Accepted Manuscript

Numerical prediction of wear in SAG mills based on DEM

simulations

Lei Xu, Kun Luo, Yongzhi Zhao

PII: S0032-5910(18)30116-5
DOI: https://doi.org/10.1016/j.powtec.2018.02.004
Reference: PTEC 13180
To appear in: Powder Technology
Received date: 16 August 2017
Revised date: 2 January 2018
Accepted date: 2 February 2018

Please cite this article as: Lei Xu, Kun Luo, Yongzhi Zhao , Numerical prediction of
wear in SAG mills based on DEM simulations. The address for the corresponding author
was captured as affiliation for all authors. Please check if appropriate. Ptec(2017),
https://doi.org/10.1016/j.powtec.2018.02.004

This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
 ACCEPTED MANUSCRIPT

Numerical prediction of wear in SAG mills based on DEM simulations

Lei Xua, Kun Luob, Yongzhi Zhao*,a

a
Institute of Process Equipment, College of Energy Engineering, Zhejiang University, Hangzhou 310027, China

b
State Key Laboratory of Clean Energy Utilization, Zhejiang University, Hangzhou 310027, China

ABSTRACT：Wear is a major operating problem for semi-autogenous grinding (SAG) mills. A credible and efficient

PT
numerical approach for accurately predicting wear within SAG mills can significantly reduce design time and improve

RI
grinding efficiency. In this paper, the 3D simulations were performed using discrete element method (DEM) combined

SC
with an erosion model, which is referred to as Shear Impact Energy Model (SIEM), to predict wear within a SAG mill.
NU
The approach is quantitively validated against the corresponding experiment reported by other researchers. The results

show that the rotation speed significantly affects the wear rate and wear distribution on the liners. In addition, the effects
MA

of changes in lifter shape on wear within the SAG mill are also obtained and analyzed. The primary reason for wear on

the liners is revealed based on the simulation results: it is the intense collisions between the particles and the liners during
D

the acceleration of the particles in the toe region that cause the severe wear. Finally, the rotation speed and lifter shape are
E
PT

synthetically evaluated from the aspect of wear and energy utilization.

Keywords: Discrete element method (DEM), Semi-autogenous grinding (SAG) mill, Wear, Grinding, Erosion.
CE
AC

* Corresponding author. E-mail address: [email protected] (Yongzhi Zhao)

1
 ACCEPTED MANUSCRIPT

1. Introduction

Wear often occurs when particles attack a surface [1]. Problems caused by wear arise frequently in

comminution devices and have significant influence on the normal operation of grinding processing systems,

which play an dominant role in many industrial fields, such as in mineral, chemical, pharmaceutical industries

[2]. To date, the development of the comminution devices is still restricted by the time and cost of the traditional

PT
design approach [3] and an efficient and credible numerical method can significantly reduce the design time and

RI
cost. Specifically, liner wear significantly affects the operating time of tumbling mills, and the replacement of

SC
worn liner can cause large economic loss [4, 5]. Moreover, the liner shape evolution caused by wear can

influence the charge motion within the tumbling mills and leads to a lower grinding efficiency [6, 7]. Therefore,
NU
an effective and efficient approach for predicting wear within the tumbling mills is of great economic
MA

significance.

To predict erosion in the comminution devices, we must accurately calculate the particle motion. Discrete
D

element method (DEM), which can provide dynamic information of individual particle and handle multiple
E

particle contacts [8-10], has become an important tool for conducting researching on large scale equipment in
PT

mineral process [11-13]. This method has been used extensively to improve the efficiency of the comminution
CE

devices [2, 14-16]. Mishra and Rajamani [17] first applied DEM to the simulations of the charge motion in a 2D
AC

ball mill. Cleary followed this work and obtained many meaningful results [2, 18, 19]. In 2002, Cleary and

Sawley simulated a 3D ball milling as a part of the DEM modeling of the industrial granular flows [20], after

which Cleary studied the axial transport within a dry ball mill [21]. Rajamani and Mishra [22], Cleary [18, 23]

and Morrison [24] are the pioneers in studying semi-autogenous grinding mills using DEM. Recently, Cleary

and Morrison proposed a method for predicting slurry transport within the grinding chamber of a SAG mill [25].

