Yamashita 2021

Minerals Engineering 170 (2021) 107036
Contents lists available at ScienceDirect
Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng
A review of modeling and control strategies for cone crushers in the mineral
processing and quarrying industries
Andre S. Yamashita a, *, Alex Thivierge b, Thiago A.M. Euzébio c
a
Luleå University of Technology, Control Engineering Group, Department of Computer Science, Electrical and Space Engineering, Luleå, Sweden
b
Université Laval, LOOP, Centre E4m, Québec, Canada
c
Instituto Tecnológico Vale, Ouro Preto, Brazil
A R T I C L E I N F O A B S T R A C T
Keywords: Run-of-mine ore is usually too large to be useful for construction or metallurgy. Large particles must be reduced
Review to specific sizes to either comply with aggregate sizing regulations, or facilitate mineral liberation; therefore, in
Cone crusher the aggregates and mineral processing industries, run-of-mine ore is broken in crushing circuits by cone crushers.
Modeling
Here, we review the cone crusher literature, focusing on the modeling and control of crushing circuits. A total of
Control
Optimization
61 works published in the primary literature, ranging from 1972 to 2020, are classified and discussed with
Crushing circuit respect to the model formulation—i.e., population balance models, empirical models, or data-driven mod
Comminution circuit els—and control strategy—i.e., proportional-integral-derivative, model predictive control, expert system, or
model-free control. The data are summarized in a table that makes locating a particular formulation or strategy
quick and easy. The discussion of the current state of the art of crushing circuit modeling and control technol
ogies consolidates the results available in the literature, as well as the challenges that we must overcome to
increase crushing performance through control and optimization. The discussion of future trends brings attention
to the discrete element method, model based control, plant-wide and mine-to-mill optimization, and machine
learning applications for data acquisition and process optimization.
1. Introduction quantitative data supporting the benefits of such strategies. A series of

