Minerals Engineering xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

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

Simulation of particle flows and breakage in crushers using DEM: Part 1

– Compression crushers
Paul W. Cleary ⇑, Matthew D. Sinnott
CSIRO Computational Informatics, Australia

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

Article history: Discrete Element Method (DEM) simulation using a breakage model is applied to study a broad range of
Received 30 July 2014 existing compression based crushers to better understand machine level operation and to demonstrate
Revised 10 October 2014 that such modelling is now possible. Predictions for five types of crushers: jaw crusher, cone crusher,
Accepted 28 October 2014
gyratory crusher, impact crusher and double roll crusher are presented. The breakage method used is
Available online xxxx
based on facture of particles using geometric rules for progeny generation and estimates of the force
applied to the particles as they flow through the crusher. From these simulations, estimates of power,
Keywords:
product size, throughput rate and crusher wear are made.
DEM
Crusher
Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.
Breakage
Comminution
Wear
Compression

1. Introduction mills (Mishra and Rajamani, 1992, 1994; Rajamani and Mishra,
1996; Cleary, 1998a,b, 2001a,b,c, 2004, 2009; Herbst and Nordell,
Crushers are typically the first stage of comminution processing 2001; Cleary et al., 2003; Djordjevic, 2003, 2005; Kalala et al.,
for run of mine ore in mineral processing and for the manufacture 2008; Powell et al., 2011; Morrison and Cleary, 2004, 2008;
of aggregates and are also used in the preparation of aggregate, Cleary et al., 2008; Kalala et al., 2008; Powell et al., 2011; Wang
limestone and coal. Many different types of crushers have been et al., 2012; Delaney et al., 2013; Khanal and Jayasundara, 2014;
developed over the years for a varied range of applications, mate- Jonsén et al., 2014) and stirred mills (Sinnott et al., 2006, 2011;
rial types and required products. Typically design, optimisation Yang et al., 2006; Jayasundara et al., 2008, 2012; Plochberger and
and customisation of crushers have been performed using experi- Avila, 2014). A recent review of the use of DEM in comminution
mental approaches. Often measurement is of high-level features science is presented by Weerasekara et al. (2013). DEM modelling
such as laser scanning of worn liner profiles (Franke et al., 2012; has the advantage of being able to simulate the flow of particles
Rosario et al., 2004) and correlation of this wear with crusher oper- through equipment under very controlled conditions and to be
ating conditions, or monitoring of mass flow for real-time optimi- able to make detailed predictions of specific outputs while provid-
sation of performance (Hulthén, 2010). However measurement can ing insight into the nature of the flow patterns and breakage
be difficult, particularly for internal information such as how the processes.
particles are flowing, where they break and how. Input and output Detailed fracture in DEM has so far mainly been studied in sim-
information such as feed rate, feed size, product size, power draw ple scenarios involving single particles and simple geometries.
and wear on the mantle and concave can be and are measured and Fracturing of agglomerate grains has been partially explored using
provide useful global information on how the crusher is operating DEM. Usually this is performed by creating clusters of DEM parti-
over longer periods of time. cles which are bonded together in various ways. Upon impact
Discrete Element Method (DEM) is a computational technique the force at the external contact is transmitted through the particle
that allows particle flows in various types of equipment to be sim- network and bonds are broken if various force and/or bending
ulated. It has been used extensively in the simulation of tumbling moment limits are reached. Broken bonds are unable to transmit
tensile force which then modifies the force transmission network.
⇑ Corresponding author at: Private Bag 33, Clayton South, Victoria 3169, This enables predictions of the tensile strength (and other such
Australia. properties) and the resulting fragment size distribution. This
E-mail address: Paul.Cleary@csiro.au (P.W. Cleary). approach has been used by Thornton (1996), Thornton et al.

http://dx.doi.org/10.1016/j.mineng.2014.10.021
0892-6875/Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
2 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

(1999), Kafui and Thornton (2000), Tomas et al. (1999) and so far prevented the modelling of a full-scale crusher. Li et al.
Schubert et al. (2005) to explore breakage of small cohesively (2014) developed a PFC model of coarse size spherical feed in a
bonded powders and for concrete agglomerates. This approach is batch, lab-scale, simplified cone crusher geometry which also used
very suitable when the material is genuinely an agglomerate and a particle replacement model for the breakage. The effect of vary-
the DEM particles used to represent them have realistic sizes and ing the closed-side setting and angular speed on the predicted
shapes. If however, over-sized spherical particles are used then degree of size reduction qualitatively agreed with experimental
the connection to the underlying real microstructure is lost and trends but the model was too coarse to enable realistic
the DEM method is being used as a zero order fracturing contin- comparisons.
uum method. In this scenario it is likely to be better to use a gen- Overall crusher performance is a combination of breakage and
uine second order fracturing continuum method such as Smoothed transport (with complex interaction between them). The overall
Particle Hydrodynamics (SPH), Das and Cleary (2010). system scale behaviour is important particularly for crushers oper-
The bonded DEM approach has been used more broadly in geo- ated in non-dilute manner where substantial multi-body interac-
mechanical applications, including simulation of material property tions are important and where particle machinery interactions
tests, particularly with Particle Flow Code (PFC) in both two and (controlled by geometry) are non-trivial.
three dimensions (Potyondy and Cundall, 2004; Jiang et al., 2011; The aim of this paper is to provide a foundation for systematic
Schöpfer and Childs, 2013) and for attempting to characterise modelling of full size crushers by applying a reasonably sophisti-
breakage properties of rocks for milling or crushing (Refahi et al., cated DEM model to a wide range of compression and impact
2007; Whittles et al., 2006). This approach has also been used by crushers and to demonstrate the types of predictions that are
several groups of researchers to study rail ballast, including now possible. To facilitate this we use a relatively simple breakage
Hossain et al. (2007), Indraratna et al. (2009), Thakur et al. model within the DEM (which is comparable to the ones used by
(2010) who used PFC2D with simple breakage models and Lu others in the recent literature) to allow us to qualitatively predict
and McDowell (2010) who similarly used PFC3D. Lobo-Guerrero all the key outputs of interest for design and optimisation. This
and Vallejo (2006) explored the effect of cyclic loading on rock bal- model uses realistic crusher geometries and feed particle size
last along the center of a rail line using a 2D model with a breakage ranges and includes all the multi-body collisions that occur in
method that is very similar to the one used here. All these models these machines so as to predict flow, breakage locations and spatial
were very small in terms of numbers of particles and very simple in distribution of particle size, power and wear. This also allows some
their geometry configurations. comparison of relative behaviour for different classes of crusher
An alternative approach for simulating the detailed fracture of a that can help to understand the potential usage and advantages
non-agglomerate particle was proposed by Potapov and Campbell of each. This study is intended to be broad across most crusher
(1994, 1996) who modelled the particle as an assembly of bonded types in order to provide a base level for DEM crusher modelling.
rigid or deforming tetrahedral elements. This produced very realis- This is needed since most crushers have not yet been modelled
tic results but at very high relative computational cost. Recently, using DEM at the full machine scale or with realistic operating con-
Eliáš (2014) used a similar bonded polyhedral model for fracture ditions. Since this model uses spherical particles with a geometric
of very small numbers non-round particles in a compression test. (but fractal like) progeny distribution for each breakage event, it is
This used a very simple progeny model with each fracturing parti- unlikely that the predicted breakage size distributions will be
cle breaking into four relatively similarly sized draughter particles. quantitatively accurate, however this should be sufficient to enable
Simulation of crushing machines is much less common in the predictions of the other key quantities needed to help in under-
literature. This is due to the combination of the standing crusher performance. Detailed validation of breakage
product size distribution will be examined in future work using a
 requirement to include the breakage directly in the simulations more data driven breakage model. This study is structured into
in order to allow the particles to transit through the equipment, two parts with this first part focusing on compression based crush-
 larger model sizes (in terms of numbers of particles in the ers and the second part on impact crushers.
crushers), and
 high level of geometric complexity of the crushers.
2. Discrete Element Method (DEM)
Particularly in compression based crushers, which all use a pro-
cess based on squeezing particles in progressively narrowing space DEM simulation involves following the motion of every particle
between opposed surfaces, the DEM particles are unable to physi- (coarser than some resolved cutoff size) in the system and model-
cally pass through the machine unless the model actually breaks ling each collision between the particles and between the particles
them into smaller particles. Since most DEM software does not and their environment, which in this case is the crusher. The
include such capabilities, there is correspondingly less simulation crusher geometry is built using a CAD package and imported as a
of crushers compared to mills. Examples of DEM simulation of triangular surface mesh in the DEM simulation software. This pro-
crushers include models of lab-scale cone crushers (Herbst and vides essentially unlimited flexibility in specifying the complex
Potapov, 2004; Quist and Evertsson, 2010) and single particle three-dimensional geometries of real machines with which the
breakage in jaw crushers (Legendre and Zevenhoven, in press; particles interact. The general DEM methodology has been well
Refahi et al., 2010) all of which use bonded particle models. established and is described in early review articles by Campbell
Herbst and Potapov (2004) proposed a Discrete Grain Breakage (1990), Barker (1994) and Walton (1994).
(DBG) model that is representative of the method that more recent In this paper, we use a conventional linear-spring and dashpot
bonded sphere models of crushers use, but with bonded tetrahe- variant of DEM which is described in more detail in Cleary
dral elements. Lichter et al. (2009) replaced the DBG model with (1998a,b, 2004, 2009). Briefly, the particles are allowed to overlap
a Fast Breakage (FB) model which uses polygonal elements to rep- and the amount of overlap Dx, and normal vn and tangential vt rel-
resent DEM particles and a micro-scale Population Balance Model ative velocities determine the collisional forces via a contact force
(PBM) for particle replacement of the breakage progeny following law. There are a number of possible contact force models for
a fracture event. This was applied to a continuous fed lab-scale inelastic collision available in the literature, approximating the col-
cone crusher model and produced realistic predictions of through- lision dynamics to various extents, which are reviewed in Thornton
put, power and product sizes, although the computational cost has et al. (2013).