Weerasekara et al. analyzed the effects of mill size and charge particle size on the energy in grinding using DEM

2
 ACCEPTED MANUSCRIPT

modelling and drew many interesting conclusions [26]. Studying the tumbling mills (e.g., the ball, SAG and AG

[27] mills) using DEM has attracted many researchers’ interests as reviewed by Weerasekara et al. [28], such as

charge motion and power draw [14, 29-32], effects of grinding media shape [33, 34], breakage of particles

[35-38] and so on. Actually, many types of mills have been successfully simulated using DEM, such as IsaMill

[39-41], tower and pin mill [42-44], centrifugal mill [45], stirred media mills [46] and so on.

PT
A particle-scale erosion model is also essential for accurately predicting wear in the comminution devices.

RI
In 1960s, several particle-scale erosion models have been proposed by Finnie [1] and Bitter [47, 48] as reviewed

SC
by Lyczkowski and Bouillard [49]. Finnie’s model can be conveniently combined with DEM [50]. However,

this model does not appear suitable for dense conditions. Ashrafizadeh and Ashrafizadeh first found the
NU
relationship between the shear impact energy and wear [51]. Zhao et al. proposed a new quantitative erosion
MA

model (shear impact energy model, SIEM) based on this relationship [52]. Unlike Finnie’s model, the SIEM can

be applied to both the dilute and dense conditions. Only the hardness of the target surface is necessary for
D

predicting wear using SIEM and it has been proved that this model can accurately predict erosion of the tubes in
E

a fluidized bed and the elbow [52, 53].

PT

Up to now, many researchers have made efforts to investigate wear in the comminution devices using
CE

numerical approach. Mishra first made an attempt to show the energy distribution on the liner of a
AC

semi-autogenous mill using Millsoft© [54]. Kalala et al. investigated the normal and tangential forces exerted

on the lifters and validated the simulation results against the data from the corresponding experiment [55]. And

they also performed a 2D DEM simulation to predict wear on the liners and the results were validated [56].

Cleary et al. investigated liner shape evolution in a HICOM® mill using a 3D DEM simulation and found that

abrasion dominate the wear in this mill [4]. Cleary and Owen investigated the effects of the liner design on the

erosion behavior of the liner within a HICOM® mill [57]. Rezaeizadeh et al. proposed a new method for

3
 ACCEPTED MANUSCRIPT

predicting wear in a SAG mill [58] based on the work of Radziszewski et al. [59] and the simulation results were

validated. Powell et al. [60] qualitatively predicted liner evolution within a long-running ball mill using DEM.

Many efforts have been made to investigate liner wear within various comminution devices [61-63] and to

develop a methodology for accurately predicting liner wear in the comminution devices is one of the foci in this

field.

PT
In this paper, the SIEM is combined with DEM to predict liner wear within a SAG mill for the first time.

RI
The approach is quantitively validated against the experimental data reported by other researchers [6, 64]. The

SC
effects of the rotation speed and lifter shape on wear of the liners within the SAG mills are obtained and

analyzed. In addition, the primary reason for the wear within the SAG mills is revealed and some efforts have
NU
been made to evaluate the effects of the rotation speed and lifter shape synthetically based on the simulation
MA

results.

2. Mathematical model and numerical method

D

2.1. Model of particles

E

When using DEM, the particle motion is given by Newton’s kinetic equation. For a particle i with mass mi
PT

and moment of inertia Ii, two types of forces are in general taken into account: the gravity mig and the contact
CE

force. Specifically, when the particle i collides with another particle j (or a wall j), the contact force consists of
AC

two components: the normal contact force Fc,n,ij and the tangential contact force Fc,t,ij. Therefore, the

translational motion of the particle i can be calculated using Eq. (1):

ni
 mi g    Fc ,n,ij  Fc ,t ,ij 
dvi
mi (1)
dt j 1

where vi is the velocity of the particle i and ni is the number of the particles in collision with the particle i.

