See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318080904

Mathematical modeling of a vertical shaft impact crusher using the Whiten

model

Article  in  Minerals Engineering · September 2017

DOI: 10.1016/j.mineng.2017.06.022

CITATIONS READS

2 290

4 authors:

Juliana Segura Salazar Gabriel Barrios

Federal University of Rio de Janeiro Federal University of Rio de Janeiro
5 PUBLICATIONS   2 CITATIONS    14 PUBLICATIONS   84 CITATIONS   

SEE PROFILE SEE PROFILE

Victor Rodriguez Luis Marcelo Tavares

Federal University of Rio de Janeiro
20 PUBLICATIONS   102 CITATIONS   
118 PUBLICATIONS   1,588 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Classification and degradation of green iron ore pellets View project

Mechanistic mill modelling View project

All content following this page was uploaded by Victor Rodriguez on 07 May 2018.

The user has requested enhancement of the downloaded file.

 Minerals Engineering 111 (2017) 222–228

Contents lists available at ScienceDirect

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

Mathematical modeling of a vertical shaft impact crusher using the Whiten MARK
model
⁎
Juliana Segura-Salazar, Gabriel Pantoja Barrios, Victor Rodriguez, Luís Marcelo Tavares
Department of Metallurgical and Materials Engineering, Universidade Federal do Rio de Janeiro – COPPE/UFRJ, Cx. Postal 68505, CEP 21941-972, Rio de Janeiro, RJ,
Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: Vertical Shaft Impact (VSI) crushers have been used as interesting alternatives to cone crushers, particularly in
Vertical Shaft Impact Crusher (VSI) the production of aggregates for the construction industry, not only due to their good energy efficiency but also
Comminution to their ability to generate more isometric and tougher particles, which is highly desirable in cement mortars and
Crushing concrete applications. Several mathematical models for the VSI crusher have been proposed in the last two
Modeling
decades or so. The Whiten crusher model, originally developed for cone crushers, has served as the basis of
Aggregates
several approaches to model VSI crushers. In the present work, the Andersen/Awachie/Whiten model has been
Simulation
used as the basis for modeling a VSI crusher operating in an industrial plant in Brazil, processing quarry rock to
product manufactured sand. Nineteen industrial experiments, covering a range of feed rates, rotor speeds, feed
distributor types and feed size distributions, have been carried out. The approach demonstrated to be capable of
providing satisfactory estimates of the VSI performance, being able to predict the product size distribution and
the specific energy consumption with confidence over a wide range of operating conditions. Since it uses a model
that is already available in commercial plant simulators, it may be used, with additional expressions, in simu-
lating any desired circuit. Model parameters such as K3 and T10 were found to be particularly influenced by key
operational variables such as feed rate and rotor frequency. The significant effect of feed rate on the performance
of the VSI crusher studied has been discussed on the basis of simulations of the material flow pattern inside its
feed distributing system, simulated using the discrete element method. According to these simulations, it has
been inferred that change in the behavior of VSI from lower to higher feed rates may be related to the transition
from material being fed predominantly to the rotor, to the increasing contribution of the cascading effect.

1. Introduction processed by both VSI crushers, in contrast to the product of a cone

Impact crushers have become particularly useful in manufactured
aggregates production because of the nature of the breakage mechan-
isms, which allow direct fragmentation of the particles without leaving
residual stress in them (Wills and Finch, 2016). As such, material pro-
duced in a particular size range, which may be controlled by the
combination of crushing and classification stages, may be directly used
in construction and building. This contrasts with the metal-mining in-
dustry, where the machine has been used in stages that are upstream
from grinding.
The issue of residual stress and damage associated to crushing
process has been studied by Briggs and Bearman (1996), who con-
ducted fundamental breakage tests in a Modified Hopkinson Pressure
Bar to quantify the level of damage in rocks crushed using two different
VSI crushers (Barmac and Canica) as well as a cone crusher. They found
evidence that the material consistently becomes more competent as it is
crusher.
Another noticeable feature of impact crushers is their ability to
improve particle shape in the product (Wills and Finch, 2016), which
can be advantageous in applications where the rheological properties of
the crushed material must be considered, such as in pumping of ore
concentrate and in concrete and cement mortars (Gonçalves et al.,
2007; Lindqvist, 2008).
A number of studies have been reported in the literature showing
the operational benefits achieved in comminution plants when either
cone crushers are replaced by VSI crushers, or when conventional
grinding circuits are enhanced with the incorporation of a pre-crushing
stage based on VSI technology. Examples of these improvements are:
• Barmac crushers replaced roller crushers in the final crushing stage
of a nepheline syenite processing plant, allowing feeding with a
coarser feed, at similar circuit capacity, greater ease of operation,