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 3

The normal force:

F n ¼ kn Dx þ C n v n ; ð1Þ
consists of a linear spring to provide the repulsive force and a dash-
pot to dissipate a proportion of the kinetic energy. The maximum
overlap between particles is determined by the stiffness kn of the
spring in the normal direction. The normal damping coefficient Cn
is chosen to give the required coefficient of restitution e (defined
as the ratio of the post-collisional to pre-collisional normal compo-
nent of the relative velocity). An easy method of calculation is given
by Thornton et al. (2013).
The tangential force is given by:
n X o
F t ¼ min lF n ; k t v t Dt þ C t v t ; ð2Þ

where the vector force Ft and velocity vt are defined in the plane
tangent to the surface at the contact point. The integral term repre-
sents an incremental spring that stores energy from the relative
tangential motion and models the elastic tangential deformation
of the contacting surfaces, while the dashpot dissipates energy from
the tangential motion and models the tangential plastic deforma-
tion of the contact. Depending on the history of the contact, it is
possible for the spring to be loading in one direction and simulta-
neously unloading in the orthogonal direction. The total tangential
force Ft is limited by the Coulomb frictional limit lFn, at which point
the surface contact shears and the particles begin to slide over each
other.
The discrete element algorithm has three principal stages:

1. A search grid is used to periodically build a near-neighbour

interaction list that contains all the particle or particle – bound-
ary pairs that are likely to experience a collision in the short
term. Using only pairs in this list reduces the force calculation
to an O(M) operation, where M is the total number of particles.
2. The forces on each pair of colliding particles and/or boundary
objects are evaluated in their local reference frame using the
spring-dashpot interaction model, and then transformed into
the simulation frame of reference.
3. All the forces and torques on the particles and boundary objects
are summed and the resulting equations of motion are inte-
grated. Time integration is performed using a second-order pre-
dictor–corrector scheme.

The computational cost of the DEM method is then linearly pro-

portional to the number of particles. When DEM is used with
breakage, the number of smaller particles can grow dramatically
due to the breakage and this will strongly affect the compute dura-
tion. Nonetheless, simulation of industrial scale crushers with rea-
sonable resolution of the feed and product size distributions is
easily feasible on current desktop computers.
Quantitative predictions of dynamic boundary stresses or loads,
wear rates and distributions, collision forces, energy spectra,
power consumption, torques and flow rates, sampling statistics,
mixing and segregation rates and many other quantities can be
made from the information available in DEM simulations. For more
details on the simulation method, data analysis and application to
mills, see Cleary (1998a,b, 2004), Cleary and Sawley (2002) and
Morrison and Cleary (2004, 2008). Fig. 1. Circular particle packing method which controls progeny size and location
within the volume of the parent particle.

3. Dynamic breakage in a DEM model

is quite different in different parts of the crusher. Here we use a
Dynamic breakage must be included in the DEM model in order replacement strategy breakage model from Cleary (2001c).
to simulate compression based crushers since the ability of parti- Most particle breakage has been modelled in DEM by construct-
cles to move through the decreasingly small gaps depends on the ing the macro-particles from smaller sub-particles. Most research-
particles becoming smaller. It is also very desirable for impact ers studying breakage have used bonded round particles since this
crushers since the rock size distribution can change rapidly and is algorithmically simple to do, see Thornton (1996), Thornton

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
4 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

et al., (1999), Kafui and Thornton (2000), Tomas et al. (1999) and Calibration of the energy and force thresholds used in determin-
Schubert et al. (2005) for examples of mostly agglomerates. ing whether breakage occurs for specific particles according to these
Potapov and Campbell (1994, 1996) in contrast used triangular/tet- mechanisms is not at all straight forward. Quantities that can be
rahedral sub-particles to model both impact and compression experimentally measured such as the impact energy for breakage
breakage of single particles. This allows very detailed analysis of (using Drop Weight or Rotary Breakage Testers) and fracture tough-
particle breakage under a range of conditions. Unfortunately the ness or other rock mechanic compressive breakage measures are not
computational cost of this approach is linearly dependent on the easily related to the calculated quantities used in DEM breakage
number of sub-elements. In an industrial scale simulation this models. Another critical issues is that for compressive breakage,
would typically increase the cost by factors of 103 to 104, depend- tests such as with uniaxial, biaxial, triaxial or Brazilian tests are all
ing on the accuracy that one wished to obtain for the breakage. performed at slow or very slow strain rates since they were designed
An alternative approach (proposed by Cleary, 2001c) is to use a for measuring rock properties in geomechanical situations. In crush-
rule base to identify the conditions under which particles would be ers, the strain rates are several orders of magnitude higher than in
expected to fracture and to then replace the fracturing parent par- traditional rock mechanics testers. Most materials demonstrate
ticles with appropriate assemblies of daughter particles packed strain rate dependent deformation and fracture properties, particu-
into the space occupied by the parent. This is possible in DEM since larly at very high strain rates and so this type of data is not obviously
complete information is known about all particles and their inter- usable. Therefore we have adopted a very pragmatic approach of cal-
particle forces. In physical terms this takes advantage of the fact ibrating the breakage energies or forces to give output behaviours
that the fracture timescale (governed by the speed of sound in for the crushers which sensibly correspond to the real operating
the rock) is orders of magnitude faster than the DEM timescale conditions of these crushers. This means that for each crusher we
(which is the timescale for the rearrangement of particles around have a known feed size, a known geometry and operating conditions
any particular rock). The fracture process is essentially instanta- and a reasonable understanding of what the product size distribu-
neous on the timescale of the particle motion and so the fracture tion should be. This information is used to choose sensible represen-
and replacement can be performed as an instantaneous operation tative breakage levels for each case. Calibrating breakage data from
in any given DEM timestep. measured input and output data is likely to be the best strategy for
The two particle breakage mechanisms used in this paper are: quite some considerable time. The development and usage of suit-
able laboratory characterisation tests (that can be used to measure
1. High energy impacts: Particles are broken if the cumulative a priori rock properties relevant to crusher strain rates) and the
energy absorption by the particle during a collision exceeds a refinement of the DEM breakage rule bases to use this measured
size dependent breakage energy. The energy absorption in a col- data and which are then able to directly predict with quantitative
lision is given by the energy dissipated by viscous and frictional accuracy the product of crushers will require significant research
elements of the contact model. This energy can be calculated for breakthroughs.
each collision and can be accumulated over the collision history In this paper, the parent and daughter fragments are all mod-
of each particle and represents the damage that has been done elled as spherical. The daughter fragment sizes are chosen using
by these collisions. If this accumulated damage energy exceeds geometrical criteria to optimally pack the space occupied by the
its breakage threshold energy then by this mechanism the par- parent particles. This process is illustrated in two dimensions in
ticle fractures. Fig. 1. In step one, the largest fragment is created first. This has a
2. Compressive fracture: Particles are broken if the total instanta- radius R1 that is at least half the radius of the parent particle
neous force applied to a particle exceeds a specified size depen- (shown as the outer disc) and is shown as the upper orange frag-
dent compressive stress or force threshold. This calculation is ment. The second fragment is chosen to have a radius R2 so that
performed as a simple sum of the magnitudes of all the contact R1 + R2 = Rparent. This particle can be seen in step 1 of Fig. 1 as the
forces at each time. In this simple model a scalar sum is used lower orange fragment. The cleavage plane between the first and
rather than a vector one in order to simplify the calculation pro- second fragments is chosen to pass through the contact point of
cess. In crushers, particles when highly stressed are typically the particles undergoing the collision that led to the breakage
compressed in only one direction by being trapped between event. This reflects the expectation that crack initiation is likely
two converging surfaces (which could be particles or the to commence at the contact point and will control the crack gener-
crusher liner). There can be contacts with other particles but ation between these two fragments. The effect of this is to control
these are typically comparatively weak. So the scalar sum is the orientation in world coordinates of the packing structure gen-
dominated by the two principle compressive contacts which erated from this process. In the canonical frame of the fracturing
determine the magnitude of the principle tensile force within particle shown in Fig. 1, the first two fragments are stacked verti-
the particle which when sufficiently large leads to crack initia- cally and the contact point is to the side leading to the fracture
tion and fracture (as demonstrated by Das and Cleary, 2010). between fragments 1 and 2.

Fig. 2. Simple breakage of two particles in a high speed collision.

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 5

Fixed jaw

Moving jaw

Fig. 3. Two views of a jaw crusher showing the front and back of the jaw at two jaw locations. The jaw on the right is the moving one.

