Minerals Engineering 103–104 (2017) 112–126

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

Increasing efficiency by selective comminution

Max Hesse, Oleg Popov, Holger Lieberwirth ⇑
TU Bergakademie Freiberg, Institute of Mineral Processing Machines, Lampadiusstraße 4, D-09599 Freiberg, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Lower ore grades and rising competition require new, more efficient approaches in ore processing.
Received 25 April 2016 Instead of blending different ore varieties into a homogeneous feed for processing plants, selective
Revised 1 September 2016 comminution may be a solution for certain mining operations to increase recovery at lower costs.
Accepted 2 September 2016
Selective comminution exploits selected ore characteristics to achieve an enrichment of valuable
Available online 5 November 2016
minerals in certain fractions. A simple classification step will already produce a pre-concentrate in this
case, while other fractions can be discarded or processed separately. Quantitative Microstructural
Keywords:
Analysis (QMA) provides valuable information on the behaviour of an ore in the comminution process.
Selective comminution
Quantitative microstructural analysis
Ó 2016 Elsevier Ltd. All rights reserved.
Impact comminution
Pre-concentration
lead, zinc, ore

1. Introduction comminution processes (Napier-Munn, 2014). Reducing the speci-

Deteriorating ore grades force mining companies to mine and
process more raw materials to supply sufficient concentrate
amounts to the refining industry. New mining projects often face
challenges of more complex ore textures or more fine-grained
valuable minerals, requiring additional grinding for sufficient
mineral liberation. Already these trends may lead to a four times
higher energy consumption for the comminution of the main metal
ores in the year 2030 compared with today (Norgate and Haque,
2010; Norgate and Jahanshahi, 2011). New deposits are often smal-
ler in size, lower in grade and located in more remote areas than
the depleted ones, which they shall replace (Schodde, 2010;
Peuker et al., 2012). For instance this may contribute more than
50 percent reduction in multifactor productivity (MFP) in the
Australian mining sector from 2002 to 2012 (CRCore, 2014).
While the world population will grow from around 7.1 billion at
present to some 8.7 billion in 2035, GDP is expected to double in
the same period (BP p.l.c., 2015). Resource consumption is more
closely linked to GDP growth than to that of the population
(Haberl, 2012; Krausmann et al., 2009).
While it is recommendable to examine the whole value chain of
metal production from exploration to metallurgy for potential
improvements in the efficiency, mineral processing should be in
the focus. Already at the present level of consumption, about 2%
of the electrical energy generated worldwide is spent on
⇑ Corresponding author.
E-mail addresses: M.Hesse@iam.tu-freiberg.de (M. Hesse), Oleg.Popov@iam.
tu-freiberg.de (O. Popov), Holger.Lieberwirth@iam.tu-freiberg.de (H. Lieberwirth).
fic energy consumption in mineral processing is an obvious task,
considering resource economics as well as the environmental
impact of mineral processing.
Selective comminution is a promising way to reduce the specific
energy consumption. If it is properly applied, it would allow to sep-
arate the portions of material with no or low mineral or metal con-
tent already at an early process stage by a simple classification
step. Thus, a pre-concentrate can be produced. No further grinding
of the earlier discarded material portion is performed, and this can
save a substantial amount of energy.
The pre-concentrate from a selective comminution often con-
tains still locked or semi-liberated particles and bigger amounts
of waste mineral. Since selective comminution in combination
with subsequent classification is a cost-effective way of generating
pre-concentrates, it is also a promising method for the treatment of
low-grade ores and old tailing dumps.
While selective comminution was occasionally applied in
processing ores and industrial minerals in the past (Puffe, 1955;
Taggart, 1956, pp. 14–23), a systematic investigation on the pro-
cess is still missing. A realistic prognosis for new plants based on
the comminution behaviour of the various mineral constituents
of the ore is therefore difficult. Since there are many, sometimes
contradictory, definitions and applications of the term ‘‘selective
comminution” in the literature, this article will provide a general
definition, which considers known material properties as well as
comminution parameters.
In-depth knowledge of properties of the mineral constituents of
an ore is essential for the understanding of selective comminution
phenomena. Geologists characterize ore formations by verbally

http://dx.doi.org/10.1016/j.mineng.2016.09.003
0892-6875/Ó 2016 Elsevier Ltd. All rights reserved.
 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126
describing their texture and structure based on the analysis of thin
or polished sections. This method has limitations as it describes the
ore microstructure usually two-dimensionally and qualitatively
only. The search for systematic relations between relevant process-
ing parameters, product characteristics and inherent textural attri-
butes, however, requires quantification of the microstructural
description of raw materials. Therefore, Quantitative Microstruc-
tural Analysis (QMA) was developed to support the process engi-
neer in selecting the appropriate type and size of machine and
determining the proper operating parameters for the most efficient
liberation and recovery.
2. Quantitative microstructural analysis of the ore
A wide range of mineral analysis methods supports mineralo-
gists today, such as the Mineral Liberation Analysis (MLA), QEMS-
CAN or computer tomography (CT). While each of these methods
has provided important new insights into the composition of min-
erals, traditional optical microscopy still maintained its impor-
tance in many applications. It allows the direct identification of
minerals rather than just chemical elements for instance. Within
the limits of optical resolution, thin sections in an optical micro-
scope may provide plenty of valuable information for efficient min-
eral processing. For this purpose, QMA is a method to structure the
information in a mathematic way. Some similarities may also be
established with methods applied for the analysis of polished sec-
tions of steel or cast iron. The details of the application as well as
most of the conclusions drawn are however different. This may
be compared with the application of the Vickers indenter used
for hardness testing. While in metallurgy the dimensions of the
indentation are measured, the length of the Palmqvist cracks
(Palmqvist, 1957; Szutkowska, 2005) are of superior interest in
mineralogy. The quantitative figures on fracture toughness of each
Fig. 1. Determination of mineral characteristics using QMA (process flow-chart, Popov, 2007).
113
mineral constituent derived as well as the behaviour of the cracks
at the grain boundary provide valuable information complement-
ing the QMA data of a sample.
A microscopic analysis of thin or polished sections forms the
basis of the QMA (Fig. 1). As ore minerals mostly have a complex
three-dimensional texture, the information that can be derived
from two-dimensional cut surfaces is often insufficient for a spatial
quantitative characterization of ores.
The three-dimensional description of minerals and ore
microstructures within QMA allows a quantitative characterization
of complex geological materials (Lieberwirth et al., 2014). The
results of mathematical-petrographic ore characterization using
statistical models are important for predicting relevant product
properties or the behaviour of an ore under certain comminution
loads. They allow to quantify correlations between properties of
the ore and certain parameters of the beneficiation process. One
of these properties may be the suitability of an ore for selective
comminution.
For QMA, a hand specimen is collected from exposures in the
field with detailed informations about the sampling location and
the geological context. Three orthogonally orientated thin or pol-
ished sections of the sample are then prepared and investigated
with optical microscopy using stereological methods (Fig. 2). The
first objective of any examination is usually the identification of
the minerals present. A variety of properties exhibited by each
mineral phase can be studied using optical microscopy.
The photomicrographs are analyzed in successive steps utilizing
point counting, linear analysis as well as area analysis (Fig. 3). The
data obtained for each of the three thin sections are used to estab-
lish area distribution histograms as well as so called ‘‘intersection
rose diagrams” (De Hoff and Rhines, 1968). The latter represents
the distribution of the number of intersections per unit length in
varying directions.
114 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126