⁎
Corresponding author.
E-mail address: tavares@ufrj.br (L.M. Tavares).

http://dx.doi.org/10.1016/j.mineng.2017.06.022
Received 13 March 2017; Received in revised form 21 June 2017; Accepted 24 June 2017
0892-6875/ © 2017 Elsevier Ltd. All rights reserved.
J. Segura-Salazar et al.
smaller space requirement, lower energy consumption and less de-
mand for maintenance with slightly higher generation of fines (i.e.
material less than 74 μm) (Sandvik et al., 1999);
• A significantly higher energy efficiency of a VSI compared to a cone
crusher in a crushing plant processing gneiss (feed size distribution
below 50 mm), with additional increase in both capacity and energy
efficiency in subsequent grinding operations, due to the production
of a considerably finer crushing product in the former (Lindqvist,
2008);
• A potential increase of 10–20% in the throughput of a conventional
cement grinding circuit (two-compartment dry ball mill in closed
circuit with an air-swept classifier), as well as potential energy
savings in the whole clinker grinding circuit with the incorporation
of a fine pre-crushing stage using a Barmac crusher, which may
result in a lower capital investment compared to a pre-crushing
stage based on HPGR (Jankovic et al., 2004).
Given its relatively wide applicability, VSI crushing has also at-
tracted researchers interested in mathematically modeling its perfor-
mance. Recent activity has been focused on attempting to model the
process in a mechanistic way, with the aid of the discrete element
method (Cunha et al., 2013, 2014; Sinnott and Cleary, 2015). Despite
the substantial advances in these mechanistic approaches, phenomen-
ological models retain much of their importance, mainly due to the
greater simplicity of the model implementation, which facilitates the
integration with other unit operations in plant simulators, in optimizing
entire size reduction circuits.
For instance, Whiten and White (1979) considered that particle
breakage in impact crushers could be described by a combination of a
classification function followed by breakage of the selected particles.
More recently, Bengtsson and Evertsson (2008) considered a similar
approach, further discriminating the mechanisms of volume breakage
and surface breakage (attrition). This model relied on the description of
the flow of particles out of the rotor given by Rychel (2001), which
demonstrated to be able to appropriately predict power and size dis-
tribution of the product.
Originally developed for modeling of cone crushers, the Whiten
crusher model (Whiten, 1972) has also been the basis for several im-
portant phenomenological modeling approaches of VSIs, with further
improvements. Nikolov (2002) considered the dynamic behavior of
impact breakage and proposed new breakage and classification func-
tions that were meant to be specific for impact crushers, which were, in
turn, coupled to the Whiten crusher model for the calculation of the
product size distribution. These functions were related to important
operational parameters, including the rotor radius and its frequency of
rotation, besides the feed rate. The researcher also established a
minimum particle size below which the probability of particles being
classified for breakage is equal to zero. This model was validated on the
basis of experimental data from pilot plant tests in a horizontal impact
crusher (hammer mill) processing limestone (Nikolov, 2002). While
potentially applicable also to VSI crushers, it was never applied to this
type of crusher.
Researchers from the JKMRC, on the other hand, proposed a
methodology for a more direct application of the Whiten crusher model
to the VSI (Napier-Munn et al., 1996; Kojovic et al., 1998), which is
briefly described as follows:
1. Set the parameters of the classification function (Eq. (1)) as: K1 = 0
(all particles have a chance of being broken); K 2 = top size of
crusher feed; K3 = 2.3:
K 2−x K3
C (x ) = 1−⎛ ⎞ In the present work, an alternative methodology is proposed with
⎝ K 2−K 1 ⎠ (1)
where x is the representative particle size.
2. Calculate the T10 parameter (fitted value of the breakage parameter
known as “fineness index”, t10 ) as a function of A and b parameters
223
Minerals Engineering 111 (2017) 222–228
(ore-specific constants obtained by non-linear regression from drop
weight test DWT data) and the specific comminution energy Ecs
(kW h/t), given by
t10 = A·(1−e−b·Ecs ) (2)
Ecs is approximated as a function of particle peripheral velocity at
the tip of the rotor, given by
1 2 2·π·N ·r 2
Ecs ≈ v =⎛ ⎞
2 ⎝ 60 ⎠ (3)
where N is the rotational speed of rotor (RPM) and r is the rotor
radius (m)
3. Calculate the product particle size distribution (→ p ) from feed par-
→
ticle size distribution ( f ), the breakage function ( B ) experimentally
obtained from DWT, and the previously defined classification
function ( C ), using the Whiten crusher model equation
→ →
p = ( I − C )·( I − B C )−1· f (4)
where I is the unit matrix
This approach (Kojovic et al., 1998) showed good agreement be-
tween the experimental and predicted product size distribution of a VSI
crusher, but was only validated for a single set of data. No information
was either provided on its ability to predict the power or energy con-
sumed by the machine in operation when coupled to an appropriate
power model.
As a further improvement, the Andersen/Awachie/Whiten (AAW)
model (JKTech, 2014) – which is based in the original Whiten model
and is already available in process simulation platforms such as
JKSimMet® – incorporates a sub-model for the estimation of power
consumption in crushers (P ), as the sum of the power drawn by the
crusher under no load (Pn ) and the product between a theoretical power
(Pp ):
P = Pn + G·Pp (5)
where G is a dimensionless factor that depends on the type of crusher
(typically corresponds to 65–80% of energy usage efficiency for con-
ventional cone and gyratory crushers, which means a G value between
1.2 and 1.55, according to Napier-Munn et al., 1996). The theoretical Pp
power has been defined as the energy necessary to comminute the feed
size distribution to a product size distribution as if all breakage were to
occur in a single particle breakage device (such as a pendulum or a drop
weight tester), given by
j
Pp = ∑ Ecst10i·Ci·x i
i=1 (6)
where Ecst10i is the specific comminution energy at the prevailing value
of t10 for size i (kWh/t), obtained from single-particle breakage tests; j is
the number of size ranges; Ci is the classification function or probability
of breakage of size i and x i is the particle mass flow of size i in the
crusher (t/h), which is, according to the Whiten model, given by
→ →
x = ( I − B C )−1· f (7)
The AAW model also differs from the original Whiten model
(Whiten, 1972) in the fact that the breakage function is experimentally
obtained from breakage tests on individual particles on a drop weight
apparatus (JKTech, 2014; Napier-Munn et al., 1996). Equally to the
original Whiten model, the AAW model solves Eq. (4) to predict the
product size distribution.
the aim of predicting the performance of the VSI in terms of both
product size distribution and power consumption for a wide operational
range, using the AAW model as a basis. A discussion on the significant
effect of feed rate on the model predictions has been carried out on the
J. Segura-Salazar et al.
basis of simulating the material flow in the feeding system using the
discrete element method.
2. Experimental
The work focused on a VSI crusher in operation in the State of São
Paulo (Brazil), dedicated to production of manufactured sand for
mortars. The feed consists of granulite, a metamorphic rock, rich in
quartz and feldspar. In the deposit in question, the material has a beige-
grayish color, phaneritic granular texture, presenting exceptional
homogeneity in situ. It has specific gravity of 2.65 g/cm3 and Los
Angeles abrasion index of 20.4%, which characterizes it as a very
competent rock.
A representative sample of the feed to the plant was collected for
breakage testing in a drop weight apparatus. Following an adaptation of
the standard drop weight test (DWT) (Napier-Munn et al., 1996), given
the size range of interest in the feed, particles contained in the size
classes 31.5–26.5 mm, 22.4–19.0 mm and 16.0–13.2 mm were pre-
pared by sieving and separated in equivalent lots for single-particle
testing. These have then been subjected to impacts in the drop weight
tester at specific energies that vary from 0.25 to 2.5 kW h/t.
The VSI crusher installed in the circuit is a Barmac-type crusher,
with partial cascading of the feed, manufactured by Shanghai Shibang
Machinery Co. Ltd. (China). It is equipped with two 132 kW motors,
that project the rock from the 1-m diameter rotor against the lateral
anvils in the crushing chamber. The feed is introduced into the feed
hopper through a distributor or dispersion system in the form of a
truncated cone (Fig. 1). In the present work, two different feed dis-
tributor cone geometries were used (Fig. 1). The discharge of the
crusher is then classified in a vibrating screen with a 2-mm opening,
with the coarse fraction returning to the crusher mixed with the fresh
feed.
Industrial surveys were carried out under different operating con-
ditions, including: material feed size distribution, mass flow in circuit
feed, feed distributor and circuit configuration (open or closed circuit),
and rotational speed of the VSI rotor. Indeed, campaigns were con-
ducted covering feeds with top sizes from 37.5 to 26.5 mm, VSI rotor
speeds from 1200 and 1520 RPM, two types of feed distributor cones,
and feed rates (fresh feed in open circuit, and fresh feed plus recycle in
closed circuit), covering the range from 28 to 213 t/h.
The collection of samples in the industrial surveys was carried out in
reverse order of the circuit flow, that is, samples of the discharge of the
VSI crusher were taken immediately prior to sampling the feed and the
recycle stream. In order to collect the discharge sample, a deviator was
installed which diverted the material flow, through which timed sam-
ples were collected. On the other hand, 1.5-m belt cut samples were
224
Minerals Engineering 111 (2017) 222–228
collected of the feed to the crusher, during which the plant operation
was interrupted.
After weighing, samples were split until an appropriate mass was
reached for size analyzes. These analyzes were carried out through a
combination of sieving in a Ro-Tap® sieve shaker for sizes ranging from
37.2 mm down to 0.425 mm and laser-based measurements for sizes
below 0.425 mm using Mythos® (Sympatec Gmbh, Germany).
Size distributions and solids flowrates collected during surveys were
reconciled through mass balancing, using the corresponding module
available in the JKSimMet® software. Some biased tests were identified
after reconciliation and analysis of all data sets, possibly due to the
instability of the process during the conduction of the sampling cam-
paigns, particularly in some of the few tests conducted with the open
circuit configuration, which is not a common operating condition for
that circuit and that was only tested for the purpose of modeling the
crusher. Such inconsistencies were mainly detected in the particle size
distributions of VSI feed and discharge. As such, 16 out of the 19 tests
were used to develop the model.
3. Results and discussion
3.1. Breakage characteristics
The breakage characteristics of the feed have been described from
drop weight tests. Fig. 2 summarizes the results from the tests, which
show only a modest size-scale effect. Disregarding this size effect and
fitting the data to the model t10 = A [1−exp(−bEcs )], the parameters
A = 60.4 and b = 0.85 are estimated, which characterize it as a rea-
sonably tough rock for fragmentation ( A ∗ b = 51.3).
3.2. Model fitting
Typical results from experiments are shown in Fig. 3, which com-
pares measured and reconciled feed and product size distributions. In
some cases, such as in the test at 1200 RPM and 29 t/h of feed rate,
reconciliation modified the product size distribution significantly.
The methodology suggested by researchers from the JKMRC
(Napier-Munn et al., 1996; Kojovic et al., 1998) and described in the
introductory part of the present work was tested in the set of data from
the present work, with unsuccessful results regarding the prediction of
the VSI crusher performance (product size distribution) under the range
of conditions studied.
Although originally developed for compressive crushing, where the
unbroken material is mixed with the fresh feed and subjected to sequential
breakage and classification stages as they move downwards in the crushing
chamber, the Whiten model (Eq. (4)) is understood to be also valid in the
Fig. 1. Scheme of the industrial VSI crusher studied, showing
the two distributor cones tested in the feed distributor.
J. Segura-Salazar et al.
Fig. 2. Drop weight test results.
Fig. 3. Measured (Exp) and reconciled (Bal) particle size distributions of the VSI feed and
product for selected tests at different feed rates.
case of VSI crushing. The authors believe so because there is a chance that
particle propelled against the anvils may be further struck by other pro-
pelled particles before they are able to leave the active crushing zone. As
such, an approach based on the AAW model was proposed, based on some
of the initial assumptions of that earlier work:
• All particles contained in the feed may be classified for breakage,
that is, the K1 parameter of the classification function (Eq. (1)) is
equal to zero. It is also assumed that the K 2 parameter may be re-
presented by the top size of particle size distribution in the VSI feed.
As such, two values of K 2 were used, according to the different top
sizes of the feed particle size distributions studied: 26.5 mm and
37.5 mm.
• The parameter K3 – which determines the shape of the classification
function – was considered, in principle, to be a fitting parameter,
besides the T10 parameter. Both were simultaneously fitted for each
test, on the basis of reconciled data from the VSI feed and discharge,
along with power measurements during sampling campaigns, and
the breakage function experimentally obtained from DWT.
Correlations were then observed between the fitted parameters and
mass flow rate, z1, as observed in Figs. 4 and 5, and, to a lesser ex-
tent, rotor speed (z2 ).
The K3 parameter of the Whiten model classification function (Eq. (1))
is critical in determining the shape of this function, so the probability of
a particle being selected for breakage in a given narrow size range tends
to increase along with K3, for a set of K1 and K 2 values. Results of
225
Minerals Engineering 111 (2017) 222–228
Fig. 4. Relationship between fitted T10 parameter and feed rate.
Fig. 5. Relationship between K3 parameter (fitted value) and feed rate.
model parameter fitting and their correlation with the feed flowrate of
the VSI are presented in Figs. 4 and 5. It was observed that T10 tends to
proportionally increase with the mass flow rate up to a certain limit
value which is close, as expected, to the parameter A , while the op-
posite occurs with K3, which decreases as the mass flow increases, and
it varies within a range between 0.6 and 2.4.As previously mentioned,
the AAW model has a sub-model for the estimation of the crusher power
consumption, involving a power factor G which is related to the
equipment configuration and also may be calibrated against a set of
experimental power measurements under different operating condi-
tions. The experimental dataset of power measurements was analyzed,
and a clear correlation was observed between this variable and the
rotor speed, independent of feed rate (Fig. 6), showing a similar trend to
a previous study (Kelly and Dawe, 1988). Following this evidence, it
was decided to adapt the power model by using values of G that vary
with rotor speed, since the no-load power (Pn ) values were very similar
for both rotor speeds. Accordingly, the power factor was manually
fitted, so that two values were selected: 1.08 for tests with rotor speed
of 1200 RPM, and 1.54 for tests with rotor speed of 1520 RPM. Those
values would correspond, by definition, to energy usage efficiencies (1/
G ) of 93% and 65%, respectively.After including the previous con-
siderations, the model parameters were refitted and correlated to the
main operating variables of the equipment (z1, z2 ) through non-linear
regression (Eqs. (8) and (9)). The goodness of fit of the estimated T10
values (Eq. (8)) in relation to the fitted T10 values for each test may be
verified by a coefficient of correlation (R2) of 0.967 and by Fig. 4. K3
model coefficients (Eq. (9)) were established manually after an iterative
J. Segura-Salazar et al. Minerals Engineering 111 (2017) 222–228

Fig. 6. Effect of feed rate and rotor speed on measured gross power for the present work (left) and the experiments of Kelly and Dawe (1988) (right).

procedure, aiming to obtain a smoothed function instead of a step

function, whose goodness of fit may be verified in Fig. 5.

z1 z2 ⎞ ⎤
T 10 = 53.9 ⎡ ⎛
⎢1−exp ⎜−5.28 z1 + 3.73 z2 ⎟ ⎥
⎣ ⎝ ⎠⎦ (8)

−4 14.9
⎡ z z ⎞⎤
K 3 = 2.38−1.72 ⎢exp ⎛⎜−⎛ 1 ⎞ ⎛ 2 ⎞
⎜ ⎟ ⎜ ⎟
⎟⎥
⎝ ⎝ z1 ⎠ ⎝ z2 ⎠ ⎠⎦ (9)
⎣
where the reference values for mass flow rate (z1 ) and rotor speed (z2 )
have been selected as 55 t/h and 1350 RPM, respectively, making the
equations dimensionless.

3.3. Verification of the overall modeling approach

Product size distributions of the VSI crusher under different oper-

ating conditions are simulated based on the proposed modeling ap-
Fig. 8. Balanced (Bal) vs. simulated (Sim)VSI product size distributions for different feed
proach (Section 3.2) and compared to reconciled data for selected tests
size distributions and feed rates at a rotor frequency of 1520 RPM.
(Figs. 7 and 8), demonstrating the very good agreement between the
two. In addition to that, the simulated percentages of material passing
according to Eqs. (2) and (3), with r equivalent to 0.505 m for the
two selected sizes, 11.2 and 1.70 mm, which represent a wide range of
equipment under study (Fig. 1), N values equivalent to 1200 and 1520
VSI size distributions of the VSI product were compared to the corre-
RPM, and ore-specific constants according to Section 3.1, which gives
sponding experimental values (Fig. 9), showing a generally good
T10 values of 22.8 and 32.1%, respectively. It is demonstrated that the
agreement between the two. Furthermore, the proposed modeling ap-
assumptions of this approach (Kojovic et al., 1998; Napier-Munn et al.,
proach has been compared to the one proposed by researchers from
1996) are too restrictive to appropriately predict the product size dis-
JKMRC (Kojovic et al., 1998; Napier-Munn et al., 1996) as described in
tribution of a VSI crusher operating over such a wide range of condi-
the introductory section, assuming K3 equal to 2.3 and estimating T10
tions (Fig. 9), highlighting that this model depends only on a single
fitting parameter. Finally, the predicted VSI power consumption esti-
mates were compared to the corresponding measured values for all tests
(Fig. 10), demonstrating a good agreement between them for the VSI
crusher operating under a wide range of conditions.

4. Discussion

Fig. 7. Balanced (Bal) vs. simulated (Sim)VSI product size distributions for different feed
size distributions and feed rates at a rotor frequency of 1200 RPM.
While the AAW model explicitly accounts for changes in feed size
distribution and breakage characteristics, through Eqs. (2), (4) and (6),
parameters of the breakage and classification functions T10 and K3
were found to significantly vary with feed rate and rotor speed. Given
the non-straightforward implication of feed rate on both of these
parameters (Figs. 4 and 5), simulations are presented in Fig. 11 to il-
lustrate its effect. It clearly shows a transition in crusher performance in
the range of feed rate from about 70 to 140 t/h. Such transition is hy-
pothesized to result from the domination of crushing of material being
propelled by the rotor at lower feed rates, to the increase in contribu-
tion of the cascading material, at higher feed rates.
In order to gain insights into this phenomenon, simulations using the

226
J. Segura-Salazar et al.
Fig. 9. Percent passing in VSI product for selected sizes for the proposed model (left) and for the model proposed by Kojovic et al. (1998) (right).
Fig. 10. Power consumption of the VSI crusher.
discrete element method were conducted using the software EDEM Version
2.7 from DEM Solutions (Edinburgh, UK). The contact model describing
interactions between materials was described by the Hertz–Mindlin model
(Weerasekara et al., 2013). The contact parameters for the model were
taken from a previous publication by the group for rock with similar
characteristics (Cunha et al., 2013). Particles were described as spheres and
were contained in the size distribution of the feed, depicted in Fig. 8
truncating the material finer than 2.36 mm to improve computing speed.
227
Minerals Engineering 111 (2017) 222–228
The shear modulus and the Poisson’s ratio of the particles were set to
10 GPa and 0.25, and for steel 70 GPa and 0.30, respectively.
The simulated geometry was drawn according to Fig. 1, which
corresponded to the real crusher design. The DEM simulations were
limited to the feed hopper and distributor system, so that the rotor and
the discharge zone were not simulated. All the particles were created
with an initial velocity of 1.5 m/s in the negative Z axis, that is, with the
purpose of describing the adopted particles acceleration due to their
projection from a conveyor belt upstream. In order to analyze the
outcomes of the simulations, two mass flow sensors were created: the
first one was located to measure the feed flow and the second one the
rotor flow (Fig. 12). The fraction of the material sent to the rotor was
determined by the ratio between the mass flow sensors in steady state
conditions. Simulations were conducted for a range of feed rates and
the proportion of material reporting to the cascading flow recorded.
Results of the DEM simulations are illustrated in Fig. 13. It shows
that below feed rates of 120 t/h all the material is being fed to the rotor
using the distributor cone A. This threshold is only 20 t/h when dis-
tributor cone B is used. Since the majority of the tests have been con-
ducted with cone A, it is inferred that the significant change in the VSI
performance in feed rates from about 70 to 140 t/h in Fig. 11 may be
explained by this change in feed distribution inside the crusher.
However, the lack of experimental data for cone B at lower feed
rates did not allow to confirm such observation in practice, so that the
K3 model was only influenced by feed rate and rotor frequency (Eq.
(9)). Such effect could be further studied in the future by using properly
designed experiments and calibrated simulations.
Fig. 11. Simulations showing the effect of feed rate on the power con-
sumption and product size distribution, considering a feed with a top
size of 26.5 mm and a rotor speed of 1520 RPM.
J. Segura-Salazar et al.
Fig. 13. DEM simulation results of the proportion of cascading material for the two
distribution cones.
5. Conclusions
The Andersen/Awachie/Whiten model of the cone crusher has been
applied in modeling the performance of a 264 kW vertical shaft impact
crusher producing manufactured sand. The model has been supple-
mented by equations describing the variation of parameter K3 in the
classification function and the T10 parameter of the breakage function,
which were both described as a function of feed rate and rotor speed.
Within the range of conditions studied, the feed distributing system did
not play a measurable role in either of these functions. These models
were able to predict, with confidence, the product size distribution and
the power consumption of the industrial-scale VSI.
Nevertheless, the significant effect of feed rate was further analyzed
on the basis of simulations of the feed distributing system using the
discrete element method. Such simulations demonstrated that, as feed
rate increases, the proportion of material entering the rotor decreases
while the cascading flow increases. The significantly lower collision
energies to which particles that report to the cascading flow receive in
comparison to those projected by the rotor (Cunha et al., 2013) may be
responsible for the observed significant reduction in the K3 parameter
228
View publication stats
Minerals Engineering 111 (2017) 222–228
Fig. 12. Cross-section (A-A) of the VSI feed
chamber simulation.
of the classification function as the feed rate increases. This effect,
however, requires further investigation.
Acknowledgements
The authors would like to thank ECOMIX Ltda., as well as CNPq and
CAPES from Brazil, for sponsoring the research. The authors also thank
DEM Solutions for providing the software EDEM® through the
Academic Programme.
References
Bengtsson, M., Evertsson, M., 2008. Modeling of output and power consumption in ver-
tical shaft impact crushers. Int. J. Miner. Process. 88, 18–23.
Briggs, C.A., Bearman, R.A., 1996. An investigation of rock breakage and damage in
comminution equipment. Miner. Eng. 9, 489–497.
Cunha, E.R., Carvalho, R.M., Tavares, L.M., 2014. A predictive model of the Vertical Shaft
Impact (VSI) Crusher. In: Comminution ’14. Cape Town, pp. 1–12.
Cunha, E.R., Carvalho, R.M., Tavares, L.M., 2013. Simulation of solids flow and energy
transfer in a vertical shaft impact crusher using DEM. Miner. Eng. 43–44, 85–90.
Gonçalves, J.P., Tavares, L.M., Toledo Filho, R.D., Fairbairn, E.M.R., Cunha, E.R., 2007.
Comparison of natural and manufactured fine aggregates in cement mortars. Cem.
Concr. Res. 37, 924–932.
Jankovic, A., Valery, W., Davis, E., 2004. Cement grinding optimisation. Miner. Eng. 17,
1075–1081.
JKTech, 2014. JKSimMet - Version 6.0 Steady State Mineral Processing Simulator.
Brisbane.
Kelly, E.G., Dawe, G.A., 1988. Efficiency of Barmac Crusher by Modified Bond Method. In:
SME Annual Meeting. Society of Mining Engineers Inc., Phoenix, pp. 1–5.
Kojovic, T., Napier-Munn, T.J., Andersen, J.S., 1998. Modeling cone and impact crushers using
laboratory determined energy-breakage functions. Comminution Practices 187–194.
Lindqvist, M., 2008. Energy considerations in compressive and impact crushing of rock.
Miner. Eng. 21, 631–641.
Napier-Munn, T.J., Morrell, S., Morrison, R.D., Kojovic, T., 1996. Mineral comminution
circuits: their operation and optimization. JKMRC Monogr. Series Mining Mineral
Process.
Nikolov, S., 2002. A performance model for impact crushers. Miner. Eng. 15, 715–721.
Rychel, R., 2001. Modellierung des Betriebsverhaltens von Rotorschleuderbrechern.
Technischen Universität Bergakademie Freiberg.
Sandvik, K.L., Fagerli, S., Frost, T., 1999. Change from pressure crushing in roller mills to
impact crushing in production of mineral sand. Powder Technol. 105, 436–442.
Sinnott, M.D., Cleary, P.W., 2015. Simulation of particle flows and breakage in crushers
using DEM: Part 2 – impact crushers. Miner. Eng. 74, 163–177.
Whiten, W.J., 1972. The simulation of crushing plants with models developed using
multiple spline regression. J. South African Inst. Min. Metall. 257–264.
Whiten, W.J., White, M.E., 1979. Modeling and simulation of high tonnage crushing
plants. In: 12th International Mineral Processing Congress. São Paulo, pp. 148–158.
Wills, B.A., Finch, J.A., 2016. Wills’ Mineral Processing Technolology. Butterworth-
Heinamannpp. 123–146.