The primary fragment diameter is chosen randomly in a range the generation of very large numbers of very fine progeny that
from Smin to Rparent where Smin is the smallest primary fragment can significantly slow such a DEM simulation. In the crusher sim-
radius allowed and Rparent is radius of the parent particle that is ulations used here we would typically seek to resolve in excess
fracturing. In general Smin is chosen to be dependent on the colli- of 90% of the progeny. The mass that is not represented explicitly
sion energy or the fracture force with it decreasing with increasing in the DEM model after fracture represents finer fragments that
intensity of the breakage event. So for weak collisions or breakage cannot be resolved. This material is automatically omitted from
events, Smin is close to Rparent but as the breakage events become the simulation. This missing mass is actually useful as an explicit
stronger Smin decreases towards Rparent/2. This means that for weak prediction from the model as it represents the mass of finely bro-
events there is a single large daughter fragment surrounded by ken product. This can be accumulated over the simulation and
finer particles but as the breakage event becomes stronger the pri- reported as part of the crusher product.
mary and secondary fragments become closer in size which also It should be noted that with this packing algorithm, the only
then increases the average size of the next several progeny gener- random variable and the only contact model dependent variable
ated. In this way the shape of the progeny size distribution is con- used are the choice of R1 and Smin respectively. All fragment radii
trolled by the intensity of the breakage event. Ri for i > 1 are fully and purely determined by the geometry of
In step 2 (see Fig. 1), two particles (coloured1 orange) are added the space left between the growing cluster of progeny and the sur-
– one on either side of the first two fragments. Their sizes are chosen face of the parent. The packing structure formed is fractal like and
to be the largest that can fit into the available space with contact is consistent with a situation where cracks radiating from the
made with both the first two fragments and also touching the outer cleavage plane sub-divide as they approach the surface of the par-
surface of the parent particle. The process is very similar in three ent producing a network of increasingly fine cracks and therefore
dimensions with a horizontal layer of particles being added at the fragments.
one height level around the periphery of the first two fragments. Fig. 2 shows an example of the operation of the breakage model
For step 2, this is four particles. for a simple high speed two particle collision in three dimensions.
In step 3, a second layer of smaller fragments (again coloured The parent particles in Fig. 2a collide and dissipate enough energy
orange in Fig. 1) is added into the gaps between the particles cre- in the normal direction to fracture and are replaced by the daugh-
ated in the previous steps. A new fragment is added between each ter fragments in Fig. 2b. These fragments are then free to move
adjacent pair of surface particles in the current cluster configura- apart (Fig. 2c) and interact with other particles and boundaries of
tion. In two dimensions, this layer generates four particles which the crusher, perhaps breaking again if conditions warrant this.
are smaller in size reflecting the decreasing free space left between In reality, rocks are not spherical and this will affect both the
the growing cluster of progeny and the surface of the parent parti- particle flow and the breakage in the crusher to some degree. This
cle. This process continues in steps 4 and 5 as a third and then can be expected to affect the quantitative accuracy of the product
fourth layer of progressively finer fragments are added between size distributions predicted. Non-round particle shape often affects
each adjacent pair of particles from the previous steps. This gener- the behaviour of particles under shear (Cleary, 2008) and the over-
ates 8 and the 16 fragments respectively for these steps. Further idealisation of particles as being round can sometimes create sig-
layers can be added if required, but typically four layers allows a nificant quantitative errors in DEM flow predictions (as shown by
very large fraction (up to 98%) of the parent volume to be repre- Cleary et al. (1998) with DEM incorrectly predicting mixing rates
sented by the progeny particles. in a rotating drum when cubic salt particles were represented as
Typically a minimum fragment size below which progeny par- discs). However, the focus here is on predicting crusher scale par-
ticles are not resolved by the DEM simulation is specified. In would ticle flows and machine outputs such as power draw and wear
be common to set this to match the minimum size used to truncate which are expected to be much less sensitive to the particle shape.
the feed size distribution. This lower cut-off is needed to control The results predicted here are qualitatively reasonable which sug-
gests that particle shape is not a dominant factor. Nonetheless, the
1
For interpretation of colour in Figs. 1, 15 and 24, the reader is referred to the web absence of realistic shape in this generation of crusher models can
version of this article. still influence the results. Quantitative prediction of the product

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
6 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

Table 1 feed as closely as possible. The breakage generated in all the com-
Key geometric and operating details for the jaw crusher. pression based crushers analysed here uses the compression
Height (available crushing region) 1920 mm breakage mechanism described in Section 3.
Feed opening 800  1010 mm For each of these crushers, predictions will be made for:
Feed size (maximum) 600 mm
Discharge opening 200  1010 mm
Inclination of moving jaw 20.56 deg
1. Power.
Clearance between moving jaw and side walls attached to 5 mm 2. Product size.
fixed jaw 3. Flow rates.
Angular speed of jaw about pivot point 200 RPM 4. Crusher wear.
Period of crushing Half-cycle
Product top size 200 mm
The details of operation of each crusher, its geometric details
and the feed size distribution are given for each crusher section
below.

Fig. 4. DEM prediction of particle flow through a jaw crusher system showing flow
along and through the grizzly feeder with coarse particles falling into the jaw
crusher on the right and finer material bypassing through the chute below. The
particles are coloured by size with red being large, green being intermediate and
blue being the smallest. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

size distribution and the effect of rock shape will be examined in

future work.

4. DEM parameters and outputs for modelling of crushers

All the simulations in this paper were performed using a linear-

spring dashpot collision model. The coefficient of restitution used
for rock–rock contact was 0.3 and for rock–steel it was 0.5. The
friction coefficient used was 0.5. These are taken from Cleary and Fig. 5. DEM prediction of particle flow through and breakage within a jaw. The
Morrison (2012). The spring constant typically used is 108 N/m. particles are coloured by size with red being large, green being intermediate and
blue being the smallest: (a) particles are loaded during the compression phase, (b)
These are fairly typical material and numerical parameters for immediately post fracture, and (c) after re-arrangement of the fragments during the
the simulation of rock ore. The particles are modelled as spheres opening part of the cycle. (For interpretation of the references to colour in this
with sizes and density chosen to be representative of the crusher figure legend, the reader is referred to the web version of this article.)

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 7

Table 2 feeder. This is a vibrating feeder with long streamwise slots that
Key simulation predictions for the jaw crusher. allow finer material to fall through into a collection chute below
Power draw 44 kW and then to flow directly onto the product conveyor below short
Throughput 250 kg/s (900 tph) circuiting the crusher. This is done to limit the flow of fine material
Crusher charge when full 1.5 tonnes into the crusher which can reduce its performance. Coarser feed
Energy absorbed by fixed jaw 12.5%
Energy absorbed by moving jaw 9.6%
material flows from the end of the grizzly feeder directly into the
Discharge speed 4 m/s opening at the top of the jaw crusher.
The rock breakage threshold force for fracture ranged from
106 N for the top size (which is comparatively weaker) to 109 N
for the bottom size. The minimum breakage fragment resolved in
5. Jaw crusher
the simulation was 10 mm. The feed size distribution is uniform
with a minimum of 100 mm and a top size of 600 mm. The slot
Jaw crushers are often used for primary crushing of hard or
width of the grizzly feeder is 100 mm. Fig. 4 shows the particle
abrasive run of mine ore at the start of a crushing circuit in order
flow through the jaw crusher system from two views. The view
to provide size reduction sufficient for further crushing stages.
from the top shows the broad size range of feed flowing along
The crusher consists of two sets of jaws, one jaw being fixed and
the grizzly feeder, the flow of the finer feed into the bypass chute
the other moving back and forth in an eccentric cyclic motion.
and the coarser feed falling into the top of the crusher. The lower
The jaws are farther apart at the top than at the bottom, forming
view shows the view from below the feeder with the collection
a tapered chute that allows material to be crushed progressively
chute rendered as transparent. The flow of dark and light blue finer
to smaller and smaller sizes as it moves down through the crusher.
feed along the chute and down onto the conveyor belt is clearly
Once a particle is small enough to escape from the bottom opening
visible. This model contains around 2200 particles with a weight
it falls from the crusher as product. The amplitude of the jaw
of 4700 kg. The computational speed, as measured by the Cundall
motion is typically quite small, producing only partial crushing
with each stroke. The jaw motion also helps transport material
down through the crusher.
Fig. 3 shows a typical jaw crusher which is used in these DEM
simulations. This one has a vertical stationary jaw and an inclined
moving jaw. The key geometric and operating information for this
crusher are given in Table 1. The model is demonstrated for one
feed rate, but it is equally applicable for all feed rates including
ones that can overload the machine. For this single toggle jaw
crusher, there is a circular motion about the upper shaft (shown
as the hole in Fig. 3) and much smaller amplitude oscillation at that
the bottom of the crusher which moves in an elliptical trajectory
with small in/out motion and larger motion along the direction
of the oscillating jaw. Mathematically this can be decomposed into
a linear oscillation along the direction of the jaw and an simple
harmonic variation of the rotation about the center of the shaft
at the top of the moving jaw. The two oscillations have the same
frequency and are in phase. The auxiliary equipment supporting
the crusher is also modelled as its operation affects the material
Fig. 7. Energy absorption spectra for the particles in a jaw crusher.
fed into the crusher. In this system, material is supplied by a grizzly

Fig. 6. Wear increment distributions on the two working surfaces of the jaw crusher from a short period of operation, (a) abrasion, and (b) impact damage.

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
8 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

Bowl

Mantle

Fig. 8. A gyratory crusher with the mantle shown at two extreme positions through its motion.