Fig. 2. Preparation of thin sections.

Fig. 3. Stereological methods (a – point analysis, b – linear analysis, c – area analysis).

Based on the 2-dimensional information from the three thin or
polished sections it is possible to synthesize a 3-dimensional
model by mathematical transformation and approximation
(Popov, 2007). In this manner, the spatial microstructure of the
material is quantified. Characteristic figures are determined for
the mode (volume percentages of the mineral phases), for the
texture (size, shape, roughness) as well as for the structure (orien-
tation, distribution, space filling; Fig. 4).
The analysis of the ore microstructure prior to breakage can
provide valuable information regarding the types of textures and
mineral assemblages that are likely to produce certain particle
sizes and characteristics that are favourable or unfavourable for
selective comminution.
2.1. Volume percentages of the minerals
The volume percentage of a mineral group is defined as the quo-
tient of the sum of the volumes of the individual mineral grains in
the sample and the total sample volume. It can be most easily
determined with the point counting method (Glagolev, 1934).
 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126 115

Fig. 4. Characterization of ores.

Fig. 5. Mineral composition (e.g. lead-zinc ore from Hermsdorf, Germany).

The percentages of the constituents are generally shown in the
form of a table (Fig. 5).
Volume percentage or derived mass percentage of the ore con-
stituents before a certain comminution step, serves as a reference
for any evaluation of selectivity achieved. Even comminution prod-
ucts may be investigated using this method after embedding the
particles in epoxy.
2.2. Grain size
Knowledge of the grain sizes in an ore is essential for judging
the potential for selective comminution, since differences in parti-
cle sizes after comminution is usually the simplest way to separate
an enriched fraction by simple classification. The size of the indi-
vidual grains varies in the ore’s microstructure, so the size of the
grains has to be approximated with a size distribution function.
Accordingly, the distribution density of the grain diameter d of a
logarithmic normal distribution function can be described with a
median d50,3 and a scatter parameter rln (Fig. 6).
2.3. Grain shape
The shape is a geometric grain characteristic that has to take all
three dimensions into account (Unland and Folgner, 1997). The
shape of a grain can be regarded approximately as an ellipsoid with
the main axes a, b and c, where a P b P c. The relationships of the
main axes to each other describe the form of the grains (Fig. 7).
For the consideration of selectivity in comminution, the grain
shape may have an impact on the sensitivity of an ore constituent
against certain loads during comminution. A mineral forming
mainly platy grains, might preferably be ground by applying shear
loads supporting the platy shape. Then, an air classifier could sup-
port the separation of the different shaped particles, taking into
account the constraints that may be imposed by the density of
the various minerals.
116 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126

Fig. 6. Grain size of the minerals (e.g. granites).

Granite Basalt Gneiss

cubic needle-shaped platy

c c c

c b b
b
b

a
a a a
E=a/b
Phase related features
Rock characterisc features
F=b/c Granite Basalt Gneiss
Elongaon E - 1.095 3.908 1.000
Flatness F - 1.061 1.365 1.834

Fig. 7. Different grain shapes, variation of the main axes.

2.4. Roughness
The ellipsoid serves as reference also for the determination of
the roughness since it is the geometrical shape with the smallest
surface relative to the volume of a body with three different
dimensions in orthogonal directions. Any increase in surface rela-
tive to this optimum shape is considered as increase in roughness
of the surface. The roughness KR is defined as the ratio of the differ-
ence between the ‘‘real” surface area SV(R) and the ellipsoid ‘‘ideal”
surface area SV(I) to the ‘‘real” surface area SV(R) of the individual
mineral phases (Fig. 8).
The calculation of the ‘‘ideal” surface area SV(I) is based on the
grain size distribution with consideration of the grain shape and
the mineral phase volume percentage. The ‘‘real” surface area SV
(R) is calculated from the 3-dimensional rose of intersections
(Kurzydłowski and Ralph, 1995).
The surface area per unit volume is especially useful as it is
linked with the determination of contacts between mineral phases
within ore. This parameter is easily calculated from the number of
intersections per unit length of test line applying one of these
stereological relationships.
The outer shape of the grains largely determines the character
of the microstructure. In selective comminution, roughness may
play an important roll through its influence on the shear strength
and fracture toughness of an ore, in particular along the grain
boundaries.
2.5. Orientation
The linear analyses are done at varying angles (usually by
rotating the stage clockwise by 15° with each step). With every
new position, all the intersections of lines with the grain
 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126 117

Sandstone Quartzite

SV(I) SV(R)

SV(R) - SV(I)
KR =
SV(R)

Phase related features

Rock characterisc features
Sandstone Quartzite
Quartz Qtz % 100 100
Mean diameter d50,3 mm 0.289 0.388
Roughness degree KR % 5 29

Fig. 8. Degree of roughness of the minerals.