works (Bouchard et al., 2017; Stange and McInnes, 1995; Duarte et al.,
Rocks, ores, and their subproducts have always been of great 1999; Pomerleau et al., 2000) reported a collection of successful control
importance for civilization. For example, as mentioned in De Re Metal strategy applications in grinding circuits, and Sbárbaro (2010), Jämsä-
lica—one of the oldest known book about mineral processing—back in Jounela (2001) reviewed advanced control strategy applications in the
the Medieval Age, male slaves and peasants, the ’breakage section’, mineral processing industry. Furthermore, surveys in the quarrying in
swung heavy mallets and hammers to break rocks to produce gravel or dustry regarding the optimization of quarrying machinery and lean
extract metals. Women and children, the ’classification section’, sepa production improvement possibilities in the quarrying industry can be
rated the product of interest from the gangue using crafted screens on found in Csöke et al. (1996) and Rylander and Axelsson (2013),
riverbeds. Gravel and aggregates were used to build castles, churches, respectively. However, little attention has been directed towards
and watchtowers, and metal was melted and transformed into silver crushing circuits and its leading equipment, the cone crusher, in either
ware, ornaments, weapons. We have come a long way regarding tech industry.
nology and labor rights; however, the fundamental need to break rocks We aim to provide a starting point for people who wish to investigate
to provide raw material for other industries still remains, and is expected the modeling and control strategies of cone crushers. This review con
to remain in the foreseeable future. tains a compilation of references from the mineral processing and
Several works from the literature have reported applications of quarrying industries, since cone crushers have similar roles in both in
process control strategies in the mineral processing industry. For dustries. In fact, crusher manufacturers used to supply the same equip
example, Wei and Craig (2009) published a survey with qualitative data ment to both industries. However, as customized models began to
from process control solutions, and Bouffard (2015) reported emerge, each industry started to use the specific models of cone crushers
* Corresponding author.
E-mail addresses: andre.yamashita@ltu.se (A.S. Yamashita), alex.thivierge.1@ulaval.ca (A. Thivierge), thiago.euzebio@itv.org (T.A.M. Euzébio).
https://doi.org/10.1016/j.mineng.2021.107036
Received 23 September 2020; Received in revised form 31 May 2021; Accepted 16 June 2021
Available online 27 June 2021
0892-6875/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
A.S. Yamashita et al. Minerals Engineering 170 (2021) 107036
to meet their own demands within their operating constraints (Quist, product shape is rarely important.
2012). Additionally, from the theoretical point of view, crushing per
formance in the quarrying and mining industries have different optima
(Lee and Evertsson, 2011). 2.4. Overview of the mechanical and operational aspects of cone crushers
This work opens with a brief review of the mechanical principles of
cone crushers. Then, in the following sections, works available in the Although the cone crusher definitions presented here are limited and
cone crusher literature, ranging from 1972 to 2020, are categorized with superficial, this section provides the reader with the necessary knowl
respect to the model structure and control strategy implementation. edge to understand the decisions made with respect to process variables
Remarks about the most important contributions of each work are pre and control strategies in the publications from the literature. Some
sented. The paper closes with an overview of the current state of the art discussions, albeit interesting and important for the complete under
cone crushers and considerations regarding their future in the mineral standing of the physics and mechanics behind operating aspects of cone
processing and quarrying industries. crushers, are beyond the scope of this paper. The reader is referred to
Svensson and Steer (1990) for a discussion about the physics involved in
2. Background rock crushing inside a cone crusher; Bearman and Briggs (1998),
Lindqvist and Evertsson (2006) for comprehensive insights into the ef
2.1. Mineral processing industry fect of liner wear and feed properties on the operational characteristics
of cone crushers; and Wills and Finch (2016a), Evertsson (2000) for a
The mineral processing industry consists of a series of processes that detailed description of the mechanical operating aspects of cone
occur in the ore benefaction chain, from the moment raw ore is extracted crushers.
in situ from ore deposits until the product, at specific grades, is delivered Fig. 1 presents a simplified diagram of a cone crusher. Cone crushers
to metallurgy industries. Comminution is an important unit operation of consist of a circular outer concave enclosing a cone-shaped mantle that
the mineral processing chain that involves rock blasting and mechanical moves eccentrically, see Fig. 1. As the distance between the mantle and
comminution (crushing and grinding). The goal is to liberate the min concave inner wall increases, rock particles fall; when the distance de
erals of interest from raw ore so that subsequent concentration processes creases, the particles are compressed and crushed.
can operate optimally (Hodouin et al., 2001). The minimum distance between the mantle and concave is defined as
the closed side setting (CSS) of the cone crusher. The CSS is easily
2.2. Quarrying industry changed online in a large variety of commercial crushers; different
principles of CSS adjustment are described in (Quist, 2017). The
The quarrying industry is fundamental for civilization because its maximum distance between the mantle and concave, on the other hand,
products are used to build roads, buildings, and various types of struc is defined as the open side setting (OSS). The difference between the OSS
tures upon which people depend. In 2017, the European Union alone and CSS is often called the stroke, and it represents the eccentric throw
produced 2.86 billion tonnes of aggregates, and Germany (591 million of the mantle (Asbjörnsson et al., 2012). Usually, cone crushers operate
tonnes) and France (463 million tonnes) were the countries that at a fixed eccentric speed. It is possible to change this speed online by
contributed the most (Uepg.eu, 2017). Although most of the commi means of a frequency drive; however, this is not a typical practice in the
nution from rocks to aggregates and gravel occurs in crushing circuits, industry since such equipment is still expensive (Hulthén, 2010). The
very similar to the ones observed in the mineral processing industry, the area indicated as b in Fig. 1 is defined as the bed thickness, and the area
production goals of aggregate plants are substantially different from indicated as s is equivalent to the difference between the OSS and
these of a crushing circuit in a mining plant (Rylander and Axelsson, CSS—the stroke—and represents the eccentric throw of the mantle
2013). The profit of a quarrying plant is heavily impacted by the product (Asbjörnsson et al., 2012). The compression ratio is defined as bs . The
properties (flakiness, shape, strength, and bulk density) (Ruuskanen, mantle surface in contact with the rocks is protected by a layer of ma
2006). terial called the liner (made of plastic, cement, zinc or epoxy resin (Wills
and Finch, 2016a)). The efficiency of a cone crusher varies with the liner
2.3. The usage of cone crushers wear, and once this wear falls below some acceptable level, worn liners
are replaced with new ones as a part of the maintenance routine.
In both the mineral processing and the quarrying industries, cone Cone crushers have a safety mechanism, which may consist of
crushers perform secondary and tertiary crushing tasks, in which the ore springs or hydraulic cylinders, to protect the equipment from extreme
diameter is reduced from as large as 250 mm to less than 10 mm. The
main differences are that in the quarrying industry, (i) the product of a
crushing circuit is the final product (crushed material may be processed
further in the mineral processing industry), (ii) the final product has a
relatively lower value than the product from the mineral processing
industry, and (iii) the final product must meet stringent requirements,
concerning both particle shape and particle size distribution (such as the
CE marking of aggregates). Thus, the quarrying industry has sought
advanced crushing technology and process control strategies to maxi
mize profit (Svensson and Steer, 1990).
On the other hand, in the mineral processing industry, the product of
a tertiary crushing circuit typically goes through several more commi
nution and concentration steps until the valuable mineral is liberated
from the raw ore. For this reason, state of the art cone crusher control
technology in the mineral processing industry seems to be lagging; the
crushing stage is overlooked because it is only a part of a mining plant,
and variations in crusher performance are rarely a cause for concern,
unless the implications are catastrophic (Bearman and Briggs, 1998).
Additionally, the mineral processing industry is mostly concerned with Fig. 1. Schematic representation of a cone crusher, inspired by Ever
the particle size distribution of the product (Evertsson, 2000); i.e., the tsson (2000).
2
operating conditions, such as when the processed material is too hard or signals that affect the process; we know that they exist, but it
some unexpected material enters the crusher chamber. In these situa takes great effort to model and predict their magnitude and fre
tions, depending on the equipment, either the concave (bowl) lifts up quency. One definition of robustness lies in the ability of a
wards or the mantle drops so that the uncrushable material can pass controller to minimize negative effects from the disturbances on
without damaging the equipment. Both principles lead to an increased the control goals.
CSS. iii. Manipulated Variables. These variables change over time during
The major industrial suppliers of cone crushers include Metso (the operation, and the operator is able to set their values to some
GP, HP and MP series), Sandvik (the CS and CH series developed from extent. They include the CSS, the eccentric speed, and the choke
the Allis Chalmers Hydrocone series), FLSmidth (Raptor® models), level. Note that this section considers only the manipulated var
Pegson, Terrex, and Kawasaki. The Metso and Sandvik models are the iables of the cone crusher itself; however, when controlling a
ones most commonly used in the academic literature. crushing circuit, other manipulated variables can be included.
iv. Controlled Variables. These variables are measured or estimated
2.5. Concept definitions operation variables that may act as a proxy for crushing perfor
mance or safety. They include the power draw and chamber
This work uses some concepts to measure, compare, and identify pressure. The former is defined in terms of the crusher current (if
superior and inferior alternatives with regard to process control goals. only an ampere meter is available) and the power factor, and the
To avoid misconceptions and arbitrary interpretations, their meanings latter is often measured as the hydraulic pressure, although strain
are rigorously defined here. gauges on the cavity are also possible. Moreover, variables such
as the product particle size distribution, product shape, and
• Product quality: In the context of the mineral processing industry, process throughput can be included in this class. This being said,
this concept refers to the particle size distribution of the product depending on circumstances, the latter could also be considered
stream. A sieve greater than 80% of the product stream (P80 ) is a either a disturbance or a mechanical variable.
common metric for particle size (Wills and Finch, 2016b). In the
quarrying industry, the concept of product quality comprises the The following paragraphs detail the variables mentioned above.
product shape, flakiness, particle size distribution (Bengtsson and
Evertsson, 2006b), and strength (Bengtsson, 2009). Standard tests, 2.6.1. Stroke
which are essentially specific sieving procedures, are used to char The stroke, or the eccentric throw, is defined as the distance the
acterize the shape and flakiness, and each test has its own index mantle veers from its axis. It also refers to the difference between the
(Bengtsson, 2009). OSS and the CSS. High stroke values translate into fewer crushing zones
• Efficiency: In both the quarrying and mineral processing industries, inside the chamber (Sandvik, 2020). This parameter is indicated as s in
efficiency is defined as the extent to which the process is able to Fig. 1. The stroke linearly affects the equipment capacity around an
convert energy into products that meet the desired product quality; operating point and has some effect on the number of fine particles.
therefore, when the term ’efficiency’ is used to grade or compare Adjusting the stroke involves a change in the eccentric bushing, which is
control strategies, it refers to the energy efficiency. not a viable online adjustment.
• Performance: In both industries, the performance of the control
platform of a crushing circuit is measured with respect to the extent 2.6.2. Chamber, mantle, and liner design
to which the circuit is able to meet some or all of its goals. Different The liner profile affects the crushing performance and must be
from efficiency, performance can be defined based on the following selected according to the desired product size distribution and the feed
goals of the crushing circuit: maximizing profit, minimizing energy properties (Moshgbar et al., 1995). Similarly, the chamber geometry
consumption, minimizing waste, obtaining an acceptable particle (the base angle of the cone, height of main shaft axis, eccentric angle and
size distribution and shape, maximizing capacity (Ma et al., 2016); rotational speed) optimization, taking into account the product size
maximizing product throughput at a given particle size, and guar distribution and shape, also affects the crushing performance (Gang
anteeing supply of a given rate of crushed material to downstream et al., 2009b).
processes (Wills and Finch, 2016a). Additionally, KPIs devised for
the manufacturing industry can also be used as performance in 2.6.3. Choke level
dicators in the process industry, particularly for crushing circuits The choke level is a function of the liner design, eccentric speed,
(Bhadani et al., 2020). stroke and CSS. It has been reported that cone crushers work best when
• Safe operation: In both industries, safe operation is determined by choke fed (Bearman and Briggs, 1998; Jacobson et al., 2010; Santos
the stabilization of the process while respecting safety margins for et al., 2020) and that operating under these conditions extends the
operations, equipment, and the workforce (Bouchard et al., 2018). durability of both the liners and the equipment.
2.6. Operating variables of cone crushers 2.6.4. Closed Side Setting (CSS)
The CSS is the most used manipulated variable in cone crusher
It is important to consider the inventory of the available process control strategies. Reducing the CSS leads to more crushing and, thus, a
variables in a crushing circuit to define the control objectives and an smaller particle size distribution in the product stream. However, this
overall control strategy. The nomenclature used in this paper is as reduction comes at the cost of a smaller processing capacity. Section 4
follows: discusses several reported works from the literature in which the CSS is
used as a manipulated variable to reach a certain control goal in
i. Operation Parameters. Mechanical design variables are fixed crushing circuits.
operating parameters that do not change over time, at least
without having to stop production. They comprise the stroke, 2.6.5. Eccentric speed
chamber and liner design (including mantle and concave angles). The eccentric speed is a candidate for the manipulated variable in
ii. Disturbances. These variables may change significantly over cone crusher control strategies; however, it is still rarely used due to the
time, and the process has little to no control over them. They absence of frequency converters that allow setting eccentric speed
include the particle size distribution and the composition and values online (Hulthén and Evertsson, 2008). When the eccentric speed
physical properties of the feed. Disturbances are exogenous is increased, the processed rocks are subject to more crushing events,
3
and thus, a smaller product particle size distribution is observed. Slower 3.1. Population balance models
eccentric speed values, however, lead to a product with a better shape
and higher crushing capacity (Quist, 2012). Population balance models (PBM) of cone crushers must consider
how the probabilistic and periodic events of breakage and the creation
2.6.6. Feed conditions of smaller particles affect the particle behavior inside the equipment. In
Cone crushers, which are present in secondary or tertiary crushing the literature, this is addressed by breakage, selection, and classification
circuits, usually have minimal control over the feed rate and properties functions. The following convention is generally adopted:
(hardness, moisture content, particle size distribution, and amount of
mixed ores). These values are often defined by upstream processes. • The breakage function (sometimes called the appearance function)
Therefore, feed conditions are treated as disturbances in the control defines the elements of a lower-triangular matrix that gives the
framework of crushing circuits. In some cases, however, the feed rate is a proportion of each size fraction a particle breaks into when a frag
manipulated variable. It can be changed either directly with a feeder in mentation event occurs;
front of the crusher or indirectly by manipulating the process • The selection function defines the elements of a diagonal matrix
throughput. The circulating load, if present, can complicate manipu containing the breakage rate of particles by size fraction; and
lating the crusher feed rate. • The classification function defines a diagonal matrix in which the
elements are the probability that particles leave a crushing zone—the
2.6.7. Cone crusher operational conditions whole crusher chamber or a fraction of it—depending on the parti
Variables such as the power draw of the crusher, the volume of cle’s size.
material inside the crusher, and the pressure inside the crusher chamber
are important, albeit the pressure is not always measured. Safe operation Simplifications and variations of the PBM are usual. In fact, the
considerations and the potential to implement soft sensors, which esti classification matrix in Eq. 1 is a combination of the discharge rate
mate important but difficult-to-measure variables based on readily matrix and the selection function. It is only valid at the steady state,
available measurements, motivate the use of these variables in control assuming that the discharge rate is proportional to the reactor content
strategies. (Sbárbaro, 2010). Another variation often observed in that the commi
Note that a complete crushing circuit consists of more than the cone nution literature is that the selection matrix is used to describe a prob
crusher alone. Conveyor belts for material transport, silos and bins for ability rather than a breakage rate (e.g., Evertsson, 2000; Anticoi et al.,
intermediate storage and buffering, and screens for classification are 2018). Although both variations make use of matrices, using the
essential for the effective operation of a crushing circuit. Therefore, breakage probability is sometimes referred to as the matrix approach,
appropriate modeling and control of crushing circuits requires modeling while using the breakage rate is occasionally called the kinetic approach
and control of conveyors, silos, screens, and the cone crusher itself. (Machado Leite, 1990). The matrix approach is typically accurate when
the comminution equipment can be considered fixed residence time
3. Cone crusher review by the model structure equipment.
Sometimes, the selection matrix is a function of the power draw,
Arguably, the most well-known cone crusher model was developed which is consistent with the classic comminution theory stipulating that
by Whiten (1972). The model was derived from the population balance decreasing the particle size requires an energy increment to the system
equation formulated by Epstein (1947). Whiten (1972) modeled the (Wills and Finch, 2016b). It is also common practice in the literature to
cone crusher by assuming that it behaves as a single mixer with a clas divide the length of the cone crusher into several crushing zones that
sifier at steady-state, as follows: have their own dynamics; however, their input and output streams are
connected, as observed in Herbst and Oblad (1985), Atta et al. (2014),
xp = (I − C)(I − BC)− 1 xf (1) Evertsson (2000).
Since the dynamics of wear evolve much slower than the fragmen
where B and C are the breakage and classification matrices, respec
tation, different authors extended the steady-state model of Whiten
tively. I is an identity matrix, f p and xf are the feed and product particle
(1972) to account for wear. Moshgbar et al. (1995) modelled the rate of
size distribution, respectively.
wear of the concave and the mantle liners and used this information to
Static models are sufficient in some applications, such as modelling a
transform the classification function parameters into time dependent
cone crusher in a plant wide context in which its dynamic and residence
functions. This new information allowed them to investigate an adaptive
times (approximately a few seconds) are negligible compared to the
control strategy in which the CSS value compensates for wear by varying
other process units. On the other hand, dynamic models allow the effects
in real time. Andersen and Napier-Munn (1990) also developed liner
of disturbances with faster dynamics to be studied, especially in circuits
wear functions but used regression analysis. They modelled the classi
where all units have low residence times (e.g., a circuit comprising
fication function parameters as varying with the CSS, the mass flow rate,
exclusively of crushers, conveyors, and screens).
the feed size, the liner length, and the expired fraction of the expected
Different models have been developed, which can be classified as
liner service life. These approaches are advantageous for studying the
either population balance, empirical, or data-driven models. It should be
wear phenomenon since the computational requirements are low. The
noted that despite being reported in the cone crusher literature, discrete
results show that the longer the liner is, the greater is the capacity of the
element method (DEM) modeling is not considered in this section. DEM
cone crusher. On the other hand, the more worn the liner is, the smaller
is a numerical technique that tracks the progress of each particle in the
is the capacity of the cone crusher. Therefore, although it is financially
simulated system over time by integrating the equations of motion of
attractive to wait as long as possible to change worn liners because of the
each entity. DEM models have been used to gain a better understanding
cost of a new liner, at some point, the production loss caused by wear
of the flow and breakage processes inside the cone crusher, and for
outweighs the cost of a new liner. To study both fast dynamics (i.e.
power draw and liner wear predictions (e.g., Cleary et al., 2017; Delaney
fragmentation) and wear, one could use the model of Sbárbaro (2010),
et al., 2015; Quist and Evertsson, 2010; Quist, 2012). These models are
which is a dynamic equivalent of Eq. (1), and combine it with the wear
computationally expensive and, to the best of our knowledge, have not
functions of either Andersen and Napier-Munn (1990) or Moshgbar et al.
been applied to the process control of cone crushers. The interested
(1995). Sbárbaro (2010) presented dynamic simulations of a whole
reader is referred to Weerasekara et al. (2013) for an extensive review of
comminution circuit, including a crushing circuit, screens, conveyor
the state-of-the-art of DEM models in the mineral processing industry.
belts and grinding mills, and studied different process control schemes.
These results could possibly be extended by including wear functions.
4
Although the model of Whiten (1972) showed adequate perfor crusher model in a mixed logical dynamic simulation of a crushing cir