Table 3 rates of these surfaces. The energy absorbed by the fixed jaw is lar-
Key geometric and operating details for the gyratory crusher. ger because its area is larger and the feed stream strikes it and par-
Concave height 3731 mm ticles initially flow more down along this surface. The combined
Concave internal diameter (maximum – top of crushing region) 2930 mm energy consumption by the crusher is nearly 1=4 of the energy sup-
Concave internal diameter (minimum – bottom of crushing 2202 mm plied to the system and is quite large reflecting the dominance of
region)
the particle-crusher surface interactions compared to the parti-
Mantle diameter (maximum) 1859 mm
Mantle height 3000 mm cle–particle ones.
Mantle side angle (from horizontal) 75 deg Fig. 6 shows the resulting wear patterns on the jaw crusher
Feed opening (b/w concave and mantle shield) [open side] 966 mm working surfaces after 10 s of operation. The abrasion wear is
Feed size (maximum) 600 mm shown in the left and the impact damage is shown on the right.
Discharge opening [closed side setting] 150 mm
Eccentric throw 90 mm
The impact damage is concentrated on the central upper section
Angular speed of mantle 500 rpm of the inclined moving jaw where the feed stream lands. The distri-
Period of crushing Continuous butions are quite uneven over short periods of operation as a result
Product top size 150 mm of the specific collisions and interactions that have occurred. Long
term wear becomes progressively more even as the specific events
are averaged out. Fig. 7 shows the energy absorption spectrum of
number (Cu) which is the number of particle timestep per cpu sec- the particles in a jaw crusher. This has a broadly normal distribu-
ond, was 230,000. tion with a peak collision frequency of 800 collisions/s at an energy
Fig. 5 shows the particle flow and breakage within the jaw level of around 1 mJ. Higher energy collisions occur less frequently
crusher itself with frames just before and after a number of break- with very high energy collisions of 300 J or above occurring with a
age events. In Fig. 5a there is a large red particle in the top right of frequency of less than 1.0 Hz. Weaker collisions are also less
the crusher and two smaller orange particles near the bottom frequent.
which are all in contact with both jaws and are being compressed.
This decreasing size of the particles with distance down into the 6. Gyratory crusher
crusher is a result of the decreasing width of the gap between
the jaws of the crusher. In Fig. 5b, all three of these particles have Gyratory crushers are also used for primary crushing operations
fractured as the compressive force applied exceeds the fracture but provide larger capacity compared to a jaw crusher. A gyratory
threshold of each particle. The large red particle has broken into crusher also uses the principle of cyclic compression with alternat-
a smaller but still quite large red particle and a collection of finer ing squeezing of the particles as the crusher working surfaces
material (light and dark blue). The fracture of the smaller orange approach each other and downward flow of the particles as they
particles is more complete with them fragmenting into fine prog- move apart. The geometry and the way in which this is achieved
eny. Note the flow of fine product material (coloured blue) falling are quite different for the gyratory crusher compared to that of a
from the bottom opening of the crusher. In Fig. 5a, the fragments jaw crusher. It consists of a cylindrically symmetric stationary
have now started to re-arrange, falling lower into the crusher in outer bowl shaped structure called the concave and an inner cylin-
preparation for the next compression cycle. The jaw crusher oper- drically symmetric inner structure called the mantle. The concave
ates in a relatively dilute mode so that product particles are rela- is oriented with its wide end upward and narrows progressively as
tively free to quickly discharge from the crusher. Fracture is one moves lower into the machine. The mantle is oriented oppo-
dominantly caused by large particles being loaded by contact with sitely with its apex at the top and increases in cross-sectional
both jaws and there is little indication of multi-particle interac- diameter with depth into the crusher. The space between the man-
tions influencing the fracture process. tle and concave is where the rocks are located and it becomes nar-
Key output predictions for this mill are given in Table 2. The rower with depth and has a minimum gap which occurs near the
power draw for this crusher is predicted to be 44 kW which is rea- bottom of the mantle. The mantle is tilted slightly to one side so
sonable for such a unit. The resident load contained in the crusher that the gap is smaller on that side and larger on the other. The
is on average, 1.5 tonnes and the throughput is predicted to be inclined axis of the mantle then rotates at a constant rate around
around 900 tph for this type of generic hard rock feed material, the crusher. When viewed from directly above, this axis rotates
which is again quite reasonable. The speed of discharge of fine in a circular motion. Note that the mantle itself does not rotate.
product at the bottom of the crusher is about 4 m/s. With DEM, As the mantle axis rotates its surface approaches the concave in
the usage of the energy supplied to the machine can also be pre- the radial direction and then retreats radially so there is no shear
dicted. In this case 12.5% of the power is absorbed by the fixed between the surfaces and no change in orientation of the mantle
jaw and 9.6% by the moving jaw. This energy controls the wear when viewed from above. The point of closest approach of the

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 9

Table 4
Key simulation predictions for the gyratory crusher.

Power draw 630 kW

Throughput 460 kg/s (1650 tph)
Crusher charge when full 13 tonnes
Energy absorbed by mantle 9.2%
Energy absorbed by concave 10.3%
Discharge speed 6 m/s

surfaces precesses around the circumference of the crusher at con-

stant speed. The size of this minimum gap is called the closed side
setting (CSS). The gap on the opposite side of the mantle is always
at a maximum and this distance is called the open side setting
(OSS). The CAD model for the gyratory crusher modelled here is
shown in Fig. 8.
Granular material is loaded into the crusher from above and
travels downward between the two cones. Such a crusher can be
operated in either batch mode with periodic high flow rate feed
added or choke feed with a steady feed from above creating a con-
stant load on top of the crusher. Gyratory crushers are typically
operated in batch mode, most typically fed by dump trucks trans-
porting large rocks directly from a mine. At any given time the
mantle is moving radially outward in a specific radial direction.
The gap to the concave decreases on that half side of the crusher
and the particles in this space are compressed. If the stress at the
contact and therefore the contact force on any of the particles is
sufficiently high then they fracture and the resulting progeny rear-
range into the surrounding pore spaces. On the other half of the
crusher the gap is increasing, creating space for material above
to fall down and form a thicker and more compact bed. The peri-
odic approaching motion causes particles to be progressively
crushed in a series of steps until they are small enough to fall
out through the gap between the two cones at the bottom of the
crusher. Since some part of the mantle is always close to the outer
cone at some location, crushing and discharge are continuous pro-
cesses. The concave and mantle shape are designed to operate with
coarse material and to generally only fracture any given particles
once or twice before being discharged. The discharge gap (at the
closed-side) is controlled by raising or lowering of the mantle.
Fig. 8 shows the gyratory crusher that was simulated using
DEM. The key geometric and operating information for this crusher
are given in Table 3. The most important attributes are the physical
size with a height of nearly 4 m, a rotation rate for the mantle of
500 rpm and a CSS of 150 mm. For this DEM model, the spring con-
stant used was k = 4  108 N/m. This is higher than for the other
crushers simulated in this paper because the forces generated in
the gyratory crusher are higher and the requirement to have phys-
ically acceptable maximum particle overlaps with the crusher
geometry necessitates the use of a higher value. The rock breakage
threshold force for fracture ranged from 5  106 N for the top size
to 109 N for the bottom size. The minimum breakage fragment
resolved in the simulation was 10 mm. This model contains only
around 680 particles with a weight of 8600 kg because of the very
large feed size that is used for this type of crusher. The Cundall
number was 45,000.
Fig. 9 shows the particle flow in the gyratory crusher operating
in a batch arrangement after a full load is dumped into the crusher.
Fig. 9a shows an oblique view from above with a section of the
crusher not shown so that the particles inside can be seen. The ori-
ginal coarse feed is shown as the red, orange and yellow particles
Fig. 9. Particle flow through a gyratory crusher. The particles are coloured by size the top. There has already been significant breakage with particles
with red being large feed and green being the smallest feed. The finer breakage below the top becoming increasingly fine with depth down into
product particles are dark blue. Frames (a) and (b) are from the same view and at
slightly different times showing the downward flow as the gap opens, and (c) shows
the crusher. The particles are light to mid-blue near the bottom
the view from below. (For interpretation of the references to colour in this figure where the mantle and concave most closely approach each other.
legend, the reader is referred to the web version of this article.) Fig. 9b shows the particle arrangement a little later. There has been

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
10 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

Fig. 10. Resident particle size distribution in a batch operation of a gyratory crusher (left) early in the operation when a significant amount of feed material is still present and
(b) later in the operation when there is a significant increase in the larger product fragments and significant reductions in the feed classes.

Concave

Cone

Fig. 11. A cone crusher. The concave (outer cone) is transparent showing the
moving mantle (inner cone) with the variable aperture gap in which the particles
are crushed.

Table 5
Key geometric and operating details for the cone crusher.

Concave height 412 mm

Concave internal diameter (maximum – top) 597 mm
Concave internal diameter (bottom of crushing region) 434 mm
Mantle diameter (maximum – crushing region) 400 mm
Mantle height 440 mm
Mantle side angle (from horizontal) 70 deg Fig. 12. Flow of particles through a choke fed cone crusher at two times with
Feed opening (top of crushing region) 58 mm breakage occurring in the lower half as particles are cyclically squeezed between
Feed size (suggested) 40 mm the mantle and the concave. The particles are coloured by size with red being large
Discharge opening [closed side setting] 12 mm feed and green being the smallest feed. The finer breakage product particles are
Eccentric throw 28 mm light blue. (For interpretation of the references to colour in this figure legend, the
Inclination of mantle for nutating motion (from vertical) 1 deg reader is referred to the web version of this article.)
Angular speed of mantle 600 rpm
Period of crushing Continuous
Product top size 12 mm
and intermediate (light blue) product produced by the cyclic com-
pression can be seen falling from the bottom of the crusher. This
downward flow is strongest on the side opposite the closest
rearrangement of the feed at the top due to the motion of the man- approach of the mantle and concave – which is located on the
tle and from the rebound forces generated by particles being upper left side of the bottom of the crusher at this time because
crushed below. The closest approach of mantle and concave is this is the area where the crusher gap is largest.
located on the front left side of the crusher at this time. Fig. 9c Key output predictions for this gyratory crusher are given in
shows the crusher from below. The closest approach is on the Table 4. The average power draw over the batch cycle is 630 kW
lower right side of the crusher at this time. The finer (dark blue) which is quite reasonable for this type of machine and a generic

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 11

rock feed. The resulting throughput is predicted to be 1650 tph. Table 6

The concave absorbs 10.3% of the input energy (leading to wear) Key simulation predictions for the cone crusher.

while the mantle absorbs slightly less at 9.2%. This crusher has a Power draw 9.5 kW
higher throughput that the previous jaw crusher but is able to take Throughput 3.2 kg/s (11.5 tph)
a significantly wider feed size distribution (with much more fine Crusher charge when full 32.5 kg
Energy absorbed by mantle 18.5%
material). Fig. 10 shows an example of one of the quantitative pre- Energy absorbed by concave 13.5%
dictions DEM with breakage can make. Fig. 10a shows the size dis- Discharge speed 4 m/s
tribution of particles within the crusher shortly after loading is
completed. This is a mix of feed and some early breakage products.
In this specific case, the most common particle size is around
150 mm. There are only small numbers of larger particles and also that no particles larger than the CSS should be able to pass
decreasing amounts of finer material. Later in the operation through the crusher without experiencing at least one breakage
(t = 8 s), there has been a significant decrease in the amount of event. The cone crusher environment is much denser and the tran-
intermediate size particles (100–200 mm particles which have sit times are much slower compared to the gyratory crusher. These
been broken) and some decrease in the larger particles (as these contribute to increasing the opportunity to break particles and to
larger feed particles are also crushed) and a large increase in the increase the product fineness. Cone crushers are usually operated
fractions of finer material (which represents product particles that in choked mode with a continuous feed and continuous product
have yet to exit the machine). Detailed validation of the predicted generation.
size distribution by comparison to specific test data will be per- Fig. 11 shows the CAD model of the cone crusher used in these
formed in future work. Here this intent is to demonstrate that DEM simulations. The step in mantle where the mantle approaches
DEM with breakage is able to model a gyratory crusher and to the concave is the starting point for where breakage occurs. The
show the types of predictions that are now possible. long narrow breakage zone below is clearly visible in this cross-
section view. The wide gap on one side (at this time on the right)
and the matching narrow gap on the other side (the left side at this
7. Cone crusher
time) are also observable. The increase and decrease of the gaps
drive the cyclic compression and result from the rotation of the
Cone crushers are mainly used for secondary and tertiary crush-
axis of the mantle around the circumference of the crusher in the
ing and are similar in structure and operation to gyratory crushers
same way as occurred for the gyratory crusher. The key geometric
but use a flatter crushing head with different profiles for the man-
and operating information for this cone crusher are given in
tle and concave. They are also smaller in size. Unlike the larger
Table 5. This crusher is around 0.6 m high, rotates at 600 rpm,
gyratory crushers, the walls of the concave in the crushing region
and has a CSS of 12 mm (which is significantly finer than the
are inclined at an angle comparable to the edge of the mantle. This
150 mm for the gyratory crusher). The rock breakage threshold
means that the vertical cross-sectional gap between these surfaces
force for fracture ranged from 105 N for the top size to 109 N for
narrows quite slowly. This is done in order to maximise the num-
the bottom size. The minimum breakage fragment resolved in
ber of multiple contacts between mantle and feed material gener-
the simulation was 2.5 mm. The breakage force for the top size is
ating more fracture events and leading to much finer crushing of
lower than for the gyratory crusher because smaller particles have
the material. In the early parts of this gap, inter-particle breakage
a lower force requirement for breakage, decreasing in proportion to
is common resulting in a high degree of fines. Lower down the nar-
their mass (but possibly moderated by the flaw density also chang-
rower gap ensures that smaller particles are able to be trapped
ing with size). This model contains around 15,600 particles with a
between the two working surfaces and also fractured. This ensures
weight of 520 kg. The Cundall number was 720,000.
Fig. 12 shows the particle flow in the cone crusher from front
view at two times. The crusher is sectioned with the front half
not shown so that the particles can be seen. The feed material (col-
oured from green to yellow to orange to red) fills the crusher to
well above the top of the mantle. Feed is continually added at a
rate that matches the discharge rate so that the fill level (and pres-
sure head) remains constant. This arrangement is termed choke fed
and ensures that the crusher is operating in a continuous steady
state mode. As the particles enter the crushing zone they become
finer and are coloured blue to dark blue reflecting the rapid break-
age of the feed particles and their conversion into finer progeny. A
dilute distribution of finer product can be seen over the surface of
the mantle as this material flows slowly downward slowed by fric-
tional contact with the mantle and concave.
Fig. 13 shows a close-up of the breakage region on the left as the
gap between the concave and mantle is opening up. Particles are
densely packed into the upper half of the breakage region where
the converging slopes of the surfaces lead to progressive breakage
of the particles. The lower half of the breakage zone (from halfway
down the visible part of the mantle to the bottom) has a very small
slope difference between the working surfaces and finer product
particles can be seen flowing down through the gap.
Fig. 13. Close-up of flow and breakage in the crushing region of the cone crusher. Key output predictions for this crusher are given in Table 6. The
The particles are coloured by size with red being large feed and green being the
smallest feed. The finer breakage product particles are blue. (For interpretation of
power draw for this quite small cone crusher is predicted to be
the references to colour in this figure legend, the reader is referred to the web 9.5 kW and the throughput is about 11.5 tph. The energy absorbed
version of this article.) by the mantle is 18.5% of the energy input while the concave

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
12 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

Fig. 14. (a) Resident particle size in the cone crusher shortly after operation commences and (b) cumulative normal energy absorption of particles measures the way in which
compressive energy is added to the particles.

Fig. 14a shows an example of the size distribution of the parti-

cles resident within the crusher. It is bi-modal reflecting the feed
distributions from 20–50 mm and the product size distribution
(5–20 mm). DEM simulation can provide information on the size
distribution within the crusher as well as the product size distribu-
tion exiting the bottom of the crusher and how these vary with
time and operating condition or material property changes.
Fig. 14b shows the distribution of the cumulative normal energy
absorbed by the particles – which can be considered as a measure

Table 7
Key geometric and operating details for the single roll crusher.

Diameter of roll 800 mm

Length of roll 500 mm
Diameter of roll teeth 80 mm
Height of roll teeth above roll surface 50 mm
Number of roll teeth 4 (axial)  20 (angular)
Feed size (maximum recommended) 250 mm
Distance from end of roll teeth to crushing plate 100 mm
surface
Fig. 15. Abrasive wear on the mantle of a cone crusher with a curved surface over a Inclination of crusher plate 10 deg
period of 7 s. The wear is very uneven depending on the specific particles broken at Width of crusher plate 500 mm
each location. Diameter of crusher plate ribs 166 mm
Height of crusher plate ribs 30 mm above crusher
plate
Spacing between crusher plate ribs (from centre to 500 mm
centre)
absorbs around 13.5%. These are noticeably higher than for the
Angular speed of roll 100 rpm
gyratory crusher and reflect the much larger number of breakage Period of crushing Continuous
events in the breakage zone which therefore create larger energy Product top size 100 mm
absorption by the crusher working surfaces.

Roll

Conveyor

Fig. 16. A single roll crusher showing the toothed single roller and the inclined crusher conveyor plates.

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 13

around 0.15 J and a range from 105 J (which has a negligible effect
on these particles) up to a maximum of 900 J (which is more than
enough energy to obliterate such particles).
Fig. 15 shows an example of the prediction of abrasive wear on
the mantle of the cone crusher over 7 s of operation. The wear is
low on the regions above the breakage zone (dark blue), is moder-
ate at the top of the outward slope of the mantle and increases
with depth into the crusher. It is important to understand that
the wear increment from processing a specific set of feed is quite
spatially variable as the details of which particles flowed through
and broke in different parts of the crusher leads to quite large cir-
cumferential variations. Long term wear can also be predicted by
running the model for longer which will average out these circum-
ferential variations. Depending on the need, one may want to cal-
culate short term wear increments or long term average wear. Only
the former is shown for this example.

8. Impact roll crusher

Roll crushers are generally of two varieties – single-roll and

double-roll. They are mostly employed in intermediate or final size
reduction stages and for softer materials such as coals and
limestone.

Fig. 17. Particle flow along an upward inclined conveyor to a roll crusher. Fine
material is pulled through the gap under the studded roll and departs on the
conveyor. Larger particles are broken as they are trapped between the roll and
conveyor plates on approach. The particles are coloured by size with red being large
feed and green being intermediate and light blue being the smallest feed. The finer
breakage product particles are dark blue. (a) Oblique view of the full crusher, and
(b) side view of the breakage region in front of and directly below the roll. (For
interpretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)

Table 8
Key simulation predictions for the single roll crusher.

Throughput 240 kg/s (864 tph)

Power draw 125 kW
Energy absorbed by conveyor 8.3%
Energy absorbed by roll 10.1%
Energy absorbed by the walls 5.1%

of impact damage of the particles. For the design and optimisation

of such crushers, it is important to understand how energy is sup-
plied to the particles and what effect this has on the particles. The
energy absorbed by each particle in transiting a crusher depends
on its specific path and contact histories. The distribution of these
cumulative absorbed energies over all the particles that have
passed through the crusher is characteristic of the overall nature
Fig. 18. (a) Impact damage on the roll is concentrated on the leading faces and tops
of its crushing action on the feed material. For this crusher the of the studs on the roll and (b) abrasion wear is concentrated on the ridges of the
energy absorption is close to normally distributed with a peak of conveyor with moderate abrasion on the intervening plates.

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
14 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

Fig. 19. (a) Energy spectra for collisions of rocks with the roll and conveyor and (b) cumulative energy absorption by the particles in crusher.

Housing Table 9
Key geometric and operating details for the double roll crusher.

Diameter of each roll 450 mm

Length of each roll 500 mm
Height of primary (P) roll teeth above roll surface 75 mm
Width of primary roll teeth 70 mm
Height of secondary (S) roll teeth above roll surface 75 mm
Width of secondary roll teeth 50 mm
Height of tertiary (T) roll teeth above roll surface 35 mm
Width of tertiary roll teeth 30 mm
Number of roll teeth 9 (axial)  12
(angular)
Pattern of roll teeth in alternating rows T P T/S T S
Minimum distance between the teeth of the two 80 mm
rolls
Feed size 300 mm
Angular speed of roll 60 rpm
Period of crushing Continuous
Product top size 150 mm

Roll
minimum breakage fragment resolved in the simulation was
5 mm. This model contains around 1700 particles with a weight
Roll
of 270 kg. The Cundall number was 510,000.
Fig. 17 shows the particle flow and breakage in this crusher at
Fig. 20. A double roll crusher views obliquely from above. The feed hopper and
crusher shell are shown in orange while the two rolls with alternating large and
one particular time after steady state operation has been achieved
small downward directed teeth are shaded grey. The front wall of the crusher shell for a feed rate of 240 kg/s. The rocks are conveyed up to the stud-
is omitted to allow the rolls to be seen clearly. (For interpretation of the references ded single roller by the inclined crusher plate (Fig. 17a). The ridges
to colour in this figure legend, the reader is referred to the web version of this both protect the gaps between the articulated plates in the con-
article.)
veyor and act to restrict particles flowing backwards. As particles
approach the roll the lower ones are forced between the roll and
the conveyor plates. The upper particles in the feed stream are
8.1. Single roll impact crusher forced upwards and form a pile in front of the roll. The size of this
pile is controlled by the balance of the weight of the pile and the
The CAD model for a single-roll crusher is shown from two pressure in front of the roll that tries to push particles upwards.
views in Fig. 16. It consists of a toothed roller that compresses The particles moving under the roll (shown in close-up in
material travelling below against an inclined moving ridged crush- Fig. 17b) are crushed by high compressive forces as they are
ing plate. Rocks are conveyed up the incline to the point where the trapped between the studs on the roll and the conveyor plate
teeth of the roller are able to bite into the rocks. Feed material is and its ridges. Finer particles (dark blue) are produced and flow
then crushed by a combination of compression, impact and shear. out from under the roll with the smaller feed (light blue) that
The crushed product is discharged over the incline onto a separate has escaped breakage. They are transported to the right by the con-
conveyor. Single-roll crushers are mainly used for breaking soft veyor and exit the crusher (on the right).
materials such as limestone. The key geometric and operating Key output predictions for this crusher are given in Table 8. The
information for this crusher are given in Table 7. This roll is power draw for the 240 kg/s flow rate is 125 kW. Of this 8.3% is
0.8 m in diameter and 0.5 m wide and has 4 rows of studs in lines consumed by wear on the conveyor, 10.1% by the roll itself and
parallel to the sides of the roll. The maximum feed size is 250 mm 5.1% by the front and back walls on either side of the conveyor
(larger particles cannot be gripped by the roll and consequently and roll. So around 1=4 of the energy input by this crusher is used
cannot be drawn into the crushing gap). The minimum gap in producing wear of the working surfaces. Short term wear predic-
between plate and roll is 100 mm and controls the product top tions for the roll and the conveyor plate are shown in Fig. 18.
size. The rock breakage threshold force for fracture ranged from Unsurprisingly, the highest wear occurs on the upper half of the
2  105 N for the top size to 109 N for the bottom size. The front faces of the studs on the rolls and on the leading half of the

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 15

Fig. 21. Flow of particles and breakage in the double roll crusher showing an oblique view of the two studded rolls beneath a feed bin which is maintained at a constant level.
The frames are at closely spaced times and show the fracture of two particles between and above the teeth. The particles are coloured by size with red being large, green being
intermediate and blue being fine. The finer breakage product particles are dark blue. (For interpretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)

top surfaces of the studs. This will generate strong rounding wear 8.2. Double roll crusher
on the leading edges of the studs due to these being the preferen-
tial contact point of the rolls with the particles and the predomi- A double roll crusher consists of a pair of oppositely rotating
nant location where very high compressive forces are generated. toothed rolls. Particles are drawn down from a hopper above and
There is essentially no wear on the other more protected parts of are compressed when they are gripped between teeth on the rolls
the studs or on the roll surface below. The studs are extremely on each side. Fig. 20 shows such a crusher. The rolls have teeth of
effective at protecting the roll surface. The highest wear on the two sizes. The larger ones have steep moderately curved leading
conveyor is on the ridges between plates. These are the points at faces and a relatively sharp tip at the outer end. Their trailing faces
which the rocks are trapped and prevented from sliding and which are highly curved. This shape enables the opposite pairs to grip
hold the particles so that they can be compressed by the studs. The quite large particles at locations above the two rolls and drag them
liner plates in between have intermediate levels of abrasive dam- down between the rolls where they are compressed and fractured.
age due to the sliding of the rock particles from different locations The large teeth are arranged to be offset from each other in the
between ridges back towards the trailing ridge where they are axial direction. This means that the capture and fracture of the par-
trapped. ticles is a fully three dimensional process. In between the large
Fig. 19 shows examples of some of the quantitative information teeth are small teeth which appear to be present to improve the
available from the DEM simulation to understand where the transport of finer material through the crusher and to protect the
energy supplied is used by the system. Fig. 19a shows the energy roll surface between the large teeth. This type of crusher is often
spectra for collisions of all rocks in the crusher. The shear energy used for sizing a granular material; that is, ensuring that all the
is each collision is modestly larger than the normal energy, indicat- particles are less than a required maximum size. For this double
ing that shear plays an important role in the breakage under the roll crusher this maximum is controlled by the diagonal distance
roll. The most common collision energy is 0.01 J which is too low between the large teeth of opposite rolls. Particles smaller than this
to create impact breakage indicating that breakage is indeed com- dimension can, in principle, fit between the teeth and then pass
pressive in nature. The collision energies decline in frequency with through the crusher. In practice, many of these near maximum par-
both increasing and decreasing energy levels. Fig. 19b shows the ticles are not so well aligned in their trajectories and are captured
distribution of energy absorbed by the particles that are currently and crushed by the roll teeth. This type of crusher is again typically
in the system. It is somewhat noisy due to the relatively small used for softer materials such as coal and limestone.
numbers of particles within this crusher at any time, but is argu- The key geometric and operating information for this crusher
ably a normal distribution with a flattened peak. This provides a are given in Table 9. The rolls are 0.5 m long and 0.45 m in diame-
measure of the total energy absorbed by each particle during its ter. The rotation rate of the rolls is relatively slow at 60 rpm. The
transit through the machine. feed size range is from 100 mm to 300 mm. The product top size

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
16 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

Fig. 23. Resident particle size distribution of the double roll crusher at two different
times in its steady state operation. The specific particles present depend on the
particles fed in, their size and strength and the degree to which they have broken.
There is significant variation in the composition of the crusher content even though
the crusher is in steady state operation.

Table 10
Key simulation predictions for a double roll crusher.

Power draw 8.4 kW

Throughput 180 (648) kg/s (tph)
Energy absorbed by housing 0.70%
Energy absorbed by the rolls 0.61%

particles exceed their breakage force threshold, the particles break.

In this case, one particle breaks shortly after the time in Fig. 21a.
Fig. 22. Particle flow in the double roll crusher at three closely spaced times, This is the red particle near the left roll. In Fig. 21b, this has been
viewed along the axis of the rolls, showing fracture of several large particles replaced by an array of light and dark blue progeny around a cen-
between and above the teeth. The particles are coloured by size with red being tral green–yellow coloured particle. The finer (blue) material from
large, green being intermediate and blue being fine. The finer breakage product
particles are dark blue. (For interpretation of the references to colour in this figure
previous breakage events has now passed between the rolls and is
legend, the reader is referred to the web version of this article.) now discharging from the crusher. The lower of the intermediate
size green particles has been trapped between the oppositely mov-
ing teeth on the rolls and is being drawn down into the center of
for this tooth configuration is 150 mm. The rock breakage thresh- the crusher. In Fig. 21c, this particle has almost reached the center-
old force for fracture ranged from 4  105 N for the top size to line of the rolls. The fragments from the earlier fracture of the red
109 N for the bottom size. The minimum breakage fragment particle in Fig. 21a have now moved apart and the finer particles
resolved in the simulation was 5 mm. This model contains around have started falling between the rolls. In Fig. 21d, the green particle
3000 particles with a weight of 725 kg. The Cundall number was directly between the rolls (in Fig. 21c) has now finally fractured
1.4  106. due to the very high compressive force as the teeth move through
Fig. 21 shows the particle flow and breakage in the double roll the horizontal plane through the roll axes (at which the maximum
crusher. The feed material is coarse and coloured from red to yel- compression is generated). This particle was only just slightly too
low to green. There are only about a dozen feed particles above large to survive the transit. If it had been slightly stronger or
the rolls in Fig. 21a. The rotation of the rolls loads these particles, slightly smaller, then it would have survived.
pressing them against each other, against the rolls and against Fig. 22 shows the flow and breakage in the crusher from a view
the lower walls of the feed hopper. When the forces on the along the axes of the rolls. In Fig. 22a, a green particle can be seen

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 17

damage. The two wear measures are very similar in magnitude

and distribution indicating that both impact and abrasion damage
are likely to be important for this crusher. The wear is highly con-
centrated on the leading faces, leading edge and sides of the larger
roll teeth. Wear on the small roll teeth is relatively weak. Wear on
the roll surfaces between the teeth is very small with no variation
from dark blue. There is some wear also visible on the lower sec-
tions of the feed chute with the abrasion component being quite
a lot stronger than the impact damage. This reflects the sliding of
the feed material on the inwardly inclined surfaces.
Fig. 25 shows various energy spectra for the double roll crusher.
The spectrum for all collisions has a most common collision energy
level of 0.01 J. This is quite similar to the single roll crusher.
However the fall off at lower energies is faster whilst the falloff
for higher energies is slower with the maximum collision energies
(in excess of 1 kJ) being significantly higher than for the single roll
(at 100 J). This reflects the substantial kinetic energy of the larger
particles as they fall through the hopper into the bed on top of
the rolls. This potentially means that some particle breakage in this
crusher is also from impact. The spectra for class 1 (the finest
material) is shown in Fig. 25b and is bi-modal with peaks at around
1 mJ and around 10 J whilst the spectra for class 10 (the coarsest
feed) in Fig. 25c is uni-model with a peak collision energy of
around 0.1 J (which is much larger than that found for all particles).
The energy spectrum for the hopper and side walls of the crusher
(object 1) is broad and peaked at around 1 mJ. The spectrum for
the right roll (Fig. 25e) is similar to that of the coarse particles,
which is not surprising as many of the higher energy contacts on
the roll arise from collisions with the larger size particles.