Fig. 9. 2D-intersection roses.

boundaries are counted (Saltykov, 1974; Schumann, 1991). The
result contains valuable information in the form of 2D-
intersection roses (Fig. 9).
From three 2D-roses of intersections which are orthogonal to
each other, with an approximation, the parameters of a spatial
rose of intersections (nis, nlin and nfl) and their orientation angle
in the space (a, b and c) are calculated. The spatial rose of inter-
sections of an ideally isometric boundary surface system is a
sphere with diameter nis, that of an ideally linear-oriented sys-
tem is a toroid with diameter nlin and for an ideally area-
oriented system it is a double-sphere with the diameter nfl
(Saltykov, 1974) (Fig. 10).
A spatial rose of intersections of an oriented microstructure, is
the superposition of the roses of intersection of ideal boundary sur-
face systems (Fig. 10). From the parameters of the spatial rose of
intersections, the orientation degrees (Kis, Klin and Kfl) of a spatial
arrangement (percentages of linear- and area-oriented as well as
non-oriented grain assemblages) can be derived.
2.6. Distribution
Grains of a mineral phase can be evenly distributed in the space,
but they can also form clusters in which the grains of a mineral
phase share boundary surfaces (Fig. 11). The degree of clustering
C is defined as the quotient of the sum of the boundary surfaces
between the grains of the same mineral group and the total bound-
ary surface of this mineral group (De Hoff and Rhines, 1968; Exner
and Hougardy, 1986; Gurland, 1966). The degree of clustering can
be calculated with the help of linear analysis. In selective com-
minution, the knowledge on clustering is important to predict
the size fraction which might show an enrichment of a certain
ore constituent.
2.7. Space filling degree
Within a raw material there are not only solid phases, but also
pores, fissures. The latter are filled with gas and liquids. The
118 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126

Fig. 10. Schematic showing the superposition principle for an oriented microstructure.

Granodiorite Granite
β
β β

α α α
α β
a b

Phase related features

Rock characterisc features
Granodiorite Granite
Quartz Qtz % 35 32
Feldspar Fds % 41 55
Mica Mc % 24 13
Mean diameter d50.3 mm 0.518 0.437
Degree of clustering C % 16 49

Fig. 11. Distribution of the grains.

volumetric portion of those gas and liquid filled volumes can be
calculated and used to determine the space filling of solid phases
eVF. It can influence the strength of the material. If only certain
mineral phases are highly porous, this may be used in selective
comminution.
2.8. Ore characteristics table
After evaluation of all mineral groups present in the ore, the
results of the evaluation are summarized in the form of character-
istic values (Table 1). The ore characteristic values are first
 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126 119

Table 1
Data sheet ‘‘Ore characteristics”, brecciated ore type.

Raw material Kind: Sphalerite-galena Mineral related features Calculated features

of raw material
Deposit: Hermsdorf Location: Saxony,
Germany
Mode Phases Kind Unit Galena Sphalerite Calcite R microbodies
Content Volumetric Portion eV % 0.4 4.7 94.9 100
Fabric Texture Size Mean diameter d50,3 mm 0.213 0.245 – 0.232
Deviation rln – 0.314 0.360 – 0.341
Grain surface Specific surface SV mm2/mm3 31.460 30.470 – 30.880
Shape Elongation E – 1.371 1.365 – 1.367
Flatness F – 1.000 1.000 – 1.000
Roughness Roughness degree KR % 71 56 – 62
Structure Orientation Degree of linear orientation Klin % 23 22 – 15
Degree of area orientation Kfl % 0 0 – 0
Degree of isotropic orientation Kis % 77 78 – 85
Distribution Degree of clustering C % 0 0 – 0
Space filling Space filling degree eVF % – – – 100

No
SC!

Material SC! Suitable for SC Machine

No
SC!
„SC“ - Selecve Comminuon

Fig. 12. Selective comminution (SC) as result of the proper combination of a material with inherent suitability and an appropriate comminution technology.

determined independently from specific applications. Then, they
are interpreted specifically to the aim of the comminution case,
to the foreseen comminution machine and the load it applies to
the ore.
The test results presented in the following deliver statistical
relationships between the ore properties (e.g. grain size, grain
shape, mineral intergrowth, etc.), selected machine parameters
(e.g. stress loading rate) and the achievable product parameters
(particle size and particle shape distribution). Not only could a
weighting of the influence of the different ore characteristic values
on individual target values be established, statistical relationships
between selected target values and the various design and operat-
ing parameters could be proven as well. The ore characteristic
values permit an estimation of the ore comminution behaviour
with respect to selectivity, crushability, product particle shape,
wear and energy consumption.
3. Selective comminution
The term ‘‘selective comminution” as used in this article
describes a method to exploit differences in the comminution
behaviour of the mineral phases of an ore. Even if there are such
differences, selectivity is only achieved if the comminution tech-
nology can exploit those differences by the kind, speed, frequency,
energy, etc. of the load it exercises onto the material. Selective
comminution may also occur unintentionally, even with adverse
influence on the mineral processing system by the incidentally
combination of a material suitable for selective comminution with
a technology exploiting that characteristics. So, selective com-
minution can also be considered as a property of a comminution
system comprising of the appropriate comminution technology
for the inherently suitable material (Hesse, 2014). While selectivity
may occur with various comminution technologies, such as blast-
ing, this article refers specifically to technologies using comminu-
tion machines.
In addition to suitable material characteristics, an appropriate
technology is fundamental to exploit those characteristics for
selective comminution. The design of the comminution machine
and its operational parameters such as the kind of load, its
frequency, velocity and energy take decisive influence on the
comminution system. Selective comminution can be expected only
if both, material properties and comminution technology are
suitable (see Fig. 12). If applicable, however, it may be an efficiency
120 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126
increasing step in the separation process of valuable minerals from
waste, ore from bedrock or different ore varieties of a deposit from
each other.
In most of the reported or investigated occurrences of selective
comminution, the mass specific content of valuable minerals in
certain size fractions of the comminution product is different from
the one in the feed material. In this case, the size serves as separa-
tion criterion between the fractions which are rich in valuable min-
erals from the rest (Gaudin, 1939; Schmidlapp, 1965). Other
separation criteria such as particle shape are also described in
the literature (Andreas and Torlach, 1958; Stöhr, 1976).
The selective comminution helps only to realize the differentia-
tion in the size of comminution products having differences in the
specific content of valuable products but with no separation. A
downstream separation step is always necessary for the separation
of different materials. In the simplest case, this is a classification.
When selecting a comminution technology, knowledge on crack
propagation within the feed material is essential. Numerous
descriptions of the crack propagation mechanisms can be found
in the literature, often using various terms for the same phe-
nomenon (Fandrich et al., 1997; Gaudin, 1939; Müller, 2011;
Puffe, 1960; Schmidlapp, 1965; Sutherland and Fandrich, 1996;
Taggart, 1956). Cracks may propagate in the ore microstructure
essentially in two fundamentally different ways, either through
the mineral grains or along the grain boundaries. Propagation
through the mineral grains is typical for preferential breakage
(Sutherland and Fandrich, 1996). Crack propagation along the grain
boundaries characterises the interfacial breakage. The properties of
the grains as well as the grain boundaries mainly determine the
way of crack propagation. Processing parameters may influence
the cracking behaviour determined by the material properties to
a certain extent. An overview about various application options
of selective comminution is shown in Table 2.
Table 2
Application examples of selective comminution.
Kind of fracture Case Target
Grain-boundary fractures A High degree of liberation
(interfacial breakage)
B Removal of superficially adhering
components
C Comminution into fragments with
different particle shapes
Grain-fractures D Selective comminution with
(preferential breakage) enrichment by classifying
E Liberation of valuable materials
at the surface
F Selective comminution with reshaping
4. Methods to evaluate selective comminution
Selective comminution in combination with a process, which
separates fractions containing different percentages of valuable
components, can be considered as a sortation process. The method
for measuring the effect of selective comminution depends on the
kind of selective comminution and the separation process. The
most common case of selective comminution presents as a differ-
ent distribution of valuable and waste minerals in dependence
on the particle size. Thus, particle size is the most important
parameter for the evaluation of the selective comminution, as
shown in the following three examples.
4.1. Integration method with cumulative particle size distributions
This integration method is independent of any separation cuts
and provides only one evaluating parameter, which makes it easy
to compare results from various test series.
Material components to be separated after selective comminu-
tion are subsequently named V for valuable component and W for
waste component (index V and W respectively). For each compo-
nent, a cumulative particle size distribution Q3 is calculated with
the mass md of each particle size class d and the content cd of
the component in each particle size class, see Fig. 13(a) and
Eq. (1.1). Finally, four Q3 functions (see Fig. 13(b)) need to be
calculated for the evaluation. Of the four functions, two functions
present the cumulative particle size distribution of valuable com-
ponents and waste in the feed and the remaining two functions
for the product.
R dx
d¼d0
md  cd
Q 3 ðlg dx Þ ¼ R dn  100% ð1:1Þ
d¼d0
md  cd
Graphical example Literature
Fandrich et al. (1997) and
Gaudin (1939)
Fandrich et al. (1997),
Puffe (1960) and
Sutherland and Fandrich (1996)
Stöhr (1976), Puffe (1955) and
Andreas and Torlach (1958)
Schmidlapp (1957) and
Schmidlapp (1965)
Miller (2003)
Müller (2011) and Gründer (1957)
 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126 121

100 100

80 80

Q3 [%]

Q3 [%]
60 60

40 40
Q3 (lg dx)
20 20

0 0
d [mm] 1 10 100 0,0 0,5 1,0 1,5 2,0
lg d
lg d 0,0 1,0 2,0 selectivity of feed material (SF)
selectivity of comminution product (SP)
Q3 of valuable component in product
d0 dx dn Q3 of waste component in product
Q3 of valuable component in feed
(a) (b) Q3 of waste component in feed

Fig. 13. (a) Example of cumulative particle size distribution Q3; (b) example of Q3 with selectivity of the feed material and comminution product between the valuable and
waste material.

The selectivity S of two components describes the difference in
a considered class feature (here particle size) of these two compo-
nents. It is calculated as the area between the cumulative particle
size distributions of the two components and may, for example, be
described for the comminution product (index P) with Eq. (1.2).
The calculation requires the logarithm of the particle size. In
this case the selectivity of the product SP (light shaded area in
Fig. 13(b)) is the area between the Q3 functions of the waste
(Q3,WP in Eq. (1.2), dashed line in Fig. 13(b)) and the valuable
component (Q3,VP in Eq. (1.2), dot-dash line in Fig. 13(b)).
R Rn 
n
g¼1
Q 3;VP ðlog dÞ  g¼1
Q 3;WP ðlog dÞ
SP ¼
log dg¼n  log dg¼1
It should be noted that, when evaluating the selectivity of a
comminution process even in the feed material (index F) the com-
ponents V and W may exhibit differences in the cumulative parti-
cle size distributions and reflect a selectivity of the feed material SF
(Fig. 13(b), dark shaded area). If this selectivity of the feed material
SF remains unchanged in the comminution product SP, no selective
comminution occurred. The magnitude of the change, SZ, is a quan-
titative value for the degree of selective comminution, Eqs. (1.3)
and (1.4).
SZ ¼ SP  SF
SZ can be determined by inserting Eq. (1.2) for SP and the
respective term for SF in Eq. (1.3). For the discretization of cumula-
tive particle size distributions, the values g and f are the numbering
of the screening cuts, ascending with larger mesh size.
R Rn  The parameters are cF - concentration in the complete feed
n
g¼1
Q 3;VP ðlog dÞ  g¼1
Q 3;WP ðlog dÞ
SZ ¼
log dg¼n  log dg¼1
R m Rm  component. Concentrate and waste are split by the separation
Q 3;VF ðlog dÞ  f¼1 Q 3;WF ðlog dÞ
 f¼1
log df¼m  log df¼1
SZ is positive in the example shown in Fig. 13(b), despite the
fact that V is already slightly enriched in the fine fractions of the
feed material, since this enrichment in the fine fractions further
increases with comminution. Selective comminution can also be
used to homogenize a material if the components occur in different
particle size distributions in the feed material and are equally dis-
tributed in the comminution product. In this case, SZ is negative.
An interpretation of SZ and conclusions with regard to beneficia-
tion processes should therefore always take into account the selec-
tivity of the feed material SF and the comminution product SP.
4.2. Ore separation degree gore
Another option for characterizing a selective comminution
effect and the selectivity of different materials in the product is
the ore separation degree gore, Eq. (1.5) (Schulz, 1970; Schubert,
1989).
gore ¼ RV  RW ð1:5Þ
ð1:2Þ
gore indicates the difference between the recovery of valuable (RV)
and waste material (RW) in the product, both as percentage. RV
represents the mass recovery of valuable component in the size
fraction below a certain separation cut td relative to the mass of
valuable component in the feed. RW represents the waste material
portion in the same size fraction relative to the mass of the waste
portion in the feed. If the concentration of valuable component
depends on the particle size of the product, then gore changes and
depends on the separation cut td respectively.
If just the concentration of the valuable component in the feed
(F), the product (P) and the waste (W) are known, gore can also be
ð1:3Þ determined by using Eq. (1.6). This equation can be found in
similar form in literature (Schubert, 1989; Taggart, 1956,
pp. 19–200; Wills, 1992).
ðcF  cW Þ  ðcP  cF Þ  100%
gore ¼  100% ð1:6Þ
cF  ð100  cF Þ  ðcP  cW Þ
material, cP - concentration in the product and cW - concentration
in the waste material, respectively as percentage of the valuable
ð1:4Þ cut td, see Fig. 14(a). The values of cW and cP change in dependence
on td, see Fig. 14(b). With this dependence on td it is simple to find
the maximum value for the ore separation degree gore,max and the
best fit for a separation respectively, see Fig. 14(c), (Hesse and
Lieberwirth, 2014; Hesse, 2014). This calculation method can also
be applied to determine the selectivity of the feed material. There-
fore, it has to be classified only, without comminution. The differ-
ence between gore of the feed material and of the comminution
product can be used to evaluate the selective comminution.
122 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126

Component V Component W 90

ore concentration cP and cW [%]

Mass portion [%]
40 ccK
P
80
30 70 cW
cB
20 60
10 50
0 40

1/2

2/5

5/10

10/15

15/20

20/30
30
20
Fraction [mm]
10
0
concentrate (cP) td waste (cW)
1 10
seperation cut td [mm]
(a) all (cF) (b)

60
ore separation degree ηore [%]

50

40

30

20

10

0
1 10
seperation cut td [mm]
(c)
Fig. 14. (a) Example for distribution of the mass portions for the components V and W; (b) distribution of the ore concentration from (a) in the concentrate cP and in the waste
cW in dependence of the separation cut; (c) graph of the ore separation degree gore for example (a).

4.3. Recovery plots The recovery of the valuable material and the waste material can
be plotted in a diagram for various separation cuts td, see Fig. 15,
A further method for evaluating the effect of selective com- for either the fine or the coarse fraction. Fig. 15 shows the results
minution is to use the Fuerstenau upgrading diagram in recovery for the fine fraction. A linear diagonal line would indicate that
plots (Leißner et al., 2014; Reichert et al., 2015) on the classified there is no selectivity in the investigated material (line 3 in
comminution product. For the application of this method, the Fig. 15). An enrichment of the valuable component in the fine
selectively comminuted material is always separated into two fractions is documented by a recovery curve positioned above
fractions, a fine fraction and a coarse fraction by td as in Fig. 15. the diagonal line (line 1 in Fig. 15). A recovery curve below the
diagonal line (line 2 in Fig. 15) documents an enrichment of the
waste material in the fine fractions. In this case, the enrichment
100
of the valuable component occurs in the coarse fraction. This
evaluating method has similar advantages and disadvantages as
ore mineral recovery [%]

80 the ore separation degree gore. However, different parameters

60
40
20
0 load and shear load. A pneumatically operated shoot apparatus is
0 50
waste recovery [%]
Fig. 15. Recovery plot as Fuerstenau upgrading curve (if applied for the fine
fraction: Line1: selective comminution with enrichment of the valuable component
in the fine fraction; Line 2: selective comminution with enrichment of the valuable
component in the coarse fraction; Line 3: no particle size related selectivity).
1 can be derived from these functions (Leißner et al., 2014).
2
5. Application example of selective comminution
3 The following example shows some results of selective com-
minution tests with a lead-zinc-ore (Stöhr, 2012; Hesse, 2015).
The investigations are designed to study the effect of selective
comminution at different kinds of load: impact load, compressive
100 used to execute the single particle comminution with impact load.
A double roll crusher is used for the compressive load investiga-
tion. A disc mill is used to comminute the single ore particles
mainly with shear load.
The ore originates from the marble mine Hermsdorf in the
Erzgebirge Mountains, Germany. Six separated calcite/dolomite
 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126 123