mance, the premise of the crusher acting as a single mixer and classifier cuit. The resulting model is a hybrid, piecewise affine model that can
remains questionable. As such, Herbst and Oblad (1985) proposed a represent both nonlinearities and discrete events, such as trucks loading
dynamic PBM for cone crushers considering the chamber volume as a run-to-mine ore in a primary crushing station stockpile.
series of zones. A Kalman filter was used to estimate the particle size in Lindqvist and Evertsson (2006, 2003), Lindqvist and Sotkovski
each intermediate zone, power draw, and product feedrate. Arguably, (2003), Lindqvist and Evertsson (2004), Evertsson and Lindqvist (2002),
this is a more realistic representation of the dynamics in a cone crusher, also based on the work by Evertsson (2000), modelled linear wear with a
but it is difficult to validate because it would require measurements of mechanistic model. The initial model of the effects on geometry and
the particle size distribution of the material inside the crushing chamber performance by liner wear was developed in Lindqvist and Evertsson
at different heights to do so. (2003), Evertsson and Lindqvist (2002). Some measurement and pre
Other interesting steady-state representations have been proposed diction discrepancies were reported in Lindqvist and Evertsson (2003),
apart from Eq. (1). For instance, Machado Leite (1990) developed a PBM and to address these issues, an improved flow and pressure model was
comprising a breakage rate function that assigns particles to be broken presented in Lindqvist and Evertsson (2004). Significant improvements
to three energy levels depending on their size with the greater particles in the prediction of the CSS, power draw and capacity were observed,
being broken at a faster rate since they are more likely to contain flaws but only minor improvements in the prediction of wear were observed.
(cracks). Their representation considers the effect of the CSS in the In Lindqvist and Evertsson (2006), the effect of shear forces along the
following two ways: (1) they calibrate the residence time as a function of crushing surfaces was incorporated into the model, which resulted in
the CSS, and (2) the selection functions depend on the CSS. Their significant mitigation of the previously observed discrepancies. The
approach contrasts with the one of Whiten (1972) in which the CSS authors emphasized that despite the positive results, the model is still
influences the residence time indirectly through a classification matrix. incomplete, as incorrect or incomplete assumptions regarding the
Moreover, their paper presents an interesting discussion regarding an operating variables and their relationships might still affect the
improved selection function with physical meaning and constraints to predictions.
facilitate calibration by numerical methods. In parallel to the works by Lindqvist on liner wear, Ma et al. (2016)
An interparticle breakage model was developed in Lee and Evertsson emphasized the importance of an accurate cone crusher model for pro
(2008) and validated in full-scale tests in Lee and Evertsson (2011). In cess optimization and proposed a heuristic model that correlates the
the latter, the authors emphasize that significant gains in both product material hardness, compression ratio, and particle size distribution with
yield and process capacity can be achieved with adequate process the liner pressure during crushing. Based on this correlation, the effect of
control. liner wear on product quality is estimated and may become an addi
A mechanistic modular model for particle flow, size reduction, and tional tool for the dynamic optimization of the process. The author re
pressure prediction in cone crushers was developed in Evertsson (2000, ported that the liner wear effect leads to significant improvement in the
1995), Evertsson and Bearman (1997), Evertsson (1998, 1999a,b). The prediction of the particle size and flakiness.
authors defined the size reduction by consecutive events, each with its
own breakage and selection function. The initial exploration of the basic 3.2. Empirical models
operation principles of cone crushers was reported in Evertsson (1995),
Evertsson and Bearman (1997), the breakage characteristics of different This section reports on works in which the breakage mechanism is
materials were investigated and included in the model; the concept of either not modelled, is not modelled with the typical breakage-selection-
the crusher performance map was introduced in Evertsson (1998), classification approach, or where the breakage is an empirical function
Evertsson (1999a), equations of motion that consider the crusher ge of other variables. We also discuss empirical models derived from the
ometry were included to describe the material flow inside the crushing analysis of operating data supported by machine learning tools.
chamber; and Evertsson (1999b) drew additional conclusions on the Bengtsson and Evertsson (2006a) developed an empirical model that
interaction between material flow and size reduction. The Ph.D. thesis of correlates the average particle size of the feed and the CSS with the
Evertsson (2000) summarized the findings of the previously mentioned product flakiness index, an important variable in the quarrying industry.
references. The resulting model was validated in full-scale tests. This work showed some discrepancies with respect to the accuracy of the
Asbjörnsson et al. (2016) used the mechanistic model described in model, which were attributed to its simplicity. In the same vein, a linear
Evertsson (2000) and the hopper level influence on crushing capacity empirical model calibrated by the least squares method (Ruuskanen,
theory from Sbárbaro (2005) to perform crushing circuit optimization 2006) was also unsuccessful.
from an operational perspective while considering the effects of wear Eloranta (1995) reported that the crusher stroke and eccentric speed
and plant stops on the production rate. Itävuo (2009), Itävuo et al. do not have a large influence on the product flakiness, based on a study
(2013) also used the model from Evertsson (2000) to develop a dynamic with a small sample size. Other authors reached different conclusions.
cone crusher model, which is discussed in more detail in the next Bengtsson and Evertsson (2006a) concluded that including mechanical
subsection. knowledge such as the chamber geometry, stroke and eccentric speed
Atta et al. (2014), based on the work by Evertsson (2000), Whiten may improve the flakiness model. Ruuskanen (2006) suggested a
(1972), Herbst and Oblad (1985), proposed a cone crusher model that structure for a sophisticated model. The particle breakage should be
takes into account several operating characteristics to calculate crusher based on the structure of the particle, and the reliability of product shape
performance maps and the dynamic response of the crusher. The and liner wear calculations can be increased by considering the mineral
considered characteristics were classification, selection and breakage, contents of the raw ore. In Bengtsson et al. (2009), the product shape
volume capacity limitation, the influence of the eccentric speed in the model from Bengtsson and Evertsson (2006a) was used to optimize the
vertical transport of material, and the influence of mixed materials in the flowsheet design and operation parameters of a crushing circuit.
input. Results were obtained for different feed characteristics and for Bengtsson (2009) reported that the model in Bengtsson and Evertsson
step inputs in the eccentric speed and CSS. The model was extended in (2006a) was extended to accommodate the CSS, feed particle size, throw
Atta et al. (2019) to account for power consumption during the crushing and eccentric speed. The model was validated in breakage tests using
stage using a modified implementation of Kick’s formula. In Atta et al. different types of rock. A new model extension considering the inter
(2013), a simplified version of the model was used for the process particle shape for vertical shaft impact crushers was developed in
control of cone crushers. A model based on the continuous function Bengtsson and Evertsson (2006b).
method was developed in Johansson (2009) to predict the product size Gang et al. (2009b,a) proposed a heuristic model of cone crusher
distribution. Karelovic et al. (2015) used Atta et al. (2014)’s cone throughput and product flakiness based on the crushing chamber
5
geometry, rotational speed, material harness, and bulk density of the equations to model the behavior of the Hydrocone and Symons cone
final product. The authors sought a simpler model representation than crushers. The parameters were validated using laboratory-scale experi
that of Evertsson (2000). The model was validated using full-scale tests ments, and the authors investigated the optimized operation of crushing
in Gang et al. (2009b). The product particle size distribution and particle circuits from both the process control and flowsheet design perspectives.
shape models were combined to obtain a flakiness prediction model. The Most references in the mineral processing literature consider that the
optimization of cone crushers was studied while considering the output processed material is ultimately a commodity, and thus, they are not
flow rate as an objective function and the flakiness and particle size concerned with the mineral concentration. However, in the case of rare
models as constraints. Their model was used to optimize the chamber minerals processing, in which the intrinsic value of process material is
dimensions in Gang et al. (2009a), and a prototype was built to validate high, it is important to have such information. The heuristic model
the results. developed by Leon et al. (2018) describes the particle size and mineral
Some modeling attempts from the literature resulted in crushing concentration as a function of the compression ratio and can provide
circuit simulation packages. Itävuo (2009) developed dynamic models valuable additional information for process optimization.
with first-order or second-order dynamics plus dead time dynamics Itävuo et al. (2017, 2014) modelled a crusher feeder-hopper from a
based on the static regression models in Ruuskanen (2006) and the combination of mass balance and empirical equations. They used their
particle flow description in Evertsson (2000). The result was a dynamic model to develop an unbiased state observer that guarantees asymptotic
simulation package of crushing circuits. Later, Itävuo et al. (2013) convergence. Airikka (2015) used a similar modelling technique that
extended this work to account for the dynamic behavior of cone crushers resulted in the implementation of a predictive proportional integrative
in the presence of input disturbances. A dynamic model combining the (PI) controller that manipulates the feeder speed for throughput control.
blasting and crushing stages in the quarrying industry was developed in None of the works mentioned in this paragraph developed a crushing
(Csöke et al., 1996). This model considers empirical blasting and the model or modelled the breakage behavior but rather provided methods
breakage function of cone crushers, along with a probability model of to better understand the mass balance in the crushing circuit.
vibrating screens, and is delivered as a computer program for optimi
zation and planning. A toolbox for the dynamic simulation of crushing 3.2.1. Machine learning based models
circuits was developed in Asbjörnsson et al. (2012), Asbjörnsson et al. The new automation paradigm called Industry 4.0 benefits from the
(2013), Asbjörnsson (2013), Hulthén et al. (2012b), Hulthén et al. large amounts of data generated in process operation due to the
(2012a), Asbjörnsson (2015). Asbjörnsson et al. (2012) revealed a dy increasing availability of all types of sensors (Wakefield et al., 2018).
namic mass balance model incorporating a wear function that was Nevertheless, the literature related to this topic and applied to cone
developed for a commercial dynamic simulation software called crushers remains scarce.
SysCAD. The Swebrec function (Ouchterlony, 2005), which was origi McCoy and Auret (2019) reported some applications of machine
nally developed to describe the particle size distribution of blasted rock, learning tools in the mineral processing industry and theorized about the
was used to describe the particle size distribution in the crusher. The current needs and future applications in this area. Most works based on
results from a previous work (Svedensten and Evertsson, 2005) verified data-driven models and methods in the mineral processing industry are
the feasibility and accuracy of the prediction. Asbjörnsson et al. (2013) related to fault detection (Wakefield et al., 2018; Suriadi et al., 2018)
performed simulations using Simulink to increase the modeling flexi and economic optimization (Lu et al., 2019; Jiang et al., 2018). How
bility. The equipment models were differential equations in which the ever, no studies have explored applications in crushing circuits. On the
parameters were obtained from fitting real plant data. The Licentiate of other hand, Hulthén (2010), Hulthén and Evertsson (2008, 2009, 2011,
Engineering Thesis and the Ph.D. thesis of Asbjörnsson (Asbjörnsson 2010) utilized process data to perform real-time optimization of cone
(2013) and Asbjörnsson (2015) respectively) summarize the previous crushing by manipulating both the CSS and the eccentric speed.
results. Applications of the toolbox for plant performance evaluation
(Asbjörnsson et al., 2013) and optimization (Hulthén et al., 2012b) and 4. Cone crusher review by the control strategy type
operator training (Hulthén et al., 2012a) were reported. Based on the
crusher model developed by Asbjörnsson et al. (2012), Li et al. (2018) Cone crushers, along with screens and other classification equip
proposed a dynamic model of a SAG mill-pebble crusher circuit. ment, bins and silos, and conveyor belts, are the essential building
Methods based on test works to estimate power draw are also worth blocks of a crushing circuit. Before discussing the references that report
mentioning due to their frequent applications in commercial simulation on the applications of control strategies in cone crushers, a brief over
platforms. For instance, the JKSimMet software computes the power view of the process control of crushing circuits is presented for the
draw of cone crushers from the estimated particle size distribution and reader’s convenience.
laboratory tests based on the model of Andersen and Napier-Munn
(1988) (JKTech, 2015). In a similar manner, USIM PAC first estimates 4.1. Process control of crushing circuits
the particle size distribution and then uses this information to estimate
the power draw of cone crushers from the Bond Work Index (Caspeo, Process control strategies can be simple (proportional-integrative-
2011). Other test work methodologies detailed in Morrell (2019) could derivative (PID) control), complex (plant wide economic model-based
likely also be used similarly. control), or any combination in between. Such strategies are among
Bearman et al. (1991) performed pilot-plant scale experiments the most reliable ways to improve the operating performance of a pro
involving a laboratory scale cone crusher to understand and predict the cess without changing its flowsheet (Garcia et al., 1989; Lotter et al.,
following two operating characteristics: power consumption during the 2018; Svensson and Steer, 1990). Control strategies seek to optimize
crushing stage; and product particle size distribution as a function of some type of performance index or goal (production rate, product
CSS, feed size and rock strength. The authors drew some conclusions quality) by considering a set of operational, financial or environmental
regarding the relationship among the operating variables and proposed constraints (Bhadani et al., 2020). Moreover, process modeling simula
heuristic prediction models for cone crusher performance and grading tion and control also provide a framework for fault detection, operator
curves. training, and plant evaluation.
Alvarado et al. (1998) performed economic optimization of a Although many works have contemplated control strategies for
comminution plant (including crushing and grinding circuits) mini grinding circuits, the literature concerning crushing circuits is limited.
mizing the energy cost of equipment, using the feed flowrate to the Since grinding and crushing operate with process-specific equipment,
grinding circuit as the single decision variable. The equipment models the results observed in the former are not straightforwardly extendable
were based on empirical formulas. Grujić (1996) used empirical to the latter. Hulthén (2010) argued that the preference for grinding
6
circuits over crushing circuits in the literature is due to the high intrinsic • Control the throughput rate to ensure that it meets production
economic value of the products. Liu and Spencer (2004) mentioned that requirements.
dynamic models of grinding circuit equipment are relatively mature and • Minimize the energy consumption per mass of product.
that researching the appropriate control of problematic units, such as • Minimize the effect of disturbances on the process performance
SAG/AG mills, is of interest. On the other hand, the final product of the (Bearman and Briggs, 1998), such as the following: feed rate and
quarrying industry is usually derived during the crushing stage, and physical properties, i.e., hardness, moisture content, size distribu
thus, process control of cone crushers is treated with high regard in the tion, and content of different ores (Itävuo et al., 2011).
quarrying literature. Notably, since some of the performance and • From a theoretical perspective, demonstrate the robustness and
product quality goals differ between the mineral processing and quar stability in the presence of a plant-model mismatch, measurement
rying industries, the control strategies adopted by these industries are noise, and disturbances.
not directly interchangeable.
The efficient process control of crushing circuits consists of the A control algorithm should be developed while considering the two
following two interacting tasks: following goals: to meet the previously defined control objectives and to
deliver safe, straightforward and reliable plant operation.
• Mass balance control: Adjustment of the level of storage equipment
and the delivery dynamics of the raw material (through conveyor
4.2. Control strategies
belts and tripper cars) to ensure that sufficient feed material is al
ways available for choke-feeding the crusher(s).
4.2.1. PID control
• Product quality control: Adjustment of the crusher operating vari
PID controllers are widespread throughout all industries because of
ables to ensure that the product particle size distribution provides a
their relatively easy and straightforward implementation procedure
suitable final product or feedrate to downstream processes or that the
(Visioli, 2012). Fundamentally, PIDs are designed for single input single
product quality in terms of shape is adequate.
output (SISO) systems; the manipulated variable values over time are
calculated to minimize or eliminate the error between the controlled
Neglecting to implement a comprehensive control strategy in
variable and its setpoint. PID is an active research field in regard to
crushing plants may lead to low plant performance or unexpected
tuning, control strategy design, and extensions that enable applications
fluctuations due to plant instability (Asbjörnsson, 2013). Moreover,
in originally intractable problems. The fundamental premise of PID
when economic optimization is of interest, simply choosing to maximize
control is the feedback principle. In Fig. 2, assume that the sensor block
one product fraction may lead to the undesirable production of fines
is negligible; therefore ym = y. The system shown in Fig. 2 is a closed-
(Svedensten and Evertsson, 2005).
loop feedback system, indicating that under some assumptions, the
Fig. 2 presents a generic representation of a process in a closed-loop
output will be driven to a desired setpoint despite the process charac
with a certain control strategy. Different algorithms may be assigned to
teristics and disturbances, which means ysp − y = e = 0. In the classic
the ’controller’ block, such as PID, model predictive control (MPC), and
PID theory, a control action u that accomplishes output tracking in
expert systems (ES) (Olivier and Craig, 2017). The basic principle un
derlying feedback control is the use of plant measurements to reduce the closed-loop feedback systems is as follows:
( )
tracking error between outputs and their setpoints. The measurements ∫
1 t de(t)
are either direct measurements of the process output, if available, or u(t) = K e(t) + e(τ)dτ + Td , (2)
Ti 0 dt
some type of estimation via observers or soft sensors. The optimal
control actions calculated in the ’controller’ block minimize a cost where t denotes time, K is the proportional gain, Ti is the integral time
function that takes into account the previously mentioned tracking error and Td is the derivative time. From Eq. (2), the control gain is the sum of
or some other operation goal. three components that are proportional to the following: the error, the
integral of the error, and the derivative of the error. For a detailed
4.1.1. Control objectives analysis of PID controllers, readers should refer to Åström and Hägglund
Most reports of industrial and academic applications of control (1995).
strategies in crushing circuits suggest that the following are the most In Itävuo et al. (2011), using simulations, the authors demonstrated
important control objectives: that simple PID control of a gyratory cone crusher can cancel all
reasonable effects of process disturbances, although full-scale tests are
• Control the particle size and shape of the product to ensure that the needed to validate the findings. The CSS of the cone crusher was used to
product meets quality requirements. control the product size and reject disturbances in the feed size and
moisture.
A tertiary crushing stage was controlled in Itävuo et al. (2017) with a
combination of a decentralized PID and a dynamic state estimator for
the hopper level by manipulating the eccentric speed. Faulty sensors
might result in corrupted level measures, and thus, a proper state esti
mator was needed for the implementation of the feedback-based control
strategy. The crusher power, circulating load, and throughput mea
surements were used in the calculation of the optimal setpoints for the
hopper level height, eccentric speed, and CSS. However, the CSS is set
manually by an operator. Operating cone crushers at the choke-feed
point is recommended by manufacturers because this approach gives a
more consistent product sizing, maximizes throughput, develops a uni
form wear profile, and extends the liner life (Bearman and Briggs, 1998;
Ma et al., 2016).
Fig. 2. Generic closed-loop representation of a process. Here, ysp , ym , and y In Bearman and Briggs (1998), a classic selector strategy using 2 PIs
denote the output setpoints, measured outputs and process outputs, respec for the crusher filling level and crusher power was implemented. The
tively; e denotes the difference between the output setpoint and measured results of a simulation considering an empirical plant model demon
output; and u denotes the control actions sent to the process. strate that the additional degree of freedom allows for a reduction in
7
dead time effects and thus provides tighter control of relevant variables. are calculated, a cost function such as Eq. 4 is designed. The control cost
The work by Asbjörnsson (2013), who consolidated the punctual defined in (4) includes the output predictions and the input increments
results presented in Asbjörnsson et al. (2012, 2013), Hulthén et al. Δu(k|k) = u(k|k) − u(k − 1|k). The notation y(k|k) implies that the value
(2012b,a), utilized PID controllers in all instances in which it was of y(k) depends on the measurement up to time step k. Variables Qy and
necessary to operate the plant in a closed loop to obtain the pursued R are weighting matrices.
results. The simulations were performed in Simulink, using its PID
x(k + 1) = Ax(k) + Bu(k)
control toolbox, tuned accordingly. However, the work focused more on (3)
y(k) = Cx(k)
the simulation and model validation and representability than the pro
cess control issue, thus providing minimal insights regarding this issue. p
∑ ∑
m− 1
⃦ ⃦
Sbárbaro (2005) proposed using the variable speed drives of V(k) = ⃦ysp (k + i|k) − y(k + i|k)⃦2 + 2
‖Δu(k + i|k)‖R (4)
Q
conveyor belts as a new degree of freedom to compensate for the dead
y
i=1 i=0
time related to material transportation inside the crushing circuit. The

control goal is to ensure that the crusher chamber remains filled and An optimal control action profile, up to the control horizon denoted as
maintain choke-feed conditions while keeping the crusher power under m, is calculated by the MPC results from minimizing (4). Finally, only
safe operating margins. the first element of the optimal vector is fed to the plant, and the pro
The predictive PI controller described in Airikka (2015) manipulates cedure is repeated at time instant k + 1. For more information regarding
the crusher feeder speed to regulate the feed rate at a given setpoint. The MPC, readers should refer to Maciejowski (2002).
author did not address the control of the cone crusher itself but discussed An adaptive control strategy based on optimal control that manip
controlling the feed rate to the primary crusher by manipulating the ulates the CSS to compensate for liner wear is proposed in Moshgbar
speed of a vibrating feeder in a feedforward capacity. Differing from et al. (1995). The algorithm compares the particle size distribution of the
feedback control, feedforward control only uses an estimate of the dis plant output with the expected predictions calculated by a plant model.
turbances to calculate the optimal control actions. Since the output The strategy is implemented as two cascading loops, and in addition to
values are not measured to compensate for the control action, a feed compensating for liner wear, it accounts for disturbances in the
forward controller is not a closed loop system. The control goal was to composition of the plant feed. Although it is not clear which control
maintain the mass flow in the secondary crushing stage as constant as algorithm was used, the general strategy resembles the idea behind an
possible. The results showed that the proposed feedback plus feedfor MPC, which is why we described it in this section.
ward approach yields lower throughput variability than a control Sbárbaro (2010) summarized some MPC applications in the mineral
alternative without the feedforward module. processing industry and the most popular commercial solutions. Jämsä-
Manlapig and Watsford (1983) implemented a variant of a PI Jounela (2001) reported some successful applications of MPC in the
controller in a crushing plant. A first-order plus dead time model was mineral processing industry. However, application exclusively in
used to dynamically compensate for the proportional term. The crushing circuits is extremely limited.
controller algorithm was implemented in a commercial control package. Karelovic et al. (2015) implemented an MPC based on a piecewise
The authors showed that compared to a standard PID, the proposed affine model with support to both logical rules and continuous dy
approach could significantly reduce the variability of the crusher current namics. Piecewise affine models represent nonlinear dynamics by a set
during the operation. of linear models identified at appropriate operating points. Under some
Building upon the principle of mass conservation and the mass bal assumptions, piecewise affine models are equivalent to mixed logical
ance model studied in Itävuo et al. (2014, 2017), Itävuo et al. (2019) dynamical systems. Such a model was able to capture both the non
developed a control strategy for crushing circuits that considers the linearities and discrete, logical variables that arise from crushing circuit
overall mass balance control and the level control of intermediate modeling. An MPC strategy for primary crushing, secondary screening
buffers (stockpiles, bins, silos and feed-hoppers) to ensure 100% utili and tertiary crushing sections was developed. The resulting MPC prob
zation at the circuit bottleneck. The authors modeled the whole crushing lem was written as a mixed-integer quadratic programming problem. In
circuit as a single unit, thereby eliminating the dead time issues caused this strategy, each section has manipulated and controlled variables and
by material transportation, and implemented a PI strategy to regulate control objectives to maximize the output mineral flow while preventing
the feed flow rate with respect to material accumulation and the realized the overflowing or emptying of a storage bin for the primary crushing
circuit throughput. A limiting control approach was selected to manip section and to minimize tripper car movement while preventing over
ulate the in-circuit flows, thus ensuring the maximum utilization of the filling of the storage bins for the primary screening and tertiary crushing
surge capacity and circuit flow consistency. As in Airikka (2015), the sections. The mass balance control of the overall circuit was enforced.
crushing circuit throughput was controlled by utilizing the feed rate, and The practical implementation and maintenance problems of MPC
no control strategy was directly applied to the cone crusher variables, may explain its scarce applications reported in the crushing circuit
such as the CSS or eccentric speed. Although PI controllers were utilized literature. MPC requires an accurate dynamic model of the plant to be
in the lower level of the control hierarchy, the authors referred to the controlled, which is difficult to obtain (Fernandez-Camacho and
proposed control strategy as mass balance control. Bordons-Alba, 1995; Zhu et al., 2013) in many cases. Crusher liner wear,
variations in the mine ore characteristics, and physical modifications
4.2.2. Model predictive control (MPC) during feeder maintenance can lead to plant-model mismatch and
Differing from the decentralized control strategies presented in the deteriorate the performance of the controller over time. It is common
previous subsection, MPC takes advantage of a full process model to that after a promising initial operation after commissioning, the MPC
reach some control objective. MPC directly addresses input and output system leads to shutdowns due to low performance (Tufa and Ka, 2016).
coupling in its formulation. Eq. 3 is a generic linear representation of Additionally, the theory behind MPC is more complex than that of
such a model, in which the plant states x(k) at time step k are calculated distributed control and expert systems, which decreases the confidence
as a function of the plant states at the previous time step and the plant of operators in the control room. The lack of knowledge of some oper
inputs. Matrices A, B, and C are state-space matrices that represent the ators is one of the key reasons why process control solutions fail to
process. The basic principle of the MPC theory is the receding horizon deliver their full potential (Li et al., 2011; Forbes et al., 2015). As dis
strategy. The calculation of the predicted process outputs is based on a cussed in the following section, these practical disadvantages of MPC are
model of the process (such as Eq. (3)) and current and past information not a concern for the expert system; a model of the process is not
regarding the operating conditions at time instant k over a prediction necessary, and the theory behind it is easily understood by those who do
horizon p. Once the output predictions, which are denoted as y(k + i|k), not work directly with process automation and control.
8
4.2.3. Expert systems different cone crushers fit better in secondary or tertiary crushing cir
Expert systems (ESs) are algorithms that combine logic and expert cuits, depending on their characteristics. In their case study, a Symons
knowledge to make decisions. ESs range from simple decision trees to crusher performed better than an Hydrocone crusher in the tertiary
more complex applications that integrate fuzzy logic, neural networks crushing position. As an additional result, the author reported an
and numerical models. ESs have been proven to be successful in the application of ES in the Majdanpek copper mine, Serbia. It was shown
mineral processing industry because they allow for the translation of the that the control strategy maximizes the crushing energy utilization,
operator’s knowledge into computational rules. Fig. 3 presents a dia prolongs the lifetime of the crusher parts, and increases the crushing
gram of an ES. The process knowledge from process engineers and op efficiency.
erators is translated into mathematical rules, which are applied to plant Hulthén and Evertsson (2009) described the implementation of a
data (for example, output measurements acquired by soft sensors) to naive pathfinder finite state machine using online measurements of the
calculate optimal control actions. The complexity of the strategy comes CSS to control the crushing performance. Hulthén and Evertsson (2008,
mainly from how the control rules are created and translated into con 2010) reported the usage of eccentric speed to control the production
trol actions. Most ESs are implemented to support decentralized control yield of a given product, and Hulthén and Evertsson (2010) reported the
strategies in an advisory capacity, for example, to provide optimal set results of a model that considers both the eccentric speed and CSS to
points for PID controllers. (While et al., 2004). The reviews, surveys, and predict the product yield. The Ph.D. thesis by Hulthén (2010) summa
references in Sbárbaro (2010) provide a comprehensive overview of the rizes the findings of previous studies. By extending previous work,
application of ESs in the mineral processing industry; however, most Hulthén and Evertsson (2011) reported the application of the two-
reported applications are applied for grinding circuit control. variable model defined in Hulthén and Evertsson (2010) for online
Among possible ES algorithms are finite state machines (FSM) and control in an evolutionary operation algorithm.
evolutionary operation (EVOP). The former is based on process
modeling limited to a finite number of states, and the latter is based on 4.2.4. Model-free control strategies
process optimization, in which past measurements are used for future In Atta et al. (2013), the authors implemented extremum seeking
improvements. The results of FSM (Hulthén and Evertsson, 2009) and control (ESC) in cone crushers based on a simplified version of the model
EVOP (Hulthén and Evertsson, 2011) application in crushing circuits in Atta et al. (2014). ESC relies on the existence of a global maximum (or
have been reported. FSM control yielded better performance than con minimum) in the steady-state relationship between the inputs and out
stant CSS on an average of 3.5% of the time; however, due to the process puts of the system. ESC was originally intended to be used in combi
settings, the cone crusher operation was mostly limited by hydraulic nation with bandpass filters to minimize the effects of signal noise; the
pressure. In the EVOP evaluation, the algorithm dynamically calculated authors proposed a modified extended Kalman filter application that
the optimal eccentric speed values while the CSS was changed periodi performed better than the original approach in simulation scenarios.
cally. The results show that compared to the manual control and fixed
value settings, the EVOP control of the eccentric speed yielded a 4.2% 4.3. Offline optimization
increase in the crushing stage throughput. The real-time optimization of
a crushing circuit using an evolutionary operation algorithm was re Some references report the application of heuristic or data-driven
ported by Hulthén and Evertsson (2011). By measuring the production models to optimize the crusher circuit and cone crushers with respect
rates of different products in a quarrying plant, the optimal setpoints for to the crusher machine parameters and feed properties. Such applica
the CSS (indirectly via power draw measurements) and eccentric speed tions differ from those previously mentioned because they are not
were calculated at each EVOP run using a heuristic model (Hulthén and designed for the online, closed-loop control of the process. Nonetheless,
Evertsson, 2009). The authors reported an improvement in product offline optimization is useful for providing optimal operating point
throughput of approximately 15% compared to that in the case of the candidates (Atta et al., 2014) and information regarding the overall
fixed set-points. operating scheme of the cone crusher circuit.
Albuquerque et al. (2019) implemented a rule-based averaging level Crusher performance maps are surface plots that correlate to the
control algorithm for the level of solid materials in storage bins, operating variables that affect the crushing performance, such as the
manipulating the tripper car speed and the number of active silos in a following: the CSS, eccentric speed, rock characteristics, and feed size
parallel configuration. According to the authors, the proposed solution is Evertsson (2000). Although these plots are not rigorous models in the
simple yet effective. sense of candidates for process control implementation, crusher per
Grujić (1996) focused on flowsheet optimization and concluded that formance maps can provide useful guidelines for flowsheet design and
can enhance our understanding of the mechanical characteristics of cone
crushers. To better understand and predict power consumption during
the crushing stage and the product particle size distribution as a function
of the CSS, feed size and rock strength, Bearman et al. (1991) performed
pilot-plant scale experiments using a laboratory scale cone crusher. The
authors drew some conclusions regarding the relationship between the
operating variables and proposed heuristic prediction models the cone
crusher performance and grading curves. Evertsson (1998) combined a
process model that describes the flow and breakage of rocks in a cone
crusher with raw rock properties to elaborate a crushing performance
map. This map can potentially help calculate optimized operating
crusher conditions for different feed characteristics, market prices and
other economic parameters. Evertsson observed that the eccentric speed
greatly affects the crusher capacity and throughput particle size distri
bution. Atta et al. (2014) used crusher performance maps to demon
strate that their proposed dynamic cone crusher model is consistent with
operational data.
Bengtsson et al. (2017, 2009) proposed the multiobjective optimi
zation of a comminution circuit comprising a cone crusher, a vertical
Fig. 3. Expert system (ES) diagram. shaft impact crusher, screens, and silos based on the mechanistic cone
9
crusher model presented in Evertsson (2000) and the shape model consumption.
presented in Bengtsson and Evertsson (2006a). This work represented a Asbjörnsson et al. (2016) performed a crushing circuit optimization
more fundamental approach to optimization than other approaches; on study with a focus on the operational perspective. Dynamic models of
the one hand, this approach contemplated a broader perspective than the cone crusher (Evertsson, 2000; Sbárbaro, 2005), screens, conveyors
that in the works reporting on the control algorithm applications for and silos, and equipment wear dynamic models were combined with a
cone crushers; on the other hand, this approach offers a general strategy. discrete event simulation model that incorporates events such as
The approach provides interesting insights into the effects of liner wear maintenance, stops, and delays. When necessary, control loops were
and mean time to failure on the capacity, quality and financial profit of closed using PI controllers; however, limited attention has been paid to
the crushing plant. Bhadani et al. (2018) used the crusher model re the implementation and tuning of control strategies. The authors aimed
ported in Asbjörnsson et al. (2012) to propose a multidisciplinary opti to demonstrate how a combination of dynamic models and discrete
mization platform for crushing plants. An example illustrates the events can provide insight into the crushing process.
optimization of the operating parameters of the secondary and tertiary Some works in the cone crusher control literature offer a broader
crushing sections. Similarly, Li et al. (2018) simulated a SAG mill-pebble perspective regarding process optimization by contemplating the cone
crusher circuit using the crushing model in Asbjörnsson et al. (2012) to crusher and all the ancillary equipment that comprises a crushing cir
investigate the effects of the CSS and eccentric speed manipulation in the cuit. A drawback of such works is that the cone crusher model is not
pebble crusher on the overall circuit to optimize the total throughput. detailed. The work by Svedensten and Evertsson (2005) did not provide
Svedensten and Evertsson (2005) reported the economic optimiza detailed information regarding the characteristics of the cone crusher
tion of a crushing circuit based on the maximization of gross profit; the model utilized. An analysis of the results indicates that the material flow
resulting problem was solved using a genetic algorithm in a simulation rate and particle size distribution are functions of the CSS and other
environment with steady-state equipment models. The authors reported parameters. The authors mentioned that a more detailed model might
successful results concerning the performance and economic aspects of benefit the proposed optimization technique. Similarly, While et al.
each production unit (cone crusher, sieve) in the circuit. Nonetheless, (2004) did not provide detailed information regarding the cone crusher
the equipment models were simple, limited, and did not consider wear. model utilized in the work either, but it seems that a steady-state model
Rylander and Axelsson (2013) used lean production guidelines to was utilized. Ruuskanen (2006) developed an optimization platform
identify potential control improvement opportunities in a quarrying based on linear models validated using laboratory-scale and full-scale
plant. Although the work did not provide new control frameworks, it is tests. The model calculates the crushing performance from the crusher
interesting because the application of the method in field studies design and operating variables and feed characteristics. Gang et al.
revealed the following: (2009a,b) utilized a flakiness and particle size distribution prediction
model as constraints to optimize the cone crusher throughput. The
• The main improvement potential was found in the areas of transport, models were also validated on full-scale and laboratory-scale tests. Lee
buffer/handling of inventory and customer delivery/loading. and Evertsson (2008) proposed an interparticle breakage model and
• The environment around the crushing site presents several compli studied the offline optimization of cone crushers with respect to their
cations to the installation of accurate sensors, especially wireless operating variables.
sensors.
5. Discussion
While et al. (2004) reported on the application of evolutionary al
gorithms for the optimal calculation of the liner profile, CSS, rotational 5.1. Summary
speed and eccentric angle of the mantle using simulation results of the
size distribution of the product, power consumed in the crusher and Tables 1 and 2 summarize the references listed in this work in
crusher capacity. This design problem is suitable for evolutionary al different contexts. Table 1 shows the number of publications by the first
gorithms in many ways because the payoff is high, the problem is too author’s institution and country. Europe leads the ranking with several
complex to solve analytically, the search space is too large to be explored
manually, a well-defined evaluation function exists for the problem, and
a straightforward representation scheme is easy to define. Single and Table 1
multiple objective fitness functions were studied, and according to the Summary of cone crusher related works arranged by the first author’s
author, the technique performs well enough to be extended to the institution.
flowsheet design of larger circuits. Institution Country Number of papers
Bengtsson et al. (2009) studied the flowsheet optimization of
Chalmers University of Technology Sweden 32
crushing circuits with the aim of reducing the number of active crushers Tampere University of Technology Finland 8
while keeping the product quality and throughput constant. A heuristic Luleå University of Technology Sweden 4
optimization algorithm was utilized to find operating settings that Shanghai Jiao Tong University China 3
maximize profit; however, according to the authors, more complete Universidad de Concepción Chile 2
Instituto Tecnológico Vale Brazil 2
models are needed that account for properties such as the chamber ge Metso Corporation Finland 1
ometry, stroke, eccentric speed, and ore type and properties. The work Hamersley Iron Pty. Ltd. Australia 1
by Bengtsson (2009), Bhadani et al. (2018) contemplates multiobjective University of Miskolc Hungary 1
optimization of crushing circuits by heavily relying on simulations. The Loughborough University of Technology UK 1
Pegson Ltd. England 1
effect of the variation in the flowsheet layout and design parameters on
University of Queensland Australia 1
the quality and capacity of the crushed product was observed, providing Mount Isa Mines Ltd. Australia 1
valuable information for the operator to achieve closer to optimal University of Utah USA 1
operation. University of Porto Portugal 1
Santos et al. (2020) proposed a simheuristic-based decision support Bougainville Copper Ltd. New Guinea 1
Majdanpek Copper Mine Yugoslavia 1
system for choosing the optimal amount of active equipment in each University of Western Australia Australia 1
crusher circuit stage. The solution is based on simheuristics since it uses Pontifica Universidad Católica de Chile Chile 1
a simulated plant model to evaluate the best solution. Real production FLSmidth A/S Denmark 1
scenarios of a Brazilian mining plant are used in the computational ex University of Chile Chile 1
Boliden Allis Sweden 1
periments. The results show a higher production rate and less energy
10
Table 2
Summary references in the literature, industries, model types, and proposed control strategies.
Reference Year of Publication Industry Model Type Subject
(Whiten, 1972) 1972 Mineral Processing Population Balance Model Validation

(Lynch et al., 1977) 1977 Mineral Processing Population Balance Simulation
(Manlapig and Watsford, 1983) 1983 Mineral Processing Other Dynamically Compensated Controller
(Herbst and Oblad, 1985) 1985 Mineral Processing Population Balance Simulation
(Machado Leite, 1990) 1990 Mineral Processing Other Offline Optimization
(Andersen and Napier-Munn, 1990) 1990 Mineral Processing Population Balance Simulation
(Bearman et al., 1991) 1991 Mineral Processing Other Model Validation
(Moshgbar et al., 1995) 1995 Quarrying Population Balance Adaptive Control
(Evertsson, 1995) 1995 Quarrying Population Balance Model Validation
(Csöke et al., 1996) 1996 Quarrying Empirical Offline Optimization
(Grujić, 1996) 1996 Mineral Processing Empirical Offline Optimization
(Evertsson and Bearman, 1997) 1997 Quarrying Population Balance Model Validation
(Alvarado et al., 1998) 1998 Mineral Processing Empirical Offline Optimization
(Evertsson, 1998) 1998 Quarrying Population Balance Offline Optimization
(Bearman and Briggs, 1998) 1998 Quarrying NA Offline Optimization
(Evertsson, 1999a) 1999 Quarrying Population Balance Offline Optimization
(Evertsson, 1999b) 1999 Quarrying Population Balance Offline Optimization
(Evertsson, 2000) 2000 Quarrying Population Balance Offline Optimization
(Evertsson and Lindqvist, 2002) 2002 Quarrying Population Balance Model Validation
(Lindqvist and Sotkovski, 2003) 2003 Quarrying Population Balance Model Validation
(Lindqvist and Evertsson, 2003) 2003 Quarrying Population Balance Model Validation
(Lindqvist and Evertsson, 2004) 2004 Quarrying Population Balance Model Validation
(While et al., 2004) 2004 Mineral Processing NA Offline Optimization
(Svedensten and Evertsson, 2005) 2005 Mineral Processing Population Balance Offline Optimization
(Sbárbaro, 2005) 2005 Mineral Processing Population Balance PID
(Ruuskanen, 2006) 2006 Quarrying Empirical Offline Optimization
(Bengtsson and Evertsson, 2006a) 2006 Quarrying Empirical Validation
(Hulthén and Evertsson, 2008) 2008 Quarrying Data Driven FSM
(Lee and Evertsson, 2008) 2008 Quarrying Population Balance Offline Optimization
(Hulthén and Evertsson, 2009) 2009 Quarrying Data Driven FSM
(Bengtsson et al., 2009) 2009 Quarrying Empirical Offline Optimization
(Gang et al., 2009b) 2009 Quarrying Empirical Offline Optimization
(Gang et al., 2009a) 2009 Quarrying Empirical Offline Optimization
(Johansson, 2009) 2009 Mineral Processing Population Balance Simulation Only
(Itävuo, 2009) 2009 Quarrying Empirical Offline Optimization
(Hulthén and Evertsson, 2010) 2010 Quarrying Data Driven Offline Optimization
(Hulthén and Evertsson, 2011) 2011 Quarrying Data Driven Evolutionary Operation
(Itävuo et al., 2011) 2011 Quarrying NA PID
(Lee and Evertsson, 2011) 2011 Quarrying Population Balance Simulation Only
(Asbjörnsson et al., 2012) 2012 Quarrying Empirical Simulation Only
(Hulthén et al., 2012b) 2012 Quarrying Empirical Offline Optimization
(Hulthén et al., 2012a) 2012 Quarrying Empirical Simulation
(Lee and Evertsson, 2011) 2012 Both Empirical Offline Optimization
(Atta et al., 2013) 2013 Mineral Processing Population Balance Extremum Seeking Control
(Asbjörnsson, 2013) 2013 Quarrying Empirical Simulation Only
(Asbjörnsson et al., 2013) 2013 Quarrying Empirical Simulation Only
(Itävuo et al., 2013) 2013 Mineral Processing Other Simulation Only
(Atta et al., 2014) 2014 Mineral Processing Population Balance Offline Optimization
(Itävuo et al., 2014) 2014 Mineral Processing Other PID
(Airikka, 2015) 2015 Quarrying Other Feed-forward PPI
(Bengtsson et al., 2015) 2015 Quarrying Empirical Offline Optimization
(Karelovic et al., 2015) 2015 Mineral Processing Population Balance HMPC
(Ma et al., 2016) 2016 Quarrying Population Balance Validation
(Asbjörnsson et al., 2016) 2016 Quarrying Population Balance Offline Optimization
(Itävuo et al., 2017) 2017 Quarrying Other PID
(Bengtsson et al., 2017) 2017 NA Offline Optimization
(Bhadani et al., 2018) 2018 Quarrying Empirical Offline Optimization
(Li et al., 2018) 2018 Mineral Processing Empirical Offline Optimization
(Albuquerque et al., 2019) 2019 Mineral Processing Population Balance ES
(Atta et al., 2019) 2019 Mineral Processing Population Balance NA
(Itävuo et al., 2019) 2019 Quarrying Other Mass Balance Control
(Santos et al., 2020) 2020 Mineral Processing Population Balance Offline Optimization
publications originating in Sweden and Finland, countries in which Table 2 summarizes the references by industry, model type and paper
mineral and aggregate production play a significant role in the economy. subject. The category ’Subject’ in the last column of Table 2 contextu
The strong presence of Metso and Sandvik in these countries, and their alizes the main contribution of the paper into the following categories:
cooperation with universities, such as Chalmers University of Technol
ogy and Tampere University of Technology, also influence the high • Model Validation
number of publications in these areas. South Africa, Australia, China, • Simulation
and Chile are the countries with the most expressive literature contri • Offline Optimization
butions, and as expected, mining also plays a significant role in these • Control Strategy: PID, FSM, MPC, etc.
countries.
11
Table 2 indicates that the cone crusher references are well dispersed Barnewold and Lottermoser (2020) shows that advanced control plat
between the mineral processing and quarrying industries. As previously forms are more common in large mining companies (revenue greater
stated, the production and, thus, the control goals differ, but both in than 1.0 billion US$). However, automation platforms, where PID blocks
dustries benefit from as many modeling and control alternatives as are easily encountered, are well-established despite the company size.
possible. The most reported control strategy was based on PID. This Maintenance of advanced process control systems is essential to ensure
finding is consistent with the results of an industrial survey of the continued performance (Bauer and Craig, 2008), which requires quali
mineral processing industry in general (Laurentz and Craig, 2017). This fied personnel and, consequently, incurs more costs (Jämsä-Jounela,
finding was expected since compared to the case of the petrochemical 2019).
industry, the state of the art of process control in the mineral processing
and aggregates industry still has much room for improvement. 5.2. Future trends
The overall analysis of the references mentioned here indicates that
there are a few well-established research lines in the cone crusher Ancillary process control instruments are among the most cited areas
literature, as follows: for improvement. For instance, the availability of novel and improved
sensors that are capable of reliably presenting product particle size
• Mass balance control of cone crushers and crushing circuits utilizing distributions could benefit the application of control frameworks in
the feeder speed at some level; crushing circuits (Hulthén, 2010). Process optimization strategies, such
• Development of a crushing performance map for crusher circuit as real time optimization, process intensification, lean production, and
flowsheet optimization; demand side management, applied in a plant-wide fashion in conjunc
• Crushing dynamics for mechanistic understanding of cone crushers tion with the intelligent usage of process data, could likely constitute be
using DEM; and the next step in processes that employ cone crushers. Therefore, it is
• Optimization of crushing circuits considering mixed-integer imperative that ancillary control instruments are available to measure
variables. and deliver feedback information regarding the crushing circuit per
formance. For example, hyperspectral cameras (Krupnik and Khan,
Wear modelling is another trend in the literature worth mentioning. 2019; Coelho et al., 2019) may be utilized in conjunction with flow rate
Andersen and Napier-Munn (1990) proposed a model for predicting how sensors to infer material hardness. If this information is available online,
liner wear affects the product particle size distribution. Moshgbar et al. control strategies might use it to improve production rates and optimize
(1995) developed a framework for controlling crusher performances crushing power consumption.
while compensating for the effect of wear on the product particle size. One possible and straightforward future industrial application of
Lindqvist and Evertsson (2003, 2006) developed a first principles model process data and machine learning techniques is the calculation of the
for predicting the effect of wear on the chamber geometry. More optimum CSS and eccentric speed settings for cone crushers, while
recently, Ma et al. (2016) developed a model that allows the effect of considering the trade-off between liner wear and crushing performance.
wear on the flakiness index and product particle size distribution to be For example, Li et al. (2018) reported an application of machine learning
predicted. techniques in a grinding circuit and pebble crusher to demonstrate that
Evidence from the application of control strategies in cone crushers changing the CSS of the crusher can indeed improve the overall pro
reported in the literature reveals that, most often, plant-wide control duction rate and energy consumption. The digital twin concept (Tao
could provide more tangible benefits than crushing stage control alone et al., 2019), which is a reliable and accurate digital model of a physical
because the operation of cone crushers may be limited by constraints counterpart, has begun to gain popularity in the context of Industry 4.0
imposed by either downstream or upstream equipment (Hulthén, 2010). and smart manufacturing. Commercial alternatives are already available
Some works from the literature already address the plant-wide control of for implementation (e.g. Emerson, 2019). As such models become more
mineral processing plants (le Roux et al., 2016; Burchell et al., 2017; Le popular and accessible, they might facilitate process control and the
Roux and Craig, 2019; Brooks et al., 2019; Thivierge et al., 2020); optimization of crushing circuits.
however, these works mostly focused on the grinding and flotation The development and popularization of reliable dynamic simulation
processes. packages not only for crushing circuits but also for mineral processing
DEM-based models of cone crushers are gaining attention in the plants in general may drastically improve operator training and enable
literature (Quist, 2012); however, their applications are limited to pre easier and affordable means of testing and improving advanced control
dicting the crusher performance from the crusher settings and feed strategies (Asbjörnsson, 2013). Similarly, Lichter et al. (2009) developed
properties in order to better understand the rock flow and breakage a virtual simulation platform considering a DEM-based model of cone
characteristics inside the crusher chamber and to aid in the design of crushers and reported on plans to improve the platform by including the
new crushing technology instead of for operation and process control. A effect of liner wear on crushing performance. Such features are already
considerable limitation of these methods is how computationally available for a virtual simulation platform developed by the same
expensive they become as the number of modeled particles increases. research group for grinding circuits. It is worth emphasizing that due to
We report many successful control applications in crushing circuits current technology limitations, DEM models are still only used for
ranging from regulatory control (PID and its variants) to advanced mechanistic understanding and equipment design.
process control (MPC, expert systems, fuzzy control, etc.). Selecting a Models will certainly continue to improve in the future, especially as
specific control strategy is not a trivial task and what to control, the cost new instruments emerge. First principles-based crusher models essen
for implementation, and maintenance simplicity should be considered. tially make assumptions regarding the following: (1) the breakage
As Hodouin (2011) highlighted, it is not easy to estimate the added value mechanism (e.g., breakage and selection functions); (2) a transport
of using advanced control strategies over decentralized PIDs. In some model (e.g., a CSTR and a classifier in Eq. 1); (3) and the wear evolution
cases, a well-designed regulatory control strategy is equivalent to or mechanism. As such, any instrument providing information on these
even more effective than advanced solutions. In general, one can say phenomena will improve their understanding and therefore allow the
that in SISO applications (e.g., ensuring that the crusher chamber re development of improved models with more accurate assumptions. For
mains filled or compensating for dead time in the crushing circuit), PID instance, since the true breakage events differ from the simplified
might be the recommended solution because of its robustness and breakage and selection representation of comminution models (King,
simplicity. On the other hand, for MIMO applications, when the crushing 2012), the existence of some device that is able to precisely measure the
circuit is treated holistically, advanced process control might be breakage process represents a prerequisite to the development of a more
encouraged. Regarding the cost of implementation, a survey by realistic representation. The transport model assumption also remains
12
difficult to validate since it remains impossible to sample the content of been used in published work. Cone crusher behavior simulations via
the crusher at different heights in real time, and thus, any device that DEM models can be classified as an intermediate between the usual
would allow this could cast some light on this hypothesis. Finally, empirical or first-principle models and prototype testing programs in
developing some sensors that could precisely map the pressure and shear terms of both representability and time consumption. As computer
stress along the whole surface of the liners could aid in the development hardware develops and becomes more powerful and algorithms become
of better wear and power draw models based on mechanistic principles more efficient, such a simulation method, which is similar in concept to
(e.g. Evertsson, 2000; Lindqvist and Evertsson, 2006). Additionally, we the commonly used computer fluid dynamics technique in other process
would like to point out that the availability of new instrumentation can industries, might be incorporated in the design and implementation of
improve cone crusher models with respect to modeling hazard events control strategies for the whole crushing circuit, or at least parts of it.
and validating transient operation. Regarding the former, in the oil and Mine-to-mill optimization strategies in which blasting, crushing,
gas literature, there are models of industrial compressors that take into grinding, and flotation processes are considered simultaneously to
account the internal pressure of the equipment to estimate when it is determine the best overall settings to maximize profit and minimize
operating dangerously close to or beyond the recommended pressure environmental, social, and political issues have the potential to drive
thresholds (Gravdahl and Egeland, 2012). In the mineral processing mineral processing operations remarkably close to a global optimum
industry, the feed of cone crushers may be contaminated with un (Bengtsson et al., 2017). The overall problem might be formulated as a
crushable elements (e.g., metal pieces, wood) that, in practice, lead to multidisciplinary optimization problem (Bhadani et al., 2018), and as
the activation of a safety mechanism to release the CSS. Regarding the suggested in the same reference, this way of formulating the problem
latter, it would be interesting to have a cone crusher model that is might improve the operation of sites in which different types of products
sensitive to the transients from equipment start-up, for example, how can be produced. The application of population-based algorithms to the
differently the cone crusher behaves when choke fed or when running flowsheet optimization of crushing circuits is a potential solution for
empty (Jacobson et al., 2010), particularly with respect to efficiency and addressing the complex iteration between the downstream and up
performance. Such a model could lead the discussion on how important stream processes of a mineral processing plant. While et al. (2004)
it is to avoid feeder interlocks in crushing circuits. proposed to extend the technique developed in their work to such an end
As electricity prices continue to increase and more demand is placed by considering a graphic representation of the crushing circuit flow
on mineral processing plants towards sustainable operations, proper sheet, which has already been successfully performed in other
control and optimization of cone crushers, which are infamously known industries.
for being energy inefficient (Holmberg et al., 2017) (although still more We close this section by summarizing the innovation drivers of
energy efficient than tumbling mills), will become more important. In process modeling and control in the mining and mineral processing in
this vein, cone crusher models that not only are able to incorporate both dustries. As stated previously, there are fewer process control imple
power consumption, flow, and crushing prediction capabilities (Atta mentations in crushing circuits than in grinding and flotation circuits;
et al., 2019; Quist, 2012; Bearman et al., 1991), but also are of a dynamic however we believe that the innovation drivers listed here currently
nature and not prohibitively expensive in terms of computational power affect, or will at some point in the near future affect, crushing circuits.
will be needed. Jämsä-Jounela (2007) argues that the future of automation in the pro
Mixed-integer models capable of predicting not only continuous cess industry lies in decentralized control strategies that contemplate
values, such as flowrate and power draw, but also discrete events and several smaller plants instead of large ones in a trend that aims to
rules, such as whether a cone crusher in the flowsheet is active and the address the localization, agility and flexibility issues sought by cus
frequency at which loading trucks deliver raw materials (Albuquerque tomers. Process control will then have to address the coupling of several,
et al., 2019; Asbjörnsson et al., 2016), are key for the implementation of multidisciplinary heterogeneous systems that will need different models
Hybrid MPC in crushing circuits. As shown in Karelovic et al. (2015), and, thus, different control strategies. In a similar vein, Pistikopoulos
once the models are sufficiently detailed, decentralized advanced con et al. (2021) argue that process intensification through modular design
trol implementations may be evaluated against plant-wide, centralized can also increase productivity and efficiency in the process industry.
implementations. Although most studies argue that the latter is the best Detailed modeling and simulation studies are required for the proper
way to optimize crushing plants, undeniably, due to problems such as understanding and usage of such process flowsheets, for example,
different dynamics between processes, availability of computational closed-loops between digital twins and their physical counterparts.
resources, and modeling complexity, decentralized strategies can be Specifically regarding the mining and mineral processing industry,
attractive if they deliver a comparable performance. Proof of this claim Rogers et al. (2019) claims that as the industry becomes more auto
is derived from the numerous reported implementations of decentral mated, each link of the production chain must contemplate whether or
ized PID control strategies over the more advanced, centralized control not distributed or centralized strategies are more advantageous. This
strategies used in many industries. decision should be guided by operating data analysis and the availability
Leon et al. (2018) developed a model for the prediction of the par of communication channels between the processes. On the other hand,
ticle size and mineral concentration as a function of the compression the mining and mineral processing industry might react differently to
ratio of crushing circuits. As the author emphasized, such a model is the trends of automation due to its inherent dependence on the local
attractive during the crushing stage optimization of rare metal ore availability of raw materials and other special operating conditions
extraction. As the final product is highly valuable, any improvement that (Barnewold and Lottermoser, 2020). Nonetheless, we believe that the
may be realized will translate into relevant profit. More research in this major drivers of process modeling automation and control innovations
area is expected in the near future. in crushing circuits are the following: (i) the paradigm of Industry 4.0
Quist and Evertsson (2010, 2012) proposed and evaluated DEM- and digitalization, more precisely, the ability to collect and analyze
based models against data from a real crusher. The authors identified large amounts of process data to understand and improve the process;
that such models are still too computationally heavy to be freely used for (ii) the political and societal pressure towards sustainability; and (iii)
such ends, especially when numerous particles are considered. It is stakeholder pressure towards competitiveness and profit. The following
possible to improve the efficiency of current computers by 10- to 100- paragraphs address each of the drivers in turn.
fold by utilizing GPU cores and CPU cores to solve the simulation; Rogers et al. (2019) reviews the state of automation in the mining
however, this enhancement could require a complete makeover of the industry and claim that automation currently suffers from a lack of
algorithm to allow for GPU parallelization. According to the authors, standards, equipment with different levels of automation, and missing
such optimized algorithms are expected to be available in the market integration. In the future, as automation becomes more prominent, plant
within a few years, but to the best of our knowledge, they have not yet operation will essentially shift the human bottleneck from the plant floor
13
to the control room. In this sense, humans will have to adequately tune important the mineral processing and quarrying industries are to the
and operate automated processes, and operation data must be used as a economy of that country and the influence of crushing equipment
source of feedback through data analysis techniques. In the context of manufacturers (Sandvik, Metso). Most works were written by first au
digitalization of the mining industry, the most important technologies thors whose institutions are located in Europe (Sweden and Finland),
are the following: automation, robotics, internet of things, big data, real Australia, South Africa and Chile. The most important research topics
time data, machine learning, artificial intelligence and 3D printing are (i) mass balance control of crushing circuits; (ii) crushing circuit
(Barnewold and Lottermoser, 2020). The mining industry views devel flowsheet optimization, (iii) mechanistic understanding of crushing
opment as a significant business risk and more than 37% of high level dynamics using DEM models, and (iv) optimization of crushing circuits
managers lack sufficient knowledge of digitalization technologies. using mixed-integer variables. As energy becomes more expensive and
Moreover, large companies (over 1 million tonnes of processed ROM per the relevance of sustainable processes increases, we expect that more
year) are more likely to experiment with digitalization, whereas small effort will be placed on modeling and controlling cone crushers and
companies need tailored solutions based on their specific needs (Bar crushing circuits in general, as this process is necessary but highly en
newold and Lottermoser, 2020). According to Jämsä-Jounela (2007), it ergy inefficient. The basic CSTR and classifier model from Whiten
is usually difficult to build detailed mechanistic models for the mining (1972) is still used in many recent works, but frameworks with more
industry; thus, data-driven models play an important role in process complex material transport models have emerged and show promising
control applications. Moreover, digital twins and feature-driven devel results. Expressing these models in the dynamic form allows for studying
opment models will advance the detailed dynamic models of commi fast disturbances. This greatly benefits the process control literature,
nution circuits (Jämsä-Jounela, 2019). especially for circuits without bins acting as buffers. In the near future,
The elusive goal of sustainability in the mining and mineral pro as computers become increasingly powerful and algorithms become
cessing industries has been sought by many research groups in the more efficient, DEM-based models, which are currently limited to un
literature. Bai et al. (2019) reported an attempt to establish a framework derstanding the fundamental aspects of rock flow and breakage inside
to measure and visualize the sustainability of stone mines. The authors the crusher chamber, will become interesting alternatives for modeling
used the concept of cleaner production, mapped as a set of 25 indicators, cone crushers. The mineral processing industry is lagging behind other
and analyzed the implementation of the proposed method in Laizhou, industries, such as the petrochemical industry, in terms of process con
China. Upadhyay et al. (2021) discusses how sustainability will affect trol and automation due to the process complexity, a lack of accurate
the mining industry from the circular economy point of view. The data sensors, and a lack of reliable models. Thus, most reported control ap
gathered from the sustainability reports of three major mining com plications consist of PID and ES implementations. As financial, envi
panies suggests that the mining sector is slowly shifting towards sus ronmental and societal needs drive the progression of the state of the art
tainable production and that emerging concerns regarding global process control and optimization of crushing circuits, solutions directed
warning, abandoning fossil fuel, and the lack of landfill availability can towards plant-wide optimization, economic optimization, and process
accelerate the transition. Eco-efficiency (Burchart-Korol et al., 2014; intensification based on advanced control strategies are likely to
Czaplicka-Kolarz et al., 2015) is another sustainability indicator that has emerge.
been used in the mining industry. The reader can find more definitions
and references in the review paper by Segura-Salazar and Tavares Declaration of Competing Interest
(2018). As sustainability, eco-efficiency, cleaner production, and the
circular economy become less abstract and more mensurable, these The authors declare that they have no known competing financial
variables can become controlled variables of integrated control prob interests or personal relationships that could have appeared to influence
lems, such as enterprise-wide optimization. the work reported in this paper.
Finally, we present some arguments that support how the pressure
for profit can drive process modeling and control in the mining and Acknowledgments
mineral processing industries. Kansake et al. (2019) claim that auto
mation should permeate the mining and mineral processing sector in the This work was supported by CNPq (Conselho Nacional de Desen
future, increasing its reach by 50% in 2030, and companies that have volvimento Científico e Tecnológico) [Grant Nos. 402759/2018-4 and
implemented automation solutions have reported a 15–20% increase in 444425/2018-7], ITV (Instituto Tecnológico Vale), Ouro Preto, Brazil,
productivity. As the mining industry depletes the natural mineral re and the SAG Conference Award Foundation. We also want to acknowl
serves, beneficiation plants must adapt to low grade raw materials edge the useful comments by the anonymous reviewers that contributed
through integrated planning of mining and mineral processing tasks to the quality of this work and Felipe Reis, who helped create the figures.
(Jämsä-Jounela, 2019). And as a side effect of the transition from fossil
energy to renewable energy, the mining industry must meet the demand References
for metals such as Neodymium, Indium, Lithium and Cobalt to build
wind turbines, electric vehicles and solar panels (Upadhyay et al., 2021). Airikka, P., 2015. Automatic feed rate control with feed-forward for crushing and
screening processes. IFAC-PapersOnLine 28, 149–154. https://doi.org/10.1016/j.
Such changes in society and the economy, especially the focus placed on ifacol.2015.10.094.
sustainability, can pose risks to the mining and mineral processing in Albuquerque, K., Silva, H., Teodoro, E., Fonseca, A., Garioli, G., Lopes, Ê., Cota, L.,
dustries but at the same time create new business opportunities (Niko Euzébio, T., 2019. Averaging level control of bulk solid material using a tripper car.
IFAC-PapersOnLine 52, 147–152.
laou and Evangelinos, 2010) that will need the support of process
Alvarado, S., Algüerno, J., Auracher, H., Casali, A., 1998. Energy-exergy optimization of
modeling and control. comminution. Energy 23, 153–158. https://doi.org/10.1016/S0360-5442(97)
00061-3.
Andersen, J., Napier-Munn, T., 1990. The influence of liner condition on cone crusher
6. Conclusions performance. Miner. Eng. 3, 105–116. https://doi.org/10.1016/0892-6875(90)
90084-O.
This work represents the effort of the authors to provide an extensive Andersen, J., Napier-Munn, T.J., 1988. Power prediction for cone crushers. In:
Australasian Institute of Mining and Metallurgy.
but not exhaustive review of the state of the art cone crusher technology
Anticoi, H., Guasch, E., Hamid, S., Oliva, J., Alfonso, P., Garcia-Valles, M., Bascompta,
while focusing on the development of process models and their appli M., Sanmiquel, L., Escobet, T., Argelaguet, R., Escobet, A., de Felipe, J.J., Parcerisa,
cation in the control and optimization of crushing circuits. The current D., Peña-Pitarch, E., 2018. Breakage Function for HPGR: Mineral and Mechanical
state of the art cone crusher literature indicates that works stem from the Characterization of Tantalum and Tungsten Ores. Minerals 8, 170. doi: 10.3390/
min8040170.
mineral processing and quarrying industries and that the number of Asbjörnsson, G., 2013. Modelling and Simulation of Dynamic Behavior in Crushing
publications per country is related to the following two factors: how Plants. Licentiate of Engineering Thesis Chalmers University of Technology.
14
Asbjörnsson, G., 2015. Crushing Plant Dynamics. Ph.D Thesis. Chalmers University of Coelho, P.A., Sandoval, C., Alvarez, J., Sanhueza, I., Godoy, C., Torres, S., Toro, C.,
Technology. Sbarbaro, D., 2019. Automatic near-infrared hyperspectral image analysis of copper
Asbjörnsson, G., Bengtsson, M., Hulthén, E., Evertsson, C.M., 2016. Modelling of discrete concentrates. IFAC-PapersOnLine 52, 94–98.
downtime in continuous crushing operation. Miner. Eng. 98, 22–29. https://doi.org/ Csöke, B., Pethö, S., Földesi, J., Mészáros, L., 1996. Optimization of stone-quarry
10.1016/j.mineng.2016.07.003. technologies. Int. J. Miner. Process. 44–45, 447–459. https://doi.org/10.1016/
Asbjörnsson, G., Hulthén, E., Evertsson, C.M., 2012. Modelling and dynamic simulation 0301-7516(95)00051-8.
of gradual performance deterioration of a crushing circuit - Including time Czaplicka-Kolarz, K., Burchart-Korol, D., Turek, M., Borkowski, W., 2015. Model Of Eco-
dependence and wear. Miner. Eng. 33, 13–19. https://doi.org/10.1016/j. Efficiency Assessment Of Mining Production Processes. Arch. Min. Sci. 60, 477–486.
mineng.2012.01.016. https://doi.org/10.1515/amsc-2015-0031.
Asbjörnsson, G., Hulthén, E., Evertsson, C.M., 2013. Modelling and simulation of Delaney, G., Morrison, R., Sinnott, M., Cummins, S., Cleary, P., 2015. DEM modelling of
dynamic crushing plant behavior with MATLAB/Simulink. Miner. Eng. 43–44, non-spherical particle breakage and flow in an industrial scale cone crusher. Miner.
112–120. https://doi.org/10.1016/j.mineng.2012.09.006. Eng. 74, 112–122. https://doi.org/10.1016/j.mineng.2015.01.013.
Åström, K., Hägglund, T., 1995. PID Controllers: Theory, Design, and Tuning, second ed. Duarte, M., Sepúlveda, F., Castillo, A., Contreras, A., Lazcano, V., Giménez, P.,
Instrument Society of America. Castelli, L., 1999. A comparative experimental study of five multivariable control
Atta, K.T., Euzébio, T., Ibarra, H., Moreira, V.S., Johansson, A., 2019. Extension, strategies applied to a grinding plant. Powder Technol. 104, 1–28. https://doi.org/
Validation, and Simulation of a Cone Crusher Model. IFAC-PapersOnLine 52, 1–6. 10.1016/S0032-5910(98)00210-1.
https://doi.org/10.1016/j.ifacol.2019.09.154. Eloranta, J., 1995. Influence of crushing process variables on the product quality of
Atta, K.T., Johansson, A., Gustafsson, T., 2013. On-line optimization of cone crushers crushed rock. Ph.D Thesis. Tampere University of Technology.
using Extremum-Seeking Control. In: 2013 IEEE International Conference on Control Emerson, 2019. Applying the digital twin to crushing circuits. https://www.emerson.co
Applications (CCA). IEEE, pp. 1054–1060. https://doi.org/10.1109/ m/documents/automation/flyer-applying-digital-twin-to-crushing-circuits-dynamic-
CCA.2013.6662891. simulation-en-5430478.pdf (accessed 28 March 2020).
Atta, K.T., Johansson, A., Gustafsson, T., 2014. Control oriented modeling of flow and Epstein, B., 1947. The material description of certain breakage mechanisms leading to
size distribution in cone crushers. Miner. Eng. 56, 81–90. https://doi.org/10.1016/j. the logarithmic-normal distribution.
mineng.2013.10.031. Evertsson, C., 1998. Output prediction of cone crushers. Miner. Eng. 11, 215–231.
Bai, S., Hua, Q., Cheng, L., Wang, Q., Elwert, T., 2019. Improve sustainability of stone https://doi.org/10.1016/S0892-6875(98)00001-6.
mining region in developing countries based on cleaner production evaluation: Evertsson, C., 1999a. Modelling of flow in cone crushers. Miner. Eng. 12, 1479–1499.
Methodology and a case study in Laizhou region of China. J. Cleaner Prod. 207, https://doi.org/10.1016/S0892-6875(99)00136-3.
929–950. https://doi.org/10.1016/j.jclepro.2018.10.026. Evertsson, C., 2000. Cone Crusher Performance. Ph.D. thesis. Chalmers University of
Barnewold, L., Lottermoser, B.G., 2020. Identification of digital technologies and Technology.
digitalisation trends in the mining industry. Int. J. Min. Sci. Technol. 30, 747–757. Evertsson, C., Bearman, R., 1997. Investigation of interparticle breakage as applied to
https://doi.org/10.1016/j.ijmst.2020.07.003. cone crushing. Miner. Eng. 10, 199–214. https://doi.org/10.1016/S0892-6875(96)
Bauer, M., Craig, I.K., 2008. Economic assessment of advanced process control–a survey 00146-X.
and framework. J. Process Control 18, 2–18. Evertsson, C., Lindqvist, M., 2002. Power draw in cone crushers. In: Minerals
Bearman, R., Barley, R., Hitchcock, A., 1991. Prediction of power consumption and Engineering Conference. Perth, Australia.
product size in cone crushing. Miner. Eng. 4, 1243–1256. https://doi.org/10.1016/ Evertsson, C.M., 1995. Prediction of Size Distributions from Compressing Crusher
0892-6875(91)90170-Z. Machines. In: EXPLO 95 Conference. Brisbane, Australia, pp. 173–180.
Bearman, R., Briggs, C., 1998. The active use of crushers to control product Evertsson, C.M., 1999b. Size Reduction in Cone Crushers. In: Minerals Engineering
requirements. Miner. Eng. 11, 849–859. https://doi.org/10.1016/S0892-6875(98) Conference ’99. Falmouth, England.
00072-7. Fernandez-Camacho, E., Bordons-Alba, C., 1995. Model predictive control in the process
Bengtsson, M., 2009. Quality-driven production of aggregates in crushing plants. Ph.D. industry. Springer.
Thesis. doi:10.13140/RG.2.1.2978.7608. Forbes, M.G., Patwardhan, R.S., Hamadah, H., Gopaluni, R.B., 2015. Model predictive
Bengtsson, M., Asbjörnsson, G., Hulthén, E., Evertsson, C.M., 2017. Towards dynamical control in industry: Challenges and opportunities. IFAC-PapersOnLine 48, 531–538.
profit optimization of comminution circuits. Miner. Eng. 103–104, 14–24. https:// Gang, D., Dongming, H., Xiumin, F., 2009a. Cone crusher chamber optimization using
doi.org/10.1016/j.mineng.2016.07.013. multiple constraints. Int. J. Miner. Process. 93, 204–208. https://doi.org/10.1016/j.
Bengtsson, M., Evertsson, C.M., 2006a. An empirical model for predicting flakiness in minpro.2009.07.005.
cone crushing. Int. J. Miner. Process. 79, 49–60. https://doi.org/10.1016/j. Gang, D., Xiumin, F., Dongming, H., 2009b. Analysis and optimization of cone crusher
minpro.2005.12.002. performance. Miner. Eng. 22, 1091–1093. https://doi.org/10.1016/j.
Bengtsson, M., Evertsson, C.M., 2006b. Measuring characteristics of aggregate material mineng.2009.03.020.
from vertical shaft impact crushers. Miner. Eng. 19, 1479–1486. https://doi.org/ Garcia, C.E., Prett, D.M., Morari, M., 1989. Model predictive control: Theory and practice
10.1016/j.mineng.2006.08.003. – A survey. Automatica 25, 335–348. https://doi.org/10.1016/0005-1098(89)
Bengtsson, M., Hulthén, E., Evertsson, C.M., 2015. Size and shape simulation in a tertiary 90002-2.
crushing stage, a multi objective perspective. Miner. Eng. 77, 72–77. https://doi. Gravdahl, J.T., Egeland, O., 2012. Compressor surge and rotating stall: Modeling and
org/10.1016/j.mineng.2015.02.015. control. Springer Science & Business Media.
Bengtsson, M., Svedensten, P., Evertsson, C.M., 2009. Improving yield and shape in a Grujić, M., 1996. Technology improvements of crushing process in Majdanpek Copper
crushing plant. Miner. Eng. 22, 618–624. https://doi.org/10.1016/j. Mine. Int. J. Miner. Process. 44–45, 471–483. https://doi.org/10.1016/0301-7516
mineng.2009.02.001. (95)00053-4.
Bhadani, K., Asbjörnsson, G., Hulthén, E., Evertsson, C.M., 2018. Application of multi- Herbst, J.A., Oblad, A., 1985. Modern Control Theory Applied to Crushing Part 1:
disciplinary optimization architectures in mineral processing simulations. Miner. Development of a Dynamic Model for a Cone Crusher and Optimal Estimation of
Eng. 128, 27–35. https://doi.org/10.1016/j.mineng.2018.08.029. Crusher Operating Variables. IFAC Proc. Vol. 18, 301–307. https://doi.org/10.1016/
Bhadani, K., Asbjörnsson, G., Hulthén, E., Evertsson, C.M., 2020. Development and S1474-6670(17)60525-1.
implementation of key performance indicators for aggregate production using Hodouin, D., 2011. Methods for automatic control, observation, and optimization in
dynamic simulation. Miner. Eng. 145, 106065. https://doi.org/10.1016/j. mineral processing plants. J. Process Control 21, 211–225.
mineng.2019.106065. Hodouin, D., Jämsä-Jounela, S.-L., Carvalho, M., Bergh, L., 2001. State of the art and
Bouchard, J., Desbiens, A., Poulin, É., 2017. Reducing the energy footprint of grinding challenges in mineral processing control. Control Eng. Pract. 9, 995–1005. https://
circuits: the process control paradigm. IFAC-PapersOnLine 50, 1163–1168. https:// doi.org/10.1016/S0967-0661(01)00088-0.
doi.org/10.1016/j.ifacol.2017.08.402. Holmberg, K., Kivikytö-Reponen, P., Härkisaari, P., Valtonen, K., Erdemir, A., 2017.
Bouchard, J., Sbarbaro, D., Desbiens, A., 2018. Plant Automation for Energy-Efficient Global energy consumption due to friction and wear in the mining industry. Tribol.
Mineral Processing. In: Awuah-Offei, K. (Ed.), Energy Efficiency in the Mineral Int. 115, 116–139. https://doi.org/10.1016/j.triboint.2017.05.010.
Industy: Best Practices and Research Directions. Srpinger. https://doi.org/10.1007/ Hulthén, E., 2010. Real-time optimization of cone crushers. Ph.D. Thesis Chalmers
978-3-319-54199-0_13. University of Technology.
Bouffard, S.C., 2015. Benefits of process control systems in mineral processing grinding Hulthén, E., Asbjörnsson, G., Evertsson, C.M., 2012a. A Training Simulator for Crushing
circuits. Miner. Eng. 79, 139–142. https://doi.org/10.1016/j.mineng.2015.06.006. Plant Operators. In: XXVI International Mineral Processing Congress, New Dheli,
Brooks, K., Westcott, M., Bauer, M., 2019. A Combined MPC for Milling and Flotation - A India, pp. 24–28.
Simulation Study. IFAC-PapersOnLine 52, 24–29. https://doi.org/10.1016/j. Hulthén, E., Asbjörnsson, G., Evertsson, C.M., 2012b. Tuning of Real-Time Algorithm for
ifacol.2019.09.158. Crushing Plants Using a Dynamic Crushing Plant Simulator. In: 8th International
Burchart-Korol, D., Korol, J., Agata, B., 2014. Development of Eco-Efficiency Evaluation Comminution Symposium. Cape Town, South Africa, pp. 17–20.
with Multicriteria Analysis for Steel Production. In: METAL Conference Proceedings. Hulthén, E., Evertsson, C.M., 2008. On-Line Optimization of Crushing Stage Using Speed
Brno, Czech Republic. Regulation on Cone Crushers. In: XXIV International Mineral Processing Congress,
Burchell, J., le Roux, J., Craig, I., 2017. Plant-Wide Control of a Base Metal Refinery: Beijing, China, pp. 2396–2402.
Top-Down Analysis. IFAC-PapersOnLine 50, 59–64. https://doi.org/10.1016/j. Hulthén, E., Evertsson, C.M., 2009. Algorithm for dynamic cone crusher control. Miner.
ifacol.2017.12.011. Eng. 22, 296–303. https://doi.org/10.1016/j.mineng.2008.08.007.
Caspeo, 2011. USIM PAC 3.2 Unit Operation Model Guidel. Hulthén, E., Evertsson, C.M., 2010. Two Variable Real-Time Algorithm for Cone Crusher
Cleary, P., Sinnott, M., Morrison, R., Cummins, S., Delaney, G., 2017. Analysis of cone Control. In: XXV International Mineral Processing Congress, Brisbane, Australia,
crusher performance with changes in material properties and operating conditions pp. 813–819.
using DEM. Miner. Eng. 100, 49–70. https://doi.org/10.1016/j.
mineng.2016.10.005.
15
Hulthén, E., Evertsson, C.M., 2011. Real-time algorithm for cone crusher control with Lynch, A., Bush, P., Unit, A., 1977. Mineral Crushing and Grinding Circuits: their
two variables. Miner. Eng. 24, 987–994. https://doi.org/10.1016/j. simulation, optimisation, design, and control. Elsevier Scientific Publishing Co, New
mineng.2011.04.007. York.
Itävuo, P., 2009. Dynamic Modeling of a Rock Crushing Process. Master’s Thesis. Ma, Y., Fan, X., He, Q., 2016. Prediction of Cone Crusher Performance Considering Liner
Tampere University of Technology. Wear. Appl. Sci. 6, 404. https://doi.org/10.3390/app6120404.
Itävuo, P., Hulthén, E., Vilkko, M., 2017. Feed-hopper level estimation and control in Machado Leite, M., 1990. Kinetic models for the simulation of crushing circuits. Miner.
cone crushers. Miner. Eng. 110, 82–95. https://doi.org/10.1016/j. Eng. 3, 165–180. https://doi.org/10.1016/0892-6875(90)90089-T.
mineng.2017.04.010. Maciejowski, J.M., 2002. Predictive control with constraints, vol. 17. Prentice Hall,
Itävuo, P., Hulthén, E., Yahyaei, M., Vilkko, M., 2019. Mass balance control of crushing London.
circuits. Miner. Eng. 135, 37–47. https://doi.org/10.1016/j.mineng.2019.02.033. Manlapig, E., Watsford, R., 1983. Computer Control of Lead/Zinc Concentrator Crushing
Itävuo, P., Jaatinen, A., Vilkko, M., 2011. Simulation and advanced control of transient Plant Operations of Mount ISA Mines Limited. IFAC Proc. Vol. 16, 465–475. https://
behaviour in gyratory cone crushers. In: 8th International Mineral Processing doi.org/10.1016/S1474-6670(17)64300-3.
Seminar, 2011. Santiago, Chile. McCoy, J., Auret, L., 2019. Machine learning applications in minerals processing: A
Itävuo, P., Väyrynen, T., Vilkko, M., Jaatinen, A., 2014. Tight feed-hopper level control review. Miner. Eng. 132, 95–109. https://doi.org/10.1016/j.mineng.2018.12.004.
in cone crushers. In: 27th International Mineral Processing Congress (IMPC). Morrell, S., 2019. Testing and calculations for comminution machines. SME Mineral
Santiago, Chile, pp. 224–229. Processing and Extractive Metallurgy Handbook, p. 529.
Itävuo, P., Vilkko, M., Jaatinen, A., Viilo, K., 2013. Dynamic modeling and simulation of Moshgbar, M., Bearman, R., Parkin, R., 1995. Optimum control of cone crushers utilizing
cone crushing circuits. Miner. Eng. 43–44, 29–35. https://doi.org/10.1016/j. an adaptive strategy for wear compensation. Miner. Eng. 8, 367–376. https://doi.
mineng.2012.07.019. org/10.1016/0892-6875(95)00002-8.
Jacobson, D., Janssen, P., Urbinatti, V., 2010. Cavity level’s effect on cone crusher Nikolaou, I., Evangelinos, K., 2010. A SWOT analysis of environmental management
performance and production. In: PROCEMIM 2010. Santiago, Chile, pp. 15–21. practices in Greek Mining and Mineral Industry. Resour. Policy 35, 226–234.
Jämsä-Jounela, S.-L., 2001. Current Status and Future Trends in the Automation of https://doi.org/10.1016/j.resourpol.2010.02.002.
Mineral and Metal Processing. IFAC Proc. Vol. 34, 9–16. https://doi.org/10.1016/ Olivier, L.E., Craig, I.K., 2017. A survey on the degree of automation in the mineral
S1474-6670(17)33175-0. processing industry. In: 2017 IEEE AFRICON. IEEE, pp. 404–409. https://doi.org/
Jämsä-Jounela, S.-L., 2007. Future trends in process automation. Ann. Rev. Control 31, 10.1109/AFRCON.2017.8095516.
211–220. https://doi.org/10.1016/j.arcontrol.2007.08.003. Ouchterlony, F., 2005. The Swebrec function: linking fragmentation by blasting and
Jämsä-Jounela, S.-L., 2019. Future Automation Systems in Context of Process Systems crushing. Min. Technol. 114, 29–44. https://doi.org/10.1179/037178405X44539.
and Minerals Engineering. IFAC-PapersOnLine 52, 403–408. https://doi.org/ Pistikopoulos, E.N., Tian, Y., Bindlish, R., 2021. Operability and control in process
10.1016/j.ifacol.2019.12.570. intensification and modular design: Challenges and opportunities. AIChE J. 67
Jiang, Y., Fan, J., Chai, T., Li, J., Lewis, F.L., 2018. Data-Driven Flotation Industrial https://doi.org/10.1002/aic.17204.
Process Operational Optimal Control Based on Reinforcement Learning. IEEE Trans. Pomerleau, A., Hodouin, D., Desbiens, A., Gagnon, É., 2000. A survey of grinding circuit
Industr. Inf. 14, 1974–1989. https://doi.org/10.1109/TII.2017.2761852. control methods: From decentralized PID controllers to multivariable predictive
JKTech, 2015. JKSimMet Version 6.1. user manual. controllers. Powder Technol. 108, 103–115. https://doi.org/10.1016/S0032-5910
Johansson, A., 2009. Modeling and Simulation of Cone Crushers. IFAC Proc. Vol. 42, (99)00207-7.
13–18. https://doi.org/10.3182/20091014-3-CL-4011.00004. Quist, J., 2012. Cone Crusher Modelling and Simulation. Master’s Thesis Chalmers
Kansake, B.A., Kaba, F.A., Dumakor-Dupey, N.K., Arthur, C.K., 2019. The future of University of Technology.
mining in Ghana: Are stakeholders prepared for the adoption of autonomous mining Quist, J., 2017. DEM modelling and simulation of cone crushers and high pressure
systems? Resour. Policy 63, 101411. https://doi.org/10.1016/j. grinding rolls. Chalmers University of Technology.
resourpol.2019.101411. Quist, J., Evertsson, C.M., 2010. Application of discrete element method for simulating
Karelovic, P., Putz, E., Cipriano, A., 2015. A framework for hybrid model predictive feeding conditions and size reduction in cone crushers. In: XXV International Mineral
control in mineral processing. Control Eng. Pract. 40, 1–12. https://doi.org/ Processing Congress (IMPC) 2010 Proceedings. Brisbane, QLD, Australia, pp.
10.1016/j.conengprac.2015.02.006. 3337–3347.
King, R.P., 2012. Modeling and simulation of mineral processing systems. Elsevier. Rogers, W.P., Kahraman, M.M., Drews, F.A., Powell, K., Haight, J.M., Wang, Y., Baxla, K.,
Krupnik, D., Khan, S., 2019. Close-range, ground-based hyperspectral imaging for mining Sobalkar, M., 2019. Automation in the Mining Industry: Review of Technology,
applications at various scales: Review and case studies. Earth Sci. Rev. 198 https:// Systems, Human Factors, and Political Risk. Min. Metall. Explor. 36, 607–631.
doi.org/10.1016/j.earscirev.2019.102952. https://doi.org/10.1007/s42461-019-0094-2.
Laurentz, E., Craig, I., 2017. Lights-Out Process Control - Analysis and Framework. In: le Roux, J., Skogestad, S., Craig, I., 2016. Plant-wide control of grinding mill circuits:
IEEE Africon. IEEE, pp. 398–403. Top-down analysis. IFAC-PapersOnLine 49, 72–77. https://doi.org/10.1016/j.
Le Roux, J.D., Craig, I.K., 2019. Plant-wide control framework for a grinding mill circuit. ifacol.2016.10.099.
Ind. Eng. Chem. Res. https://doi.org/10.1021/acs.iecr.8b06031. Ruuskanen, J., 2006. Influence of Rock Properties on Compressive Crusher Performance.
Lee, E., Evertsson, C., 2008. Yield and energy optimizations of compressive crushing. In: Ph.D. Thesis. Tampere University of Technology.
XXIV International Minerals Processing Congress, Beijing, China, pp. 316–324. Rylander, D., Axelsson, J., 2013. Lean Method to Identify Improvements for Operation
Lee, E., Evertsson, C., 2011. A comparative study between cone crushers and Control at Quarry Sites. In: 30th International Symposium on Automation and
theoretically optimal crushing sequences. Miner. Eng. 24 https://doi.org/10.1016/j. Robotics in Construction (ISARC 2013). Montreal, Canada: International Association
mineng.2010.07.013. for Automation and Robotics in Construction (IAARC). doi: 10.22260/isarc2013/
Leon, L.G., Bengtsson, M., Evertsson, C.M., 2018. Analysis of the concentration in rare 0010.
metal ores during compression crushing. Miner. Eng. 120, 7–18. https://doi.org/ Sandvik, 2020. What is the eccentric throw? https://www.rocktechnology.sandvik/en/cr
10.1016/j.mineng.2018.01.041. ushology/the-knowledge-hub/crushing-chambers/what-is-the-eccentric-throw/
Li, H., Evertsson, C.M., Lindqvist, M., Hulthén, E., Asbjörnsson, G., 2018. Dynamic (accessed 30 March 2020).
modeling and simulation of a SAG mill-pebble crusher circuit by controlling crusher Santos, M.S., Pinto, T.V., Júnior, Ê.L., Cota, L.P., Souza, M.J., Euzébio, T.A., 2020.
operational parameters. Miner. Eng. 127, 98–104. https://doi.org/10.1016/j. Simheuristic-based decision support system for efficiency improvement of an iron
mineng.2018.07.010. ore crusher circuit. Eng. Appl. Artif. Intell. 94, 103789.
Li, X., McKee, D., Horberry, T., Powell, M., 2011. The control room operator: The Sbárbaro, D., 2005. Control of crushing circuits with variable speed drives volume 16.
forgotten element in mineral process control. Miner. Eng. 24, 894–902. IFAC. doi:10.3182/20050703-6-CZ-1902.01692.
Lichter, J., Lim, K., Potapov, A., Kaja, D., 2009. New developments in cone crusher Sbárbaro, D., 2010. Dynamic Simulation and Model-based Control System Design for
performance optimization. Miner. Eng. 22, 613–617. https://doi.org/10.1016/j. Comminution Circuits. In: Sbárbaro, D., del Villar, R. (Eds.), Advanced Control and
mineng.2009.04.003. Supervision of Mineral Processing Plants Advances in Industrial Control. Springer,
Lindqvist, M., Evertsson, C., 2003. Prediction of worn geometry in cone crushers. Miner. London, London. doi: 10.1007/978-1-84996-106-6.
Eng. 16, 1355–1361. https://doi.org/10.1016/j.mineng.2003.08.011. Segura-Salazar, J., Tavares, L.M., 2018. Sustainability in the minerals industry: Seeking a
Lindqvist, M., Evertsson, C., 2004. Improved flow- and pressure model for cone crushers. consensus on its meaning. Sustainability (Switzerland) 10. https://doi.org/10.3390/
Miner. Eng. 17, 1217–1225. https://doi.org/10.1016/j.mineng.2004.06.028. su10051429.
Lindqvist, M., Evertsson, C., 2006. Development of wear model for cone crushers. Wear Stange, W., McInnes, C., 1995. The Intelligent Control of an ROM Milling Circuit. IFAC
261, 435–442. https://doi.org/10.1016/j.wear.2005.12.010. Proceedings Volumes 28(17), 359–365. https://doi.org/10.1016/S1474-6670(17)
Lindqvist, M., Sotkovski, P., 2003. Work hardening in cone crusher liners. In: Fourth 46785-1.
International Conference on Computer Applications in the Minerals Industries. Suriadi, S., Leemans, S.J., Carrasco, C., Keeney, L., Walters, P., Burrage, K., ter
Calgary, Canada. Hofstede, A.H., Wynn, M.T., 2018. Isolating the impact of rock properties and
Liu, Y., Spencer, S., 2004. Dynamic simulation of grinding circuits. Miner. Eng. 17, operational settings on minerals processing performance: A data-driven approach.
1189–1198. https://doi.org/10.1016/j.mineng.2004.05.018. Miner. Eng. 122, 53–66. https://doi.org/10.1016/j.mineng.2018.03.025.
Lotter, N., Baum, W., Reeves, S., Arrué, C., Bradshaw, D., 2018. The business value of Svedensten, P., Evertsson, C.M., 2005. Crushing plant optimisation by means of a genetic
best practice process mineralogy. Miner. Eng. 116, 226–238. https://doi.org/ evolutionary algorithm. Miner. Eng. 18, 473–479. https://doi.org/10.1016/j.
10.1016/j.mineng.2017.05.008. mineng.2004.08.008.
Lu, X., Kiumarsi, B., Chai, T., Jiang, Y., Lewis, F.L., 2019. Operational Control of Mineral Svensson, A., Steer, J., 1990. New cone crusher technology and developments in
Grinding Processes Using Adaptive Dynamic Programming and Reference Governor. comminution circuits. Miner. Eng. 3, 83–103. https://doi.org/10.1016/0892-6875
IEEE Trans. Industr. Inf. 15, 2210–2221. https://doi.org/10.1109/ (90)90083-N.
TII.2018.2868473. Tao, F., Zhang, M., Nee, A., 2019. Digital Twin Driven Smart Manufacturing. Academic
Press. https://doi.org/10.1016/C2018-0-02206-9.
16
Thivierge, A., Bouchard, J., Desbiens, A., 2020. Plant-wide economic model predictive Wei, D., Craig, I.K., 2009. Grinding mill circuits - A survey of control and economic
control application in mineral processing. In: Proceedings of the 59th Conference of concerns. Int. J. Miner. Process. 90, 56–66. https://doi.org/10.1016/j.
Metallurgists. Toronto, Canada. minpro.2008.10.009.
Tufa, L.D., Ka, C.Z., 2016. Effect of model plant mismatch on mpc performance and While, L., Barone, L., Hingston, P., Huband, S., Tuppurainen, D., Bearman, R., 2004.
mismatch threshold determination. Procedia Eng. 148, 1008–1014. A multi-objective evolutionary algorithm approach for crusher optimisation and
Upadhyay, A., Laing, T., Kumar, V., Dora, M., 2021. Exploring barriers and drivers to the flowsheet design. Miner. Eng. 17, 1063–1074. https://doi.org/10.1016/j.
implementation of circular economy practices in the mining industry. Resour. Policy mineng.2004.06.018.
72, 102037. https://doi.org/10.1016/j.resourpol.2021.102037. Whiten, W., 1972. The simulation of crushing plants with models developed using
Visioli, A., 2012. Research Trends for PID Controllers. Acta Polytechnica 52, 114–150. multiple spline regression. In: J. S. Afr. Inst. Min. Metall., pp. 257–264.
Wakefield, B.J., Lindner, B.S., McCoy, J.T., Auret, L., 2018. Monitoring of a simulated Uepg.eu. UEPG - estimates of aggregates production data, 2017. http://www.uepg.eu/s
milling circuit: Fault diagnosis and economic impact. Miner. Eng. 120, 132–151. tatistics/estimates-of-production-data/data-2017 (accessed 27 March 2020).
https://doi.org/10.1016/j.mineng.2018.02.007. Wills, B.A., Finch, J.A., 2016a. Chapter 5 - Comminution. In: Wills’ Mineral Processing
Weerasekara, N., Powell, M., Cleary, P., Tavares, L., Evertsson, C.M., Morrison, R., Technology, eighth ed. Elsevier. doi:10.1016/B978-0-08-097053-0.00005-4.
Quist, J., Carvalho, R., 2013. The contribution of DEM to the science of Wills, B.A., Finch, J.A., 2016b. Chapter 6 - Crushers. In: Wills’ Mineral Processing
comminution. Powder Technol. 248, 3–24. https://doi.org/10.1016/j. Technology, eighth ed. Elsevier. doi: 10.1016/B978-0-08-097053-0.00006-6.
powtec.2013.05.032. Zhu, Y., Patwardhan, R., Wagner, S.B., Zhao, J., 2013. Toward a low cost and high
performance mpc: The role of system identification. Comput. Chem. Eng. 51,
124–135.
17

Yamashita 2021

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Yamashita 2021

Uploaded by

Copyright:

Available Formats

Minerals Engineering 170 (2021) 107036

Contents lists available at ScienceDirect

1. Introduction quantitative data supporting the benefits of such strategies. A series of

time related to material transportation inside the crushing circuit. The

(Whiten, 1972) 1972 Mineral Processing Population Balance Model Validation

You might also like