9. Discussion

In all the crusher models considered here, the motion of crusher

surfaces is kinematically specified and all other degrees of freedom
are held constant. In practice, small movements of machine com-
ponents in response to the very high forces generated are possible
and are sometimes common. An example would be the location of
Fig. 24. Wear damage of the double roll crusher due to: (a) impact and (b) abrasion. the mantle in a cone crusher. Typically its vertical position is con-
The roll teeth experience greatest degree of damage with abrasion and impact trolled by hydraulics and this is used to set the closed side setting
damage levels being comparable.
and therefore the top size of the product size distribution. Given
that the hydraulics have finite limits on the pneumatic pressures
they can generate and the machine stiffness is finite (albeit large),
trapped between teeth of both rolls. This particle is then broken there is always deformation of the structure and vertical move-
(Fig. 22b). A large red particle on the right is falling towards the ment of the mantle in response to very high forces within the
right roll in the first two frames. This breaks upon contact with crusher. So if a particular rock configuration leads to a spike in
the teeth of the roll on the right creating a large collection of fine force, the CSS of the machine will temporarily increase as the man-
and intermediate sized progeny. The flow of product down from tle is forced down. This means that the mantle dynamics can affect
between the rolls is clearly visible. Fig. 23 shows predictions for the particle breakage and particularly can act to limit damage to
the resident size distribution within the crusher at two times dur- the crusher liner from short lived but extreme force conditions.
ing its continuous steady state operation. The specific particles In the current DEM models, the machines are infinitely strong
present depend on the particles fed in, their size and strength (rigid) and the finite pressures of the pneumatics have not been
and the degree to which they have broken. There is significant var- considered. The inclusion of machine dynamics is straight forward
iation in the composition of the crusher content even though the in DEM with the only change being the need to integrate the equa-
crusher is in steady state operation. Key output predictions for this tion of motion for the degrees of freedom that are of interest based
crusher are given in Table 10. The power draw for the 180 kg/s on the instantaneous force applied by the particles. This has previ-
throughput is 8.4 kW. Of this 0.7% is consumed by wear on the ously been done to model fully dynamic structures such as dragline
housing and 0.61% by the two rolls. So only around 1.3% of the buckets in excavation (Cleary, 1998b) and partially dynamic
energy input by this crusher results in wear of the working sur- objects such as the rolls in a Grinding Table or Vertical Roller mill
faces. This is an order of magnitude less than the wear predicted (Cleary et al., 2008).
for a single roll crusher. The configuration of teeth on the two rolls The same issues explicitly apply to modelling High Pressure
closely controls the motion of the feed rock into the breakage zone Grinding Rolls (HPGR) where the particle forces and breakage
so that there is less relative motion of the rock with the roll sur- within the machine balance the pneumatic force on the moving
faces which could lead to abrasive wear. roll to dynamically control the gap size and therefore the through-
Fig. 24 shows the normal and shear power for this crusher after put and product size distribution. Fig. 26 shows a simple example
10 s of operation. These are measures of the impact and abrasion of DEM modelling of a HPGR in two dimensions. The feed particles

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
18 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

Fig. 25. Energy spectra for collisions of rocks with the double rolls and walls, (a) all collisions, (b) smallest rock size, (c) largest rock size, (d) hopper and crusher side walls,
and (e) one of the rolls.

are coloured by size with red being the coarsest and green being they are sufficient to enable the particles to break in generally
the feed. These flow down from a feed hopper above. The right roll the right locations and therefore enable predictions of all the other
is fixed while the left roll is free to move sideways in response to key quantities needed to help understand compression based
the balance of applied force from the hydraulics and contact forces crushers. Detailed validation of breakage product size will be
from the particle bed. Particles break under compressive load in examined in future work using a more data driven version of this
the converging region above the minimum gap between the rolls replacement breakage model. This model uses measured progeny
creating a stream of fine product particles (dark blue) that fall from size distributions for non-round particles which are then packed
between the rolls. The gap size is then controlled by the balance into the volume of the non-round parent particle. The method for
between the weight of particles in the hopper, the applied roll generating the progeny and performing the packing process is
pressure and the force generated in the particle bed. Breakage of described in Delaney et al. (2010). This next generation model will
the particles limits the force transmitted to the rolls. Breakage address the issues that arise from approximating the particles as
occurs both above the rolls due to compression and shear from spherical and with using real material dependent data to control
the roll rotation and between the rolls where the bed pressure is the breakage product generation. Our expectation is that this
maximal. model will offer the prospect of quantitatively accurate prediction
Since the breakage model used in this paper is based on spher- of the product and will be suitable for detailed validation against
ical particles with a geometric (but fractal like) progeny distribu- measured crusher data.
tion for each breakage event, it is unlikely that the predicted Particle breakage is accompanied by the release of elastic
breakage size distributions will be quantitatively accurate, but energy that has been stored in contacts between particles and in

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx 19

such a DEM model is able to simulate the crushing process and

can predict key outputs such as power draw, product size, through-
put rate and equipment wear. The model also provides the ability
to determine quantitative information such as the resident load of
the crusher (mass and size distribution), exit conditions such as
discharge velocity and how the energy supplied is being used
(either for particle breakage or equipment wear) which are nearly
impossible to obtain from physical experiments. The availability of
this computationally derived information provides greater oppor-
tunity to understand the machine level operation of crushers, the
contribution of multi-body contacts and to demonstrate that such
crusher modelling is now possible.
Specific observations from the models for the different types of
crushers considered include:

 The jaw crusher operates in a relatively dilute mode so that

Fig. 26. Demonstration of DEM modelling of a HPGR. The particles are coloured by product particles are relatively free to quickly discharge from
size with red being coarse and green being the finest feed from the hopper above. the crusher. Fracture appears to be dominantly caused by parti-
The breakage product is coloured light blue through to dark blue for the finest
cles being loaded by contact with both jaws and there is little
progeny. The left roll is dynamically controlled and can move laterally. (For
interpretation of the references to colour in this figure legend, the reader is referred indication of multi-particle interactions influencing the fracture
to the web version of this article.) process.
 Gyratory crushers fracture particles only a small number of
times leading to a broad product size distribution below the
contacts between particles and parts of the crusher. In the DEM CSS dimension. The broader feed distribution with much more
model, the elastic energy in contacts is determined by the overlap fine material means that power draw is much higher relative
Dx of the particles (which is also used in Eq. (1) to give the normal to the feed rate than for a jaw crusher.
contact force). By densely packing the daughter particles into the  The shallower approach of the mantle and the concave in the
volume of the parent, any overlaps between breaking particles crushing region promotes many more breakage events, includ-
and either other particles or boundary elements are preserved ing multi-particle loading through stress chains between the
and the elastic energy is then also preserved for the resolved prog- working surfaces. The relatively longer and shallower breakage
eny particles. The daughter particles usually have reasonable zone leads to relatively much finer product than for the gyra-
mobility because they are sufficiently small to be able to move into tory crusher. Wear increases with depth into this crusher.
the voids between the surrounding larger particles. This enables  In roll crushers, the teeth or studs on the rolls significantly pro-
the strongly elastically loaded and now small fragments that over- tect the rolls from wear. Particles break predominantly when
lap the crusher boundary elements to rapidly and often explosively loaded by the opposing working surfaces of the crusher rather
unload as this elastic energy is converted to kinetic energy. Due to than via multi-particle contacts.
the small masses of the fragments this leads to often very high
velocities for these particles. In dense regions, this ‘‘explosive’’
unloading of the boundary contacts is quickly absorbed by sur- References
rounding particles. But in less dense regions this can lead to obser-
vable motion in the crusher and sometimes ejection of high speed Barker, G.C., 1994. Computer simulations of granular materials. In: Mehta, Anita
fragments from the crusher. For example in the gyratory crusher (Ed.), Granular Matter: An Interdisciplinary Approach. Springer-Verlag, NY.
Campbell, C.S., 1990. Rapid granular flows. Annu. Rev. Fluid Mech. 22, 57–92.
and the cone crusher when it is not full the upward ejection of par- Cleary, P.W., 1998a. Predicting charge motion, power draw, segregation, wear and
ticles can visibly agitate the surface of the bed of particles inside. particle breakage in ball mills using discrete element methods. Miner. Eng. 11,
The most extreme case is a Horizontal Shaft Impactor (HIS) that 1061–1080.
Cleary, P.W., 1998b. The filling of dragline buckets. Math. Eng. Ind. 7, 1–24.
regularly ejects particles out of the feed opening requiring protec-
Cleary, P.W., 2001a. Modelling comminution devices using DEM. Int. J. Numer. Anal.
tive screens of hanging chains to be deployed. These are all real Meth. Geomech. 25, 83–105.
effects and these are correctly observed in the crusher models Cleary, P.W., 2001b. Charge behaviour and power consumption in ball mills:
shown here. The reason that these effects are captured is the sensitivity to mill operating conditions, liner geometry and charge composition.
Int. J. Miner. Process. 63, 79–114.
preservation of the elastic energy at contacts when breakage Cleary, P.W., 2001c. Recent advances in DEM modelling of tumbling mills. Miner.
occurs – which is a very positive attribute of this type of replace- Eng. 14, 1295–1319.
ment breakage model in DEM. Cleary, P.W., 2004. Large scale industrial DEM modelling. Eng. Comput. 21, 169–
204.
Cleary, P.W., 2008. The effect of particle shape on simple shear flows. Powder
10. Conclusions Technol. 179, 144–163.
Cleary, P.W., 2009. Industrial particle flow modelling using DEM. Eng. Comput. 26,
698–743.
The DEM method, including a particle replacement breakage Cleary, P.W., Morrison, R.D., 2012. Prediction of 3D slurry flow within the grinding
model, has been successfully demonstrated using a broad range chamber and discharge from a pilot scale SAG mill. Miner. Eng. 39, 184–195.
Cleary, P.W., Sawley, M.L., 2002. DEM modelling of industrial granular flows: 3D
of existing compression based crushers. Predictions for five types case studies and the effect of particle shape on hopper discharge. Appl. Math.
of crushers: jaw crusher, cone crusher, gyratory crusher, impact Modell. 26, 89–111.
crusher and double roll crusher were presented. The breakage Cleary, P.W., Metcalfe, G., Liffman, K., 1998. How well do discrete element granular
flow models capture the essentials of mixing and segregation processes? Appl.
method used is based on facture of particles using geometric rules
Math. Model. 22, 995–1008.
for progeny generation and estimates of the forces applied to the Cleary, P.W., Morrison, R., Morrell, S., 2003. Comparison of DEM and experiment for
particles as they flow through the crusher. Unlike for most particle a scale model SAG mill. Int. J. Miner. Process. 68, 129–165.
processing equipment, DEM cannot be used for modelling crushers Cleary, P.W., Sinnott, M.D., Morrison, R.D., 2008. DEM prediction of particle flows in
grinding processes. Int. J. Numer. Methods Fluids 58, 319–353.
without a breakage model because the particles will be too big to Das, R., Cleary, P.W., 2010. Effect of rock shapes on brittle fracture using smoothed
transit through the equipment. The examples given, show that particle hydrodynamics. J. Theor. Appl. Fract. Mech. 53, 47–60.