horizons are located in the deposit. The mainly dolomitic, partly
calcitic horizon K1 contains a larger concordant lead-zinc-
mineralisation consisting of two types of ore. One type is banded,
the other type is as brecciated as a part of the dolomite in the
K1-horizon. The brecciation leads to a coarser recrystallization of
a part of the ore mineralisation. The wall rock is phyllite. The whole
deposit contains a lot of larger dislocations. Since the 16th century
marble was mined in the deposit. In 1956 lead-zinc ores were sent
to the smelter in Freiberg. An intensive prospection of the
lead-zinc-mineralisation took place in the years 1958, 1960 and
1963. The high variability of the thickness of the ore layer and
the relatively low amount of ore prevented the economic ore min-
ing. The lead-zinc-mineralisation is largely triaged by the marble
mining (Hoth et al., 2010).
Both ore types were investigated. Some results of the investiga-
tions with the brecciated ore type are shown below. First, the
texture and structure properties were determined with the QMA
(Popov, 2007) on polished sections, see Fig. 16. The galena and
sphalerite are mostly fine crystalline grains in a calcitic matrix.
The average grain size is about 0.21 mm for sphalerite and 0.25
for galena. The calcite grain size could not be evaluated from the
polished section. The literature gives values of 0.03–0.1 mm
(Hoth et al., 2010). For a quantitative evaluation of the clustering,
a larger number of polished sections is required.
In addition to QMA, the Vickers hardness number and fracture
toughness were measured to further quantify the difference in
the comminution properties, see Table 3. The fracture toughness
was derived from the Palmqvist crack length. Cracks could be
generated with the Vickers indenter only in the calcite. The low
hardness indicates calcite. The point-load-index IS(50) = 4.04 MPa
for the ore was determined as per ISRM (Broch and Franklin,
1972; Brook, 1985; Chau and Wong, 1996; ISRM, 1985) with a
variation coefficient of 23.5%. The unconfined compressive
strength of the ore can be derived respectively to be around
101 MPa. The density of the ore is about 2.84 g/cm3. The ore con-
tent is about 4.7% for sphalerite and 0.4% for galena.
The results of the QMA investigations in combination with the
Vickers hardness tests allow the prognoses that a good selectivity
can be expected for galena, since its hardness is much lower than
that of the other components. Its comparable high roughness how-
ever leads to the conclusion that the breakage will be rather of
preferential than of interfacial nature. For the sphalerite, much less
selectivity is to be expected as result of the comminution process.
The part of sphalerite locked with the galena will be enriched
together with the galena in the fine fractions. The low roughness
between galena and sphalerite is advantageous for liberating both
minerals. The other part of sphalerite locked with the calcite with a
high roughness and more often smaller grained will be hardly
enriched in the fine fractions by selective comminution. It will stay
locked or semi-liberated with the calcite in the coarse fractions.
The minimum amount of sample mass was calculated to ensure
a representative sample (Schubert, 1984) based on a confidence
level of 80%. Finally, the representative sample was generated with
a sample divider. Only the narrow fraction 10/12.5 mm was chosen
for the tests in order to eliminate influences of selectivity in the
feed material on the evaluation of selectivity in the comminution
process.
Some results from the investigation with single particle impact
comminution in a shoot apparatus will be shown first. The sample
contains 499 particles. The particles were crushed individually in a
pneumatically operated shoot apparatus by perpendicular impact
load at 60 m/s against a hard metal plate. The comminution pro-
duct of all particles was classified (see Fig. 17) and the lead and
zinc content were determined for each size class. The ore separa-
tion degrees were calculated separately for galena and sphalerite
according to Eq. (1.6) for different separation cuts td.
Further tests were conducted with other samples of the same
material on a double roll crusher (compressive load), a pin mill
(impact load with high loading frequency) and a disc mill (shear
load with subordinated compressive load). The samples for the
tests on the other machines were prepared in the same way as
for the shoot apparatus. The gap of the double roll crusher was
set to 0.25 mm and the peripheral velocity to 2.7 m/s. The smooth
rolls have a diameter of 0.45 m. The pin mill has one rotating plate
with two lines of 4 pins each. Between and outside of the rotating
pin lines are grizzlies with the smallest mesh size of 10 mm. The

Fig. 16. Polished sections of the brecciated concordant lead-zinc-mineralization in the K1-horizon, Hermsdorf marble deposit, Erzgebirge mountains, Germany from QMA
analysis; light grey: galena; middle grey: sphalerite; dark grey: calcite; image width: 1.6 mm.

Table 3
Properties of sphalerite, galena and dolomite; average ± standard deviation (*logarithmic standard deviation, see Chapter 2.2).

Mineral Content Grain size* Vickers hardness number Specific surface Roughness Fracture toughness
(%) (mm) (0.025 HV) (N/mm2) (mm2/mm3) (%) (MN/m3/2)
Sphalerite 4.7 ± 0.12 0.25 ± 0.36 195 ± 21 30 56 –
Galena 0.4 ± 0.08 0.21 ± 0.31 75 ± 4 31 71 –
Calcite 95 0.03–0.1 159 ± 11 – – 0.53 ± 0.06
124 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126

Cumulative passing distribution Q3 [%]

100
Feed material 10/12.5 mm
90
80 Shot apparatus
70
Pin mill
60
50 Roller crusher
40
Disc mill
30
20 Average grain size galena
10 Average grain size sphalerite
0
0,1 1 10 100
Mesh size d [mm]
Fig. 17. Particle size distribution of the comminution products from different machines, brecciated ore.

Shot apparatus Roller crusher

Pin mill Disc mill
Average grain size sphalerite Average grain size galena

20 50
ore separation degree ηore [%]
ore separation degree ηore [%]

40
15

30

10

20

5
10

0 0
0,1 1 10 100 0,1 1 10 100
(a) separation cut t d [mm] (b) separation cut d t [mm]

Fig. 18. Ore separation degree gore for the brecciated ore (a) sphalerite; (b) galena.

peripheral velocity at the outer pin line is 32 m/s. The gap between to a high increase of the lead content from 0.4% in the feed material
the discs was set to 0.3 mm and the relative velocity of the disc mill to 4.3% in the fractions <0.125 mm in the product of the shoot
at the outer rim of the 200 mm discs was 4.5 m/s. apparatus. Here, the fraction <0.125 mm contains 15% of the galena
The reduction ratio e is defined with mesh size d50 respectively in only 2% (mass) of the product. The coarsest fraction >4 mm,
at 50% of the Q3-functions of the feed material (F) and the com- making up 58% of the mass of the product, has a content of
minution product (P), see Eq. (1.7). 0.14% galena only. This clear displacement of the galena into the
fine fractions leads to the high ore separation degree gore shown
d50;F in Fig. 18(b).
e¼ ð1:7Þ The best ore separation degree for the galena enrichment can be
d50;P
achieved with the shoot apparatus in coarse fractions at td = 4 mm,
For the comminution with the shoot apparatus e is about 2. Fig. 18(b). The selective comminution effect at the pin mill starts
Thereby, only about 5% (mass) of the comminution product is in also at coarse fractions, increases continuously with decreasing td
the same range as the grain size of galena and sphalerite, see and reaches a maximum at the finest fraction, Fig. 18(b). The roll
Fig. 17. The comminution at the other machines generates a more crusher shows a similar development, but at much lower values.
finely disperse product. The products from the roll crusher and the All three distributions show no significant change near the grain
disc mill show a similar particle size distribution. The finest pro- size of the galena (Table 3). Two interpretations are possible: First,
duct comes from the disc mill. The selective comminution leads the selective comminution effect comes mainly from the
 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126
preferential breakage. Hence, the selective comminution leads to a
preferred crack propagation in the galena. In the coarse fractions
dominates the case E of Table 2. That changes with smaller
fractions where the case D dominates. This can be explained by
the relative low hardness, low toughness due to the cleavage and
higher roughness, information retrieved by the QMA and the
Vickers hardness test, and in this case only qualitatively evaluated
mineral clusters of galena. Second, the selectivity during breakage
highly depends on the kind of load. This explains the significant
difference in the ore separation degree curves for the different test-
ing machines, which introduce different kinds of load into the
material. Only the ore separation degree of the disc mill product
shows a significant change in the range of the galena grain size
and a constant level at smaller separation cuts. The gap between
the discs (0.3 mm) seems to prevent further comminution of the
galena. Thus, a higher portion of interfacial breakage may be
possible.
As expected from QMA and Vickers hardness test, the selective
comminution effect for sphalerite is much smaller; gore reaches a
maximum of 10% only, Fig. 18(a). One reason is the much lower
difference of the Vickers hardness value between sphalerite and
calcite. The biggest changes of gore appears in the range of the
sphalerite grain size (red1 dotted line). There are again two interpre-
tations: First, the selective comminution of sphalerite seems to come
mostly from interfacial breakage. This can be explained by the high-
est hardness among the three minerals, the typical higher toughness
in comparison to galena and the well rounded grain shape, quanti-
fied with a lower roughness especially of the sphalerite grains in
the galena. The Vickers hardness difference between sphalerite and
galena is much bigger than between sphalerite and calcite (Table 3).
For sphalerite an interfacial breakage can be expected preferably at
the grain boundary to the galena. This can explain the sharp increase
of gore with a decreasing td below the sphalerite average grain size,
see red dotted line in Fig. 18(a). The selective comminution of the
sphalerite is a by-product of the selective comminution of the
galena. In the range of the grain size, case A of Table 2 may dominate
in combination with the galena. With smaller particle sizes, case D
may become more and more important. Second, the effect is nearly
independent from the kind of load. The maximum of the gore for
sphalerite at the shoot apparatus results from the sphalerite locked
in the galena. The sphalerite is also enriched in the fractions (1 until
4 mm) with the selective comminution of the galena. If the spha-
lerite locked in the galena is neglected, the distribution would be
similar to the ones realised with the other comminution machines
shown in Fig. 18(a). The best result from the disc mill is achieved
in the finest fraction.
6. Conclusions
The main aim of selective comminution is the separation of the
valuable content from waste material, thereby relieving the subse-
quent beneficiation process. The different comminution behaviour
of ore constituents is exploited to generate a product, which shows
an enrichment of the valuable content in certain fractions. With
particle size often being the preferred parameter to differentiate
these fractions, a simple classification step may be sufficient in
those cases to separate a waste fraction and produce a pre-
concentrate.
In order to select the appropriate kind and magnitude of load to
be applied and thus the most suitable comminution machine, the
mineral has to be investigated. With the QMA the paper presents
a proven scientific approach to characterize ores in a mathematic
1
For interpretation of color in Fig. 18, the reader is referred to the web version of
this article.
125
way, oriented to the terms of geology to identify opportunities for
selective comminution. By determining the texture (grain size,
grain shape, roughness, specific surface) of the minerals and their
structure (orientation, clustering) in an ore and in combination
with further parameters such as Vickers hardness, toughness, Is
(50) strength index or density, a method is generated to support
the selection of the comminution machine and the operation
parameters of the system.
First results in this new field of scientific investigation into
selective comminution are presented. Parameters are defined to
quantify selectivity S and ore separation degree gore. The selective
comminution factor SZ quantifies the change in selectivity
between feed material and product.
Test results with a lead-zinc-ore confirm that selective
comminution requires a systemic approach, comprising of the
comminution behaviour of the ore under a certain load, the type of
comminution machine and a number of operation parameters of
the system. Further investigation is required to establish correla-
tions between the quantified ore properties, the machine properties
and the selective comminution results, for a better understanding of
the influence of certain ore characteristics on its comminution beha-
viour under various loads. Another task is to relate the new param-
eter sets to indices widely used in the resource industry nowadays
such as UCS or the various Bond indices. The new approach was
applied and tested with a number of ores already. Future investiga-
tions with materials from other sources, however, will certainly help
to refine the method.
References
Andreas, E., Torlach, A., 1958. Prallmühlen in der Asbestaufbereitung. Erzmetall
Band XI (7), 333–335.
BP p.l.c., 2015. BP Energy Outlook 2035 Retrieved from <http://www.bp.com/
en/global/corporate/about-bp/energy-economics/energy-outlook.html>.
Broch, E., Franklin, J., 1972. The point-load strength test. Int. J. Rock Mech. Min. Sci.
Geomech. 9, 669–697.
Brook, N., 1985. The equivalent core diameter method of size and shape correction
in point load testing. Int. J. Rock Mech. Min. Sci. Geomech. 22, 61–75.
Chau, K.T., Wong, R.H., 1996. Unaxial compressive strength and point load strength
of rocks. Int. J. Rock Mech. Min. Sci. Geomech. 33, 183–188.
CRCore, 2014. Developing an Integrated response to Current Challenge Retrieved
July 14, 2016, from <http://crca.asn.au/wp-content/uploads/2014/06/03-
CRCORE-Overview-for-CRCA-Conference-May-2014.pdf>.
De Hoff, R., Rhines, F.N., 1968. Quantitative Microscopy. McGraw Hill Inc.
Exner, H.E., Hougardy, H.P., 1986. Einführung in die Quantitative Gefügeanalyse.
Oberursel.
Fandrich, R.G. et al., 1997. Mineral liberation by particle bed breakage. Miner. Eng.
10 (2), 175–187.
Gaudin, A.M., 1939. Principals of Mineral Dressing. McGraw-Hill Book Company Inc,
p. 77.
Glagolev, A., 1934. Quantitative analysis with the microscope by the point method.
Eng. Min. J. 135, 339–340.
Gurland, J., 1966. A study of contact and contiguity of dispersion in opaque samples.
Trans. Met. Soc. AIME 236, 642–646.
Gründer, W., 1957. Arbeitsgebiete und Grundlagen der bergmännischen
Verfahrenstechnik. Chem.-Ing.-Tech. 29 (3), 201–206.
Haberl, H., March 2012. Addicted to resources. Global Change magazine. IGBP.
Retrieved from <http://www.igbp.net/news/features/features/
addictedtoresources.5.705e080613685f74edb800059.html>.
Hesse, M., Lieberwirth, H., 2014. Selektive Zerkleinerung zur Fluorit/Baryt-
Anreicherung bei der Aufbereitung von Bergbauhalden. Chem.-Ing.-Tech.,
797–805
Hesse, M., 2014. Selective comminution for pre-concentration at mineral
processing. In: International Forum-Competition of Young Researchers,
working group ‘‘Mining equipment and mechanical engineering”. National
Mineral Resources University, St. Petersburg, Russia.
Hesse, M., 2015. Selective comminution for beneficiation and pre-concentration of
lead-zinc-ores. Ueo'rokoubz b pawboyakmyoe yelpogokmpodaybe: on yayrb r
gparnbre, <EKUOPOLCRBQ UOCELAPCTBEHHSQ HAWBOHAKMHSQ
BCCKELOBATEKMCRBQ EHBBEPCBTET, Belgorod, Russia.
Hoth, K., Krutský, N., Schilka, W., & Schellenberg, F., 2010. Marmore im Erzgebirge.
Bergbau in Sachsen. Bergbaumonografie, Band 16, Landesamt für Umwelt,
Landwirtschaft und Geologie, Freiberg.
ISRM, 1985. Suggested method for determining point load strength. Int. J. Rock
Mech. Min. Sci. Geomech. 22, 51–60.
Kurzydłowski, K.J., Ralph, B., 1995. The Quantitative Description of the
Microstructure of Materials. CRC Press, Boca Raton, Florida.
126 M. Hesse et al. / Minerals Engineering 103–104 (2017) 112–126
Krausmann, F. et al., 2009. Growth in global materials use, GDP and population
during the 20th century. Ecol. Econ. 68, 2696–2705.
Leißner, T., Mütze, T., Peuker, U.A., 2014. Nutzung der direkten Messung des
Aufschlussgrades in Sortierkennfeldern. Chem. Ing. Tech. 86 (6), 899–905.
Lieberwirth, H., Popov, O., Folgner, T., 2014. Quantitative characterization of rocks
to predict the influence of the rock on relevant product properties and system
parameters. Miner. Process. 55 (7–8), 76–88.
Miller, J.D., 2003. Ultimate recovery in heap leaching operations as established from
mineral exposure analysis bay X-ray microtomography. Int. J. Miner. Process.
72, 331–340.
Müller, H., 2011. Beitrag zur systematischen Untersuchung von Reinigungs- und
Läuterprozessen Dissertation. Technische Universität Bergakademie Freiberg.
Napier-Munn, T., 2014. Is progress in energy-efficient comminution doomed? In:
Proceedings, Comminution 2014, Capetown.
Norgate, T., Haque, N., 2010. Energy and greenhouse gas impacts of mining and
mineral processing operations. J. Clean. Prod. 18, 266–274.
Norgate, T., Jahanshahi, S., 2011. Reducing the greenhouse gas footprint of primary
metal production: where should the focus be? Miner. Eng. 24, 1563–1570.
Palmqvist, S., 1957. Metod att bestämma segheten hos spröda material, särskilt
hardmetaller. Jernkontorets Ann. 141 (7), 300–307.
Peuker, U.A. et al., 2012. Aufbereitungstechnik – Aktuelle verfahrenstechnische
Fragestellungen für die Aufbereitung von mineralischen, nachwachsenden und
sekundären Rohstoffen. ProcessNet.
Popov, O., 2007. Beitrag zur mathematisch-petrographischen Gefügecharakterisierung
für die Beurteilung der Festgesteine hinsichtlich ihrer Aufbereitung und ihrer
Produkteigenschaften Dissertation. Technische Universität Bergakademie Freiberg.
Puffe, E., 1955. Erfahrungen bei der Prallzerkleinerung in Bergbau-, Aufbereitungs-
und Hüttenbetrieben. ERZMETALL, VIII, pp. 570–580.
Puffe, E., 1960. Selektive Prallzerkleinerung als Mittel zur Senkung der
Betriebskosten und zur Erhöhung des Ausbringens in Aufbereitungsanlagen.
Glückauf 96 (1), 20–30.
Reichert, M. et al., 2015. Research of iron ore grinding in a vertical-roller-mill.
Miner. Eng. 73, 109–115.
Saltykov, S.A., 1974. Stereometrische Metallographie. VEB Deutscher Verlag für
Grundstoffindustrie, Leipzig.
Schmidlapp, K., 1957. Erfahrungen mit neuzeitlichen Mühlen in der Kaliindustrie.
Kali und Steinsalz, Kaliverein e.V., Hannover, Band 2, pp. 73–88.
Schmidlapp, K., 1965. Selektive Kalisalzzerkleinerung mittels Prallmühlen,
insbesondere Trennung von Ton und anderen schlammbildenden
Bestandteilen aus Kalisalzen. Aufbereitungstechnik 1, 7–10.
Schodde, R., 2010. The key drivers behind resource growth: an analysis of the
copper industry over the last 100 years. In: MEMS Conference Mineral and
Metal Markets Over the Long Term, SME Annual Meeting, Phoenix.
Schubert, H., 1989. Aufbereitung fester mineralischer Rohstoffe, Bd. 1. VEB
Deutscher Verlag für Grundstoffindustrie, Leipzig.
Schumann, H., 1991. Metallographie. Deutscher Verlag für Grundstoffindustrie,
Leipzig.
Schulz, N.F., 1970. Separation efficiency. Trans. Am. Inst. Min. Metall. Pet. Eng. 247,
81–87.
Stöhr, F., 2012. Untersuchung zur selektiven Zerkleinerung von Blei-Zink-Erzen in
metamorphen Karbonatgesteinen. Studienarbeit. Institut für
Aufbereitungsmaschinen, TU Bergakademie Freiberg.
Stöhr, R., 1976. Möglichkeiten der Trockenaufbereitung von Asbest unter
besonderer Berücksichtigung der Zerkleinerungsverfahren.
Aufbereitungstechnik 2, 61–68.
Sutherland, D., Fandrich, R., 1996. Selective breakage of ores. In: Proceedings,
Mining Technology Conference, Fremantle, Australia, pp. 188–194.
Szutkowska, M., 2005. Modified indentation methods for fracture toughness
determination of alumina ceramics. In: 13th International Scientific
Conference on Achievements in Mechanical and Materials Engineering,
Gliwice-Wisla, Poland.
Taggart, F., 1956. Handbook of Mineral Dressing – Ores and Minerals. John Wiley &
Sons Inc.
Unland, G., Folgner, T., 1997. Automatische Kornformbestimmung durch
photooptische Partikelanalyse. Die Naturstein-Industrie 5, 20–28.
Wills, B.A., 1992. Mineral Processing Technology. Pergamon Press, Oxford.