The particle i will also rotate due to torques. Tc,ij is the torque exerted on the particle i during the impact and

4
 ACCEPTED MANUSCRIPT

the rotational motion of the particle i is given by the following:

di ni
Ii   Tc ,ij (2)
dt j 1

where ωi is the angular velocity of particle i.

For the calculation of the contact force, Cundall and Strack [8] proposed the linear spring-dashpot model.

PT
Using this model, the two components of the contact force between two particles i and j (or a particle i and a wall

j), Fc,n,ij and Fc,t,ij, is given by the following:

RI
Fc,n,ij  kn n,ij n vn,ij (3)

SC
Fc,t ,ij  kt t ,ij t vt ,ij NU (4)

where δn,ij and δt,ij are the normal and tangential displacements between two particles i and j (or a particle i and

a wall j), respectively, vn,ij and vt,ij represent the normal and tangential relative velocities, respectively. kn is the
MA

normal stiffness of spring and kt represents the tangential stiffness of spring, and ηn and ηt represent the normal

and tangential damping coefficients, respectively. Moreover, Ting and Corkum [65] proposed several equations
D
E

to calculate the damping coefficient with the known restitution coefficient. Once the condition below is met,
PT

Fc ,t ,ij  f s Fc ,n,ij (5)

CE

the tangential contact force is given by the Coulomb friction model:

Fc,t ,ij   f s Fc,n,ij  t ,ij  t ,ij (6)

AC

where fs is the sliding friction coefficient.

In this paper, only the tangential contact torque is taken into account:

Tc,ij  rn  Fc,t ,ij (7)

where n is the normal unit vector, and r represents the radius of the spherical particle.

2.2. Erosion model

5
 ACCEPTED MANUSCRIPT

The SIEM is used in this work. This model connects the shear impact energy with the volume removed

from the surface W:

EShear
W (8)
4.0 p

in which,
t1
EShear   Fc,t ,ij  vt,ij dt , when Fc,t ,ij  vt,ij  0 and Fc,t ,ij  vi  0

PT
(9)
t0

where t0 is the start time of the contact and t1 represents the end time. The shear impact energy will be

RI
accumulated to calculate wear only when Fc,t,ij·vt,ij < 0 and Fc,t,ij·vi < 0. EShear is the shear impact energy absorbed

SC
by the surface during the impingement, p is the plastic flow pressure of the target surface and is generally 1~5
NU
times the Vickers hardness. According to the experimental data of Finnie et al [66], the plastic flow pressure is

assumed to be 1.5 times the Vickers hardness in this paper. More details on the SIEM can be found in the
MA

literatures [52, 53].

2.3. Solution and simulation conditions

E D

For this work, the self-developed DEM code was used to simulate the particle motion of the milling process
PT

within the SAG mill. A standard case was performed based on the experiment, which is a wet grinding [6, 64].
CE

Fig. 1 shows the virtual image of the whole 3D SAG mill. The 9.75×4.88 m SAG mill rotates with a constant

rotation speed of 10.5 rpm. The inside shell of the mill is covered with 60 cast chrome-molybdenum steel rail
AC

type liners, and the height of the lifter bar is 152 mm. Note that the effect of the corrosion under wet grinding

conditions is ignored in this paper. Because of the large number of the particles within the whole SAG mill, only

a slice of the mill was simulated, applying the periodic boundary conditions in both ends. There are 60 rail type

liners within the mill; therefore, 60 sets of wear data were obtained from the simulation results.

In the simulations, both the ore particles and the balls are modelled as spheres. According to the experiment

6
 ACCEPTED MANUSCRIPT

[6, 64], the ore particle size range was 1.5 ~ 17.5 cm and the ball size was 12.5 cm in the simulations. Actually,

smaller particles (below 1.5 cm) also exist in the experiment. However, these particles have little influence on

the charge motion [23] and were ignored in the simulations. The ore particle size distribution and the contact

parameters used in the simulations was estimated based on the particle size distribution estimated by Cleary [23],

which is listed in Table 1. And there are approximately 400,000 particles within the slice of the mill in the

PT
simulations. Apart from the standard case, six cases were also performed to investigate the effects of the rotation

RI
speed and lifter shape on wear of the liners. The first four rotations were used to ensure that the charge motion

SC
achieved a quasi-steady-state condition in the simulations. Then, the mill rotated another four rotations for

predicting wear. A 64-bit computer completed all of the simulations in a month, which has two Intel Xeon
NU
six-core 2.60 GHz CPUs and 16.0 GB RAM. Other details can be found in Table 2.
MA

3. Results and discussion

3.1. Charge behavior in the SAG mill

D

The charge behavior of the 10.5 rpm (N = 77%) rotation speed case, which is performed based on the
E

experiment and referred to as the standard case, is first analyzed. Specifically, Fig. 2(a) shows the distribution of
PT

the materials (balls in black and ore particles in yellow). We can see that the balls are in general uniformly
CE

distributed in the ore particles, despite the fact that the dilute cataracting stream falling from the shoulder carries
AC

few balls. Therefore, the balls in general cannot reach the liners adjacent to the toe. Actually, it can be clearly

observed that the steep face angle lifters generally trap the ore particles rather than the balls. Fig. 2(b) shows the

distribution of the velocities of the particles. The cataracting stream departs from the shoulder to the impact toe

in high ballistic trajectories and reaches the liners adjacent to the toe region with a velocity of over 12 m/s. The

well-defined bands of the cataracting materials can be observed at this rotation speed, which has been reported

by Cleary [2]. The velocities of the particles are almost 0 m/s in the low velocity region (indicated by the blue).

7
 ACCEPTED MANUSCRIPT

Moreover, the cascading flow of the free surface departing from the shoulder has a moderate velocity of

approximately 8 m/s. Actually, the characteristics of the charge motion in SAG mills have been described in

great detail by earlier work on DEM simulations of tumbling mills [2, 23, 67, 68]. Therefore, we will not give a

detail discussion in this paper.

3.2. Validation of the approach

PT
When using the SIEM, the particle is counted as a rigid body (i.e. the hardness of the particle is infinite) and

RI
all of the shear impact energy will be absorbed by the surface. However, the hardness of the ore particles in the

SC
mills is lower than of the liners (HVore : HVliner ≈ 3 : 7). It is widely accepted that the wear is inversely

proportional to the hardness of the softer material [69]. However, the hardness of the ore particles is not
NU
significantly lower than of the liners (which is the reason for why the liners undergo wear). Therefore, it seems
MA

to be a reasonable guess that 30% of the shear impact energy will be transformed into wear on the liners. Fig. 3

shows the predicted wear contour of the liners and its local details. It can be clearly observed that wear generally
D

occurs on the lifters, especially on the top right positions of the trapezoidal lifters.
E

The predicted wear profiles are compared with the experimental data [6] to validate the approach as shown
PT

in Fig. 4. The experimental wear rate was calculated based on the evolution of the lifter profile (no. 3) as shown
CE

in Figure 6 of the literature [6]. Because the lifter (no. 3) is in the middle of the mill, the effect of the wall in both
AC

ends of the mill can be ignored. The evolution of the lifter shape due to wear is not calculated in the simulations.

Therefore, only the wear rate of the first 2120 h is calculated to obtain the experiment curve. As can be seen

from Fig. 4, the simulation agrees with the experiment well. With increasing liner width (from 0 to

approximately 0.1 m), the wear rate increases sharply, i.e. the greater positions of the left steep face undergo

larger wear. Then, the wear rate will not increase until the liner width is approximately 0.15 m, which is adjacent

to the top right of the lifter. The maximum wear rate occurs at the position of approximately 0.2 m, i.e. the top

8
 ACCEPTED MANUSCRIPT

right of the lifters, which is in line with the wear contours shown in Fig. 3. Then, the wear rate decreases sharply

with increasing liner width, i.e. the greater positions of the right steep face also undergo larger wear. In addition,

the positions in the range of 0.25~0.5 m, i.e. the liners between the lifters, undergo smaller wear.

3.3. Effect of the rotation speed on wear

Fig. 5 shows snapshots of the charge motion within the mills at the rotation speeds of 9 rpm (N = 66%) and

PT
12 rpm (N = 88%).The particle size distribution used is the same as that of the standard case. For the rotation

RI
speed of 9 rpm, the shoulder position is lower than of the standard case and the bulk toe position is higher.

SC
Comparing with the standard case, the bi-linear free surface is better-defined and the semi-circular band of the

low speed particles is worse-defined. For the 12 rpm case, the shoulder position is higher than of the standard
NU
case and the bulk toe is lower. And the number of the cataracting particles increases strongly, which are in higher
MA

ballistic trajectories and the particles in general collide with the backs of the descending lifters. Comparing with

the standard case, the free surface is much steeper and the semi-circular band of the low speed particles is
D

better-defined. Actually, the effect of the rotation speed on the charge motion within the tumbling mills also has
E

been investigated and reported by earlier work in this field [19].

PT

Fig. 6 shows the predicted wear profiles of the liners at different rotation speeds. It can be clearly observed
CE

that the wear rate increases with increasing rotation speed. All of the three cases have their maximum wear rates
AC

at the liner width of approximately 0.2 m, i.e. the position near the top right of the lifers. Intriguingly, the

difference between the maximum wear rates of the standard case (Ma10.5) and the 12 rpm case (Ma12) is smaller

than the difference between the maximum wear rates of the 9 rpm case (Ma9) and the standard case (Ma10.5). The

reason will be given in the section 3.5 in this paper. In addition, all of the three cases have their second

maximum wear rates at the liner width of approximately 0.1 m, i.e. the position near the top left of the lifters. As

shown in Fig. 6, the maximum wear rates of the 9 rpm, 10.5 rpm, 12 rpm cases are respectively defined as Ma9,

9
 ACCEPTED MANUSCRIPT

Ma10.5, Ma12, and the second maximum wear rates of the 9 rpm, 10.5 rpm, 12 rpm cases are respectively defined

as Mb9, Mb10.5, Mb12. The wear distribution of 10.5 rpm case can be roughly described by R10.5 =

(Ma10.5-Mb10.5)/Ma10.5. Obviously, R10.5 > R12 > R9, i.e. the rotation speed also has effects on the wear distribution.

3.4. Effect of the lifter shape on wear

Four lifter shapes are used to investigate the effects of the lifter shape as shown in the insets of Fig. 7. Fig.

PT
7(a) and (b) shows snapshots of the charge motion within the mills at the rotation speed of 10.5 rpm when the

RI
angles of the lifter faces are 30° and 0°. Clearly, the ballistic trajectories of the cataracting materials are lower for

SC
the 30° case and higher for the 0° case comparing with that of the standard case (14°). Fig. 7(c) and (d) shows

snapshots of the charge motion within the mills at the rotation speed of 10.5 rpm when the heights of the lifters
NU
are 100 mm and 200 mm. For the 100 mm case, the ballistic trajectories of the cataracting stream are lower than
MA

of the standard case (152 mm). For the 200 mm case, the ballistic trajectories of the cataracting stream are higher

than of the standard case (152 mm). Effect of the lifter shape on the charge behavior also has been reported by
D

earlier work [19, 70].

E

Fig. 8 shows the wear rates of the liners with different lifter shapes at the rotation speed of 10.5 rpm. It can
PT

be observed that all of the cases have their maximum wear rates at the top right of the lifers and their second
CE

maximum wear rates near the top left of the lifters. Again, the maximum wear rates of the 30°, 0°, 100 mm, and
AC

200 mm cases are respectively defined as Ma30°, Ma0°, Ma100, Ma200. And the maximum wear rate of the standard

case is defined as Ma14° or Ma152. The second maximum wear rates of the 30°, 0°, 100 mm and 200 mm cases are

respectively defined as Mb30°, Mb0°, Mb100, Mb200. And the second maximum wear rate of the standard case is

defined as Mb14° or Mb152. Obviously, the maximum wear rate increases with decreasing lifter face angle and

increasing lifter height (i.e. Ma30° < Ma14° < Ma0°, Ma100< Ma152 < Ma200). Moreover, Mb30° and Mb0° are similar to

Mb14° (Mb30° ≈ Mb14° ≈ Mb0°). Finally, Mb200 is the largest of the five second maximum wear rates and Mb100 is the

10
 ACCEPTED MANUSCRIPT

smallest (Mb100< Mb152 < Mb200).

3.5. Analysis of primary reason for wear

To reveal the main reason for the occurring of wear within the SAG mill, the wear of a single lifter at

different moments in one circle were extracted to calculate the transient wear rate. Because the initial position of

the lifter is known, it is convenient to calculate the changing of its position in time. Then, the variation of the

PT
transient wear rate of the lifter along circumference of the shell as described in the inset of Fig. 9(a) (the top

RI
point of the SAG mill is 0°) is obtained. Because the of cataracting material on the liners are not continuous but

SC
bunched together, data of 10 lifters are extracted and averaged. As shown in Fig. 9(a), the trends of the curves

are similar to each other. For the standard case, the transient wear rate begins to increase sharply when the
NU
position of the lifter is approximately 110°. At this position, the lifter is close to the toe region and a few falling
MA

ore particles can attack its surface. Then, the transient wear rate increases sharply with increasing position and

reaches its maximum value when the position is in the range of approximately 130° to 160° where the lifter is
D

under the particles of the toe region. Therefore, the maximum transient wear rate is likely caused by the intense
E

collisions between the particles and liners at the beginning of the grinding process. Actually, Cleary has reported
PT

that the largest amounts of shear occur in the toe region where material abruptly changes its motion from
CE

cascading down the surface to being pulled under the bulk of the charge by the lifters [19], which indicates that
AC

the wear has a positive relation with the amounts of shear. Finally, the transient wear rate decreases to almost 0

rapidly. Moreover, all the transient wear rates increase a little in the range of approximately 270° to 300°. The

reason is likely that when the particles are lifted to the shoulder, the slight sliding of the particles across the

liners become more intense under the effect of the gravity. Additionally, the transient wear rate in general

increases with increasing rotation speed, which is in line with the results shown in Fig. 6. It can also be clearly

seen that the maximum wear rate occurs later with increasing rotation speed and the reason is that the bulk toe

11
 ACCEPTED MANUSCRIPT

moves lower with increasing rotation speed. Moreover, the range of the positions undergoing wear becomes

wider with increasing rotation speed. The reason is that both the shoulder and impact toe become higher, though

the bulk toe becomes lower (which is the reason for the lower transient wear rate of the 12 rpm case in the range

of 110° to 130° comparing with the other two cases). Intriguingly, the maximum transient wear rate of the 12

rpm case is smaller than of the other two cases and the reason will be given later in this section.

PT
In general, particles with larger kinetic energy will cause larger erosion. Therefore, we analyzed the time

RI
domain average kinetic energy of the particles in the target region at different rotation speeds as shown in Fig.

SC
9(b). The target region in this section is defined as the region with a width of 2 times the height of the lifters

adjacent to the liners along the circumference of the shell as described in the inset of Fig. 9(b). Specifically, the
NU
kinetic energy increases sharply in the range of approximately 130° to 160°. Then, the kinetic energy basically
MA

remains unchanged in the range of approximately 160° to 270° before decreases rapidly in the range of

approximately 270° to 300°. Moreover, the maