations

 to increase the capacity and efficiency of crushing machines by preventing the

entry of undersize material

Types of screens

A screen is a surface with a number of apertures of given dimensions. There are

many types of screens, which are broadly classified as either stationary or moving.

These are some examples:

 Stationary screens

 sieve bends

 the grizzly

 Moving screens

 revolving screens

 shaking screens

 reciprocating screens

 gyratory screens

 vibrating screens

18
2.5.2 Particle size distribution (PSD) representation methods

After particles have been size reduced and screened, the next step in the process is

size analysis. The concept of defining particle size in minerals processing is

ambiguous because ore particles do not have definite shapes, like spheres and

cubes. The size of a spherical particle is defined by its diameter, and a cubical

particle by the length of a side. Even for particles of regular shape, particle size does

not have a unique meaning. In minerals processing an indirect measure of size is

used, and the size of a particle is defined as the smallest whole opening in a square

mesh screen through which a particle will fall. As previously noted, if the particles are

too small to be measured using sieves, other methods like laser diffraction or light

scattering can be used for the purpose.

It is often necessary to divide the particle population into groups, each identified by

the smallest and largest size in the group. Screening using mesh sizes in geometric

series is normally used to achieve this classification. The concept of particle groups

or classes allows for the description of the behaviour of classes of particles rather

than that of individual particles. A representative size is associated with each particle

size class, and it is assumed that all particles in the class will behave as if they had a

size equal to the representative size.

PSD can be measured experimentally using sieves. The data obtained by screening

is normally presented in tabular form that shows mesh size against the fraction

smaller than that mesh. Graphical representations are generally preferred because

of their convenience in assessing and comparing particle size distributions. A variety

of different graphical coordinate systems have become popular, some with a view to

19
making the particle size distribution function plot a straight line or close to a straight

line. In some of these graphical plots, the particle size axis is plotted on a logarithmic

coordinate scale, and various options are used for the ordinate scale. In other

graphical plots, specially ruled graph papers are available for PSD plots

(Vdocuments [n.d.]).

2.5.2.1 Mean particle size

The expression of the particle size of a powder in terms of a single linear dimension

is often required. For coarse particles, the size of the opening through which 80% of

the material will pass is arbitrarily chosen. This size d80 is a useful rough comparative

measure for the size of material which has been through a crusher. A mean size will

describe only one particular characteristic of the powder and it is important to decide

what that characteristic is before the mean is calculated. Thus, it may be desirable to

define the size of particle such that its mass or its surface or its length is the mean

value for all the particles in the system. It is usually assumed that each of the

particles in the system has the same shape. Consider a unit mass of particles

consisting of n1 particles of characteristic dimension d1, constituting a mass fraction

x1, then
3
x 1=n1 k 1 d 1 ρs (2.6)

¿ : ∑ x 1=1= ρs k 1 ∑ ( n1 d 31 )

2.5.2.2 Mean particle size based on surface

Instead of fraction of total mass, if the surface in each fraction is plotted against size,

the mean abscissa ds is then the mean surface diameter (ds).

20
The mean surface diameter is defined as the size of particle which is such that if all

the particles are of this size, the total surface will be the same as in the mixture:

d s=
∑ 3
n1 d1
(2.7)
∑ n1 d 21
2.5.3 Graphical representation of sizing analyses

Shown in this section are some of the different methods of charting sizing tests. An

example size distribution from a set of test sieves is shown in table 2.2. Size

fractions are normally identified by their top size and bottom size, as shown in

column (A). The last size fraction, in this case -45 micro-metres, is material that

goes through all of the screens and is collected in the bottom pan (the final pan

product).

Table 2.2: Raw particle size distribution data from a set of test sieves

(A) (B) (C) (D) (E) (F)

Sieve size Nominal sieve Individual weight Individual % wt Cumulative % Cumulative %
range (µm) size (µm) retained (grams) retained passing retained
- 250 250 0.02 0.05 99.95 0.05
-250/-180 180 1.32 2.96 96.99 3.01
-180/+125 125 4.23 9.5 87.49 12.51
-125/+90 90 9.44 21.19 66.3 33.7
-90/+63 63 13.1 21.41 36.89 63.11
-63/+45 45 11.56 25.95 10.94 89.06
- 45 0 4.87 10.94 0 100
Total 44.54 100

The nominal sieve size, column (B), is the size of the openings in the sieves. The

individual weight retained, column (C), is the weight in grams that is retained on

each sieve in the stack. Column (D) is calculated from the values in column (C) by

dividing each individual weight by the total weight and multiplying by 100%. The

cumulative % passing, column (E), is the fraction of the total weight that passes

through each individual screen. It is calculated for a given size by subtracting the

21
material retained on all of the coarser sieves from 100%. For example, in table 2.2,

the cumulative % passing 63 µm is 100% - (0.05 + 2.96 + 9.50 + 21.19 + 29.41)%

= 36.89%.

The last entry in column E (corresponding to the final pan product) must be equal

to zero if no mistakes were made in the calculation. The cumulative % retained,

column (F), is the fraction of the total weight that has been retained by a given

screen and all screens coarser than it. Cumulative % retained plus cumulative %

passing at any given size must equal 100%. Individual % weight retained can be

directly plotted against the sieve size, as shown in figure 2.4. This is a frequency or

probability density function (PDF) plot and is plotted either on arithmetic axes or

semi-logarithmic axes.

In a normal arithmetic plot, the numbers are uniformly spaced on the axis. In a

logarithmic plot, the numbers are spaced according to their logarithms, so the

larger numbers are closer together than the smaller numbers. The logarithmic plot

is preferred for size distributions, because it allows a wide range of particle sizes to

be plotted without crowding together the points for the finer size fractions. The PDF

plot is used to determine what size fractions contain most of the material. The

cumulative % passing or cumulative distribution function (CDF) can also be plotted

versus size, as shown in figure 2.4. Normally, only the cumulative % passing or

cumulative % retained is plotted, not both, since one is the inverse of the other.

22
 Figure 2.4: Probability density function (PDF) and cumulative distribution function
(CDF) plots (Danha 2013)

Figure 2.5: The Gates–Gaudin–Schumann plot

Figure 2.5 shows the Gates–Gaudin–Schumann plot, which is a graph of the

cumulative percentage of material passing a sieve size versus the sieve size. In this

23
type of plot, both axes use the logarithmic scale, and most of the data points should

lie in a straight line. The equation for the straight line section of the plot is given by
a
y=(x /k ) (2.8)

Where

y = Cumulative % passing

x = Particle size

k = Size modulus

a = Distribution modulus

If we take logs on both sides of equation 2.8, it converts to a straight line given by

equation 2.9:

log y=a × log ⁡(x)+(2−a × log ( k ) ) (2.9)

Where

a = Slope of the line

( 2−a × log ( k ) ) = y–intercept

The size modulus is a measure of how coarse the size distribution is, and can be

determined from the graph by extrapolating the straight-line section of the plot up to

the 100% passing mark and finding the corresponding size value. The distribution

modulus is a measure of how broad the distribution is, and can be obtained by

calculating the slope of the straight-line section of the plot.

24
 Figure 2.6: A typical Rosin–Rammler plot (Danha 2013)

Figure 2.6 shows a typical Rosin–Rammler plot that is frequently used to describe

the PSD of powders of various types and sizes. The function is particularly suited to

representing particles generated by grinding, milling and crushing operations. The

conventional Rosin–Rammler function is described by equation 2.10.

[ ( )]
n
D
R=exp − (2.10)
Dm

Where

R = Retained weight fraction of particles with a diameter greater than D

D = Particle size

Dm = Mean particle size

n = Measure of the spread of particle sizes

25
All the above PSD representation methods are limiting in that they do not provide

connectivity between the different states of a grinding process. The attainable region

(AR) method, which is briefly discussed in the next section, was developed at the

Centre of Material and Process Synthesis (COMPS), University of the

Witwatersrand, South Africa, to adequately represent the state of dynamic grinding

systems.

2.5.4 The attainable region (AR) method

The AR method has been successfully applied in the field of chemical reactor

engineering for choosing and optimising reactor configurations. It was first proposed

as a novel approach to the problem of reactor synthesis by Horn (1964), who

addressed the issue of finding the optimal reactor structure for a given set of

competing reactions and associated kinetics. He noted that for given kinetics and

feeds, it might be possible to find the set of all possible output concentrations from all

possible reactor systems. He called the set of all possible products the attainable

region (Danha et al 2015). There are two reasons why all the possible products have

to be found:

 If an objective function is complicated, it might not be known where the

optimum will occur.

 One of the products may be an intermediate that is required to produce the

optimum. What the global optimum is will not be known until all the possible

intermediate products have been found.

What follows covers some of the terminology that is frequently used in the

description, analysis and understanding of the AR technique.

26
 o State variable

A state variable is one of the parameters that is used to describe the mathematical

"state" of a dynamic system. Intuitively, the state of a system describes enough

about the system to determine its future behaviour.

o The objective function

The objective function is the function that needs to be optimised. This may be the

maximising of a given product, the minimising of some amount of energy utilised or

any other state variables.

o The characteristic vector

The listed coordinates of a point that specify the product of a system are termed the

characteristic vector. From vector mathematics, we know that the coordinates of a

point can also be considered as a vector from the origin. The characteristic vector

characterises the state (or composition) of a stream, and must contain sufficient

variables to describe fully the reaction kinetics and the objective function.

o The reaction vector

The reaction vector contains the rates of formation of the components and thus the

kinetics of all the reactions. The reaction vector gives the instantaneous change in

state of a mixture when it undergoes a batch reaction.

o The mixing vector

27
The only other process that may occur in any given system is mixing, which can

occur in two modes: on its own or inside a reactor. The former occurs when two

process streams combine to form a single process stream. Mixing, like reaction, is

represented by a vector that points from the stream being considered to the stream it

is being mixed with.

Finding the AR requires an iterative construction process. However, there are a set

of necessary conditions or "rules" that make it possible to check whether all the

possible products have been found to help construct the region. The next step,

finding the optimum, entails the relatively simple procedure of looking for the point on

the boundary where the objective function is optimised. This will usually be found

where one of the lines of constant value for the objective function just touches the

boundary of the AR. In most cases the objective function will be at a tangent to the

boundary it touches. After that, based on knowledge of the processes that are

required to reach the optimal point, the engineer can determine an optimal flow sheet

(Danha 2016).

Since milling can also be considered a rate process in which the various size classes

break from the larger sizes to the smaller in a manner analogous to reactor systems,

the AR approach has been extended and successfully applied to comminution. The

power of the AR approach is that it describes the behaviour of different size classes

throughout the milling process, and can represent PSDs as single points in space.

This allows the connectivity of the points to be used for process description and

optimisation purposes.

28
The AR is a fundamental approach that is equipment-independent and can be used

in the field of comminution to describe breakage processes. A complete AR analysis

requires us to describe the set of all possible outputs for a system. The work done by

Khumalo et al (2006) and Metzger et al (2009) used the AR method to optimise

comminution processes carried out under dry conditions only. These researchers

investigated the application of the approach to the milling of a test material, silica

sand for the former and silica sand and quartzite for the latter.

An AR plot is assembled by focusing on the essential features of the process, what

is desired and what is currently available. These can include, but are not limited to,

grinding energy, grinding time and mass fraction of different size classes. Any

number of size classes can be considered, but typically three are used (Danha

2015), grouped as

 the feed size class, which is taken as the top size class or size class 1 (M1)

 the middle size class, which is the result of a moderate extent of breakage

and is termed size class 2 (M2)

 the fines size class, which is the result of a relatively large extent of breakage

and is termed size class 3 (M3)

29
 0.4

0.3
Mass fraction in size class 2

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Mass fraction in size class 1

Figure 2.7: Typical AR plot of mass fraction of material in the intermediate size class
versus mass fraction of material in the feed size class (Danha 2013)

An example of an AR plot is presented in figure 2.7. It is constructed by following the

procedure outlined by Khumalo et al (2006) and Metzger et al (2009) and shows the

mass fraction of material in size class 2 and size class 1 in size class 2 versus size

class 1 phase space. Each point on the graph corresponds to a different duration of

mill operation, starting from a single feed point. This simple plot provides some very

important information on the process, including the following (Metzger et al 2009):

 The boundary curve describes the processes used, and can be interpreted in

terms of pieces of equipment, implicitly identifying the optimal equipment

required.

 The turning point of the curve isolates an optimum solution if the objective is

to maximise the mass in the intermediate size class. This solves the

30
 optimisation problem and provides the process control policy needed to fulfil

that objective, in other words the run time required to achieve that optimum.

2.6 Mineral liberation

Apart from being relatively wasteful of energy, tumbling mills are also inefficient with

regard to mineral liberation because of the indiscriminate nature of the grinding

force. Ideally, comminution forces should be directed at the mineral crystal

boundaries. Accordingly, engineers have made efforts to use tensile forces in an

attempt to draw existing fractures further apart (Hlabangana et al 2017). One of the

earlier versions was the Snyder process (Cavanaugh & Rogers 1974), which was a

method of reducing the size of material by applying air, gas or steam under pressure

in a pressure chamber. The material is broken by the stresses induced when the

sudden release of that pressure causes the material to be ejected along a duct and

into a receiving chamber. This process was believed to improve the liberation of

minerals at generally coarser sizes.

The next attempt to bring about selective liberation of minerals from an ore entailed a

novel non-contact comminution process concept (Parekh, Epstein & Goldberger

1984). This process involved the application of electric and ultrasonic energy in order

to separate minerals from gangue particles. A third attempt at enhancing the

preferential liberation from an ore introduced a new idea developed at the Royal

School of Mines in London (Anon 1986). This comminution process reversed

traditional assumptions by causing rocks to fragment from the inside rather than

fracturing them by compressive forces. This method utilised high-voltage pulses to

generate plasma in the rocks and thus explode them from within.

31
The fourth attempt was based on an idea developed at the Camborne School of

Mines in England (Wills, Parker & Binns 1987), and entailed heat treatment of the

crushed ore prior to conventional grinding. A hard-rock tin ore was heated to 600 oC

before being quenched in water. The differential expansion and contraction

promoted intergranular fracturing, and a significant reduction in grinding resistance

was achieved owing to the micro-cracks produced in the ore minerals. This resulted

in greatly improved liberation of the cassiterite.

The fifth attempt at encouraging preferential breakage and liberation of valuable

minerals was also developed at the Camborne School of Mines in England (Clarke &

Wills 1989), where preliminary test work had shown that high-pressure roller

comminution may improve the liberation of cassiterite from its gangue minerals.

Considerable interest has recently been focused on the high-pressure roller press,

which was developed in Germany. This press is essentially a highly sophisticated

type of rolls crusher capable of generating very high interparticle pressures. There is

evidence to show that liberation can be improved by using these devices, which are

currently being used successfully to reduce the size of cement and limestone, and

they are expected to be useful in grinding industrial ores.

2.7 Mineralogical analysis

A mineral ore is described by a mixture of value minerals and its associated gangue

material. The degree of interlocking, grain size distribution, dissemination, texture

and relative hardness of an ore are collectively known as the deportment of an ore.

32
Knowledge of an ore’s deportment is vital in designing the optimal beneficiation

route.

A great deal of information about an ore is required to enable the engineer to

understand and predict its behaviour during processing.

Techniques such as particle sizing, elemental, diffraction and image analysis are

means of obtaining the information required. Some of the methods used to provide

the quantitative data about mineral composition, association and liberation of the

sample material are petrographic and image analysis. Imaging and image analysis

are fundamental to mineral liberation analysis, and a high resolution image is a

prerequisite for mineral identification and quantification (Fandrich 2007). Optical

image analysis systems are capable of differentiating between mineral species on

the basis of the distinctive reflectivity of different species.

Simple scanning electron microscope (SEM) back-scatter colour image analysis can

acquire images rapidly, but falls short in that it cannot identify the actual mineral

species or quantify them (Benvie 2007). This is why it is normally used in

combination with an energy-dispersive spectrometer equipped scanning electron

microscope (EDS-SEM) to ensure accurate discrimination between mineral species.