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021
20 P.W. Cleary, M.D. Sinnott / Minerals Engineering xxx (2014) xxx–xxx

Delaney, G., Cleary, P.W., Sinnott, M.D., Morrison, R.D., 2010. Novel application of Morrison, R.D., Cleary, P.W., 2008. Towards a virtual comminution machine. Miner.
DEM to modelling comminution processes. In: Proc. 9th World Congress on Eng. 21, 770–781.
Computational Mechanics, 19th July, Sydney, IOP Conf. Series: Materials Science Plochberger, D.I.T., Avila, M.B., 2014. Development of an energy optimized stirred
and Engineering, vol. 10, p. 012099. media grinding mill. BHM Berg-und Hüttenmännische Monatshefte 159, 253–
Delaney, G.W., Cleary, P.W., Morrison, R.D., Cummins, S., Loveday, B., 2013. 258.
Predicting breakage and the evolution of rock size and shape distributions in Potapov, A.V., Campbell, C.S., 1994. Computer simulation of impact-induced particle
AG and SAG mills using DEM. Miner. Eng. 50, 132–139. breakage. Powder Technol. 81, 207–216.
Djordjevic, N., 2003. Discrete element modelling of the influence of lifters on power Potapov, A.V., Campbell, C.S., 1996. A three dimensional simulation of brittle solid
draw of tumbling mills. Miner. Eng. 16, 331–336. fracture. Int. J. Mod. Phys. C7, 717–729.
Djordjevic, N., 2005. Influence of charge size distribution on net-power draw of Potyondy, D.O., Cundall, P.A., 2004. A bonded particle model for rock. Int. J. Rock
tumbling mill based on DEM modelling. Miner. Eng. 18, 375–378. Mech. Min. Sci. 41, 1329–1364.
Eliáš, J., 2014. Simulation of railway ballast using crushable polyhedral particles. Powell, M.S., Weerasekara, N.S., Cole, S., LaRoche, R.D., Favier, J., 2011. DEM
Powder Technol. 264, 458–465. modelling of liner evolution and its influence on grinding rate in ball mills.
Franke, J., Stewart, M.P., Lichti, D., 2012. System and Method for Measuring and Miner. Eng. 24, 341–351.
Mapping a Surface Relative to a Reference. U.S. Patent Application, 13/707, 335. Quist, J., Evertsson, C.M., 2010. Application of discrete element method for
Herbst, J.A., Nordell, L., 2001. Optimization of the design of sag mill internals using simulating feeding conditions and size reduction in cone crushers. In: XXV
high fidelity simulation. In: Barratt, D.J., Allan, M.J., Mular, A.L. (Eds.), International Mineral Processing Congress (IMPC) 2010 Proceedings/Brisbane,
Proceedings of the SAG Conference, vol. IV. University of British Columbia, QLD, Australia/6–10 September 2010, pp. 3337–3347.
Vancouver, BC, pp. 150–164. Rajamani, R.K., Mishra, B.K., 1996. Dynamics of ball and rock charge in sag mills. In:
Herbst, J.A., Potapov, A.V., 2004. Making a discrete grain breakage model practical Proc. SAG 1996, Department of Mining and Mineral Process Engineering,
for comminution equipment performance simulation. Powder Technol. 143, University of British Columbia.
144–150. Refahi, A., Rezai, B., Aghazadeh Mohandesi, J., 2007. Use of rock mechanical
Hossain, Z., Indraratna, B., Darve, F., Thakur, P.K., 2007. DEM analysis of angular properties to predict the Bond crushing index. Miner. Eng. 20, 662–669.
ballast breakage under cyclic loading. Geomech. Geoeng.: Int. J. 2, 175–181. Refahi, A., Aghazadeh Mohandesi, J., Rezai, B., 2010. Discrete element modeling for
Hulthén, E., 2010. Real-time Optimization of Cone Crushers. Doctoral Thesis, predicting breakage behavior and fracture energy of a single particle in a jaw
Chalmers University of Technology. crusher. Int. J. Miner. Process. 94, 83–91.
Indraratna, B., Thakur, P.K., Vinod, J.S., 2009. Experimental and numerical study of Rosario, P.P., Hall, R.A., Maijer, D.M., 2004. Liner wear and performance
railway ballast behavior under cyclic loading. Int. J. Geomech. 10, 136–144. investigation of primary gyratory crushers. Miner. Eng. 17, 1241–1254.
Jayasundara, C.T., Yang, R.Y., Yu, A.B., Curry, D., 2008. Discrete particle simulation of Schöpfer, M.P., Childs, C., 2013. The orientation and dilatancy of shear bands in a
particle flow in IsaMill – effect of grinding medium properties. Chem. Eng. J. 135 bonded particle model for rock. Int. J. Rock Mech. Min. Sci. 57, 75–88.
(1), 103–112. Schubert, W., Khanal, M., Tomas, J., 2005. Impact crushing of particle–particle
Jayasundara, C.T., Yang, R.Y., Yu, A.B., 2012. Effect of the size of media on grinding compounds – experiment and simulation. Int. J. Miner. Process. 75, 41–52.
performance in stirred mills. Miner. Eng. 33, 66–71. Sinnott, M., Cleary, P.W., Morrison, R., 2006. Analysis of stirred mill performance
Jiang, M.J., Yan, H.B., Zhu, H.H., Utili, S., 2011. Modeling shear behavior and strain using DEM simulation: Part 1 – media motion, energy consumption and
localization in cemented sands by two-dimensional distinct element method collisional environment. Miner. Eng. 19, 1537–1550.
analyses. Comput. Geotech. 38 (1), 14–29. Sinnott, M.D., Cleary, P.W., Morrison, R.D., 2011. Is media shape important for
Jonsén, P., Pålsson, B.I., Stener, J.F., Häggblad, H.Å., 2014. A novel method for grinding performance in stirred mills? Miner. Eng. 24, 138–151.
modelling of interactions between pulp, charge and mill structure in tumbling Thakur, P.K., Vinod, J.S., Indraratna, B., 2010. Effect of particle breakage on cyclic
mills. Miner. Eng. 63, 65–72. densification of ballast: a DEM approach. In: IOP Conference Series: Materials
Kafui, K.D., Thornton, C., 2000. Numerical prediction of impact breakage of spherical Science and Engineering, vol. 10, no. 1. IOP Publishing, p. 012229.
crystalline agglomerate. Powder Technol. 109, 113–132. Thornton, C., 1996. Numerical simulation of the impact fracture and fragmentation
Kalala, J.T., Breetzke, M., Moys, M.H., 2008. Study of the influence of liner wear on of agglomerates. J. Phys. D 29, 424–435.
the load behaviour of an industrial dry tumbling mill using the Discrete Thornton, C., Ciomocos, M.T., Adams, M.J., 1999. Numerical simulations of
Element Method (DEM). Int. J. Miner. Process. 86, 33–39. agglomerate impact breakage. Powder Technol. 105, 74–82.
Khanal, M., Jayasundara, C.T., 2014. Role of particle stiffness and inter-particle Thornton, C., Cummins, S., Cleary, P.W., 2013. An investigation of the comparative
sliding friction in milling of particles. Particuology 16, 54–59. behaviour of alternative contact force models during inelastic collisions.
Legendre, D., Zevenhoven, R., 2014. Assessing the energy efficiency of a jaw crusher. Powder Technol. 233, 30–46.
Energy (in press). Tomas, J., Schreier, M., Groger, T., Ehlers, S., 1999. Impact crushing of concrete for
Li, H., McDowell, G., Lowndes, I., 2014. Discrete element modelling of a rock cone liberation and recycling. Powder Technol. 105, 39–51.
crusher. Powder Technol. 263, 151–158. Walton, O.R., 1994. Numerical simulation of inelastic frictional particle–particle
Lichter, J., Lim, K., Potapov, A., Kaja, D., 2009. New developments in cone crusher interaction. In: Roco, M.C. (Ed.), Particulate Two-phase Flow, pp. 884–911
performance optimization. Miner. Eng. 22 (7), 613–617. (Chapter 25).
Lobo-Guerrero, S., Vallejo, L.E., 2006. Discrete element method analysis of railtrack Wang, M.H., Yang, R.Y., Yu, A.B., 2012. DEM investigation of energy distribution and
ballast degradation during cyclic loading. Granular Matter 8, 195–204. particle breakage in tumbling ball mills. Powder Technol. 223, 83–91.
Lu, M., Mcdowell, G.R., 2010. Discrete element modelling of railway ballast under Weerasekara, N.S., Powell, M.S., Cleary, P.W., Tavares, L.M., Evertsson, M., Morrison,
monotonic and cyclic triaxial loading. Geotechnique 60, 459–467. R.D., Quist, J., Carvalho, R.M., 2013. The contribution of DEM to the science of
Mishra, B.K., Rajamani, R.J., 1992. The discrete element method for the simulation of comminution. Powder Technol. 248, 3–24.
ball mills. Appl. Math. Modell. 16, 598–604. Whittles, D.N., Kingman, S., Lowndes, I., Jackson, K., 2006. Laboratory and numerical
Mishra, B.K., Rajamani, R.K., 1994. Simulation of charge motion in ball mills. Part 1: investigation into the characteristics of rock fragmentation. Miner. Eng. 19,
exp. verifications. Int. J. Miner. Process. 40, 171–186. 1418–1429.
Morrison, R., Cleary, P.W., 2004. Using DEM to model ore breakage within a pilot Yang, R.Y., Jayasundara, C.T., Yu, A.B., Curry, D., 2006. DEM simulation of the flow of
scale sag mill. Miner. Eng. 17, 1117–1124. grinding media in IsaMill. Miner. Eng. 19 (10), 984–994.

Please cite this article in press as: Cleary, P.W., Sinnott, M.D. Simulation of particle flows and breakage in crushers using DEM: Part 1 – Compression
crushers. Miner. Eng. (2014), http://dx.doi.org/10.1016/j.mineng.2014.10.021