Back-scattered electrons (BSE) image analysis, in which a series of BSE images are

collected and processed to produce liberation data, is the most basic liberation

analysis tool.

33
2.8 Worked examples

Example 2.1
a) Complete the following size analysis table:

Sieve Individual wt. Individual Cumulative % Cumulative %

size retained (g) % wt. retained wt. passing

(µm) wt. retained

600 1.25

424 15.52

300 45.1

212 160.4

150 125.2

106 97.7

75 64.3

53 50.6

0 (Pan) 70.5

Total:

b) Plot cumulative % wt. passing versus size, and estimate the 80% passing

size.

c) Draw a Gates–Gaudin–Schumann plot, and determine the values for the size

modulus and the distribution modulus in the Schumann equation.

d) Produce an attainable region plot of M2 versus M1, and comment on the

turning point of the graph.

Example 2.2

34
The size distribution of a dust sample as measured by a microscope is as shown in

the table below. Convert this data to obtain the distribution on a mass basis and then

calculate the specific surface. Assume that the particles are spherical and have a

density of 2 650 kg/m3.

Size range Number of particles in range

0–2 2 000

2–4 600

4–8 140

8 – 12 40

12 – 16 15

16 – 20 5

20 – 24 2

Solution

The mass fraction of particles of size d1 is given by

3
x 1=n1 k 1 d 1 ρs

Where

k1 = Constant

n1 = Number of particles of size d1

ρs = Density of the particles = 2 650 kg/m3

∑ x 1=1, and hence the mass fraction is

35
 3
n 1 k 1 d 1 ρs
x 1=
∑ nk d 3 ρs

In this case:

3
d n nk d ρs x
1 2 000 5 300 000k 0.011
3 600 42 930 000k 0.09
6 140 80 136 000k 0.168
10 40 106 000 000k 0.222
14 15 109 074 000k 0.229
18 5 77 274 000k 0.162
22 2 56 434 400k 0.118
∑ ¿ 477 148 400 k ∑ ¿1.0

The surface mean diameter is given by

d s=
∑ 3
n1 d1
∑ n1 d 21
And hence

2 3
d n nd nd
1 2 000 2 000 2 000
3 600 5 400 16 200
6 140 5 040 30 240
10 40 4 000 40 000
14 15 2 940 41 160
18 5 1 620 29 160
22 2 968 21 296
∑ ¿21 968 ∑ ¿180 056

180 056
Thus :d s= =8.20 µm
21968

36
 π 3 3
The volume of a particle 8.20 μm∈diameter= ×8.20 =288.7 μm
6

2 2
The surface area of a particle 8.20 μm∈diameter=π ×8.20 =211.2μm

2
211.2 μm
¿ hence the specific surface=
288.7 μm3

2
μm
¿ 0.731 3
μm

2
6 m
¿ 0.731 ×10 3
m

37
References

Anon, R. 1986. New ideas in minerals processing. World Mining Equipment, 10:4-19.
Austin, LG & Rogers, RSC. 1985. Powder technology in industrial size reduction.
Powder Technology, 42:91-109.
Bailey, D & Wright, E. 2003. Fiber optic cable construction. In Bailey, D & Wright, E
(eds). Practical fiber optics. Amsterdam: Elsevier.
Beeck, R. 1970. Bestimmung der Antriebsleistung von Rohrmühlen auf Grund von
Betriebswerten. Zement-Kalk-Gips.
Benvie, B. 2007. Mineralogical imaging of kimberlites using SEM-based techniques.
Minerals Engineering, 20:435–443.
Cavanaugh, WJ & Rogers, DJ. 1974. Applications of the Snyder process. Tenth
International Mineral Processing Congress, London.
Chimwani, N, Mulenga, FK, Hildebrandt, D, Glasser, D & Bwalya, MM. 2014. Scale-
up of batch grinding data for simulation of industrial milling of platinum group
minerals ore. Minerals Engineering, 63:100–109.
Clarke, AJ & Wills, BA. 1989. Technical note: Enhancement of cassiterite liberation
by high pressure roller comminution. Minerals Engineering, 2:259–262.
Danha, G. 2013. Identifying opportunities for increasing the milling efficiency of a
Bushveld Igneous Complex (BIC) Upper Group (UG) 2 ore. PhD thesis.
Johannesburg: University of Witwatersrand.
Danha, G, Hildebrandt, D, Glasser, D & Bhondayi, C. 2015. A laboratory scale
application of the attainable region technique on a platinum ore. Powder Technology,
274:14–19.
Danha, G, Legodi, D, Hlabangana, N, Bhondayi, C & Hildebrandt, D. 2016. A
fundamental investigation on the breakage of a bed of silica sand particles: an
attainable region approach. Powder Technology, 301:1208–1212.
Deniz, V. 2004. The effect of mill speed on kinetic breakage parameters of clinker
and limestone. Cement and Concrete Research, 34:1365–1371.
Dong, H & Moys, MH. 2003. Load behaviour and mill power. International Journal of
Mineral Processing, 26:141–151.
Fandrich, R. 2007. Modern SEM-based mineral liberation analysis. International
Journal of Mineral Processing, 84:310–320.
Gupta, RI. 2003. Chemical metallurgy: principles and practice. Weinheim: Wiley.
Harris, CC, Schnock, EM & Arbiter, N. 1985. Grinding mill power consumption.
Mineral Processing and Technology Review, 1:297–345.

38
Hlabangana, N, Danha, G, Hildebrandt, D & Glasser, D. 2016. Use of the attainable
region approach to determine major trends and optimize particle breakage in a
laboratory mill. Powder Technology, 291:414–419.
Hlabangana, N, Danha, G, Bwalya, MM, Hildebrandt, D & Glasser, D. 2017.
Application of the attainable region method to determine optimal conditions for
milling and leaching. Powder Technology, 317:400–407.
Hogg, R & Fuerstenau, DW. 1972. Power relationships for tumbling mills.
Transactions of the Society of Mining Engineers of AIME, 252:418–423.
Horn, F. 1964. Attainable and non-attainable regions in chemical reaction technique.
3rd European Symposium on Chemical Reaction Engineering. Oxford, UK: Pergamon
Press.
Khumalo, N, Glasser, D, Hildebrandt, D, Hausberger, B & Kauchali, S. 2006. The
application of the attainable region analysis to comminution. Chemical Engineering
Science, 61:5969–5980.
Metin, U, Güleç, O & Cingitaş, M. 2015. Wet grindability of calcite to ultra-fine sizes
in conventional ball mill. Particulate Science and Technology, 33:342–348.
Metzger, MJ, Glasser, D, Hausberger, B, Hildebrandt, D & Glasser, BJ. 2009. Use of
the attainable region analysis to optimize particle breakage in a ball mill. Chemical
Engineering Science, 64:3766–3777.
Parekh, BK, Epstein, HE & Goldberger, WM. 1984. Novel comminution process that
uses electric and ultrasonic energy. Mining Engineering, 36:1305–1309.
Redstarmining.indocoal.com. [n.d.]. [Accessed on: 15 November 2018].
Tangsathitkulchai, C & Austin, LG. 1989. Slurry density effects on ball milling in a
laboratory ball mill. Powder Technology, 59:285–293.
Vdocuments. [.n.d.] vdocuments.site. [Accessed on: 15 November 2018].
Veasey, TJ & Will, BA. 1991. Review of methods of improving mineral liberation.
Minerals Engineering, 4:747–752.
Vizcarra, TG, Wightman, EM, Johnson, NW & Manlapig, EV. 2010. The effect of
breakage mechanism on the mineral liberation properties of sulphide ores. Minerals
Engineering, 23:374–382.
Wills, BA. 1990. Comminution in the minerals industry - an overview. Minerals
Engineering, 2:3–5.
Wills, BA. 1991. The development of minerals engineering in the 20th century.
Minerals Engineering, 4:643–653.
Wills, BA. 2005. Mineral processing technology. 5th ed. Oxford: Butterworth–
Heinemann.
Wills, BA & Finch, JA. 2016. Grinding mills. Amsterdam: Elsevier.

39
Wills, BA, Parker, RH & Binns, DG. 1987. Thermally assisted liberation of cassiterite.
Minerals and Metallurgical Processing, 4:94–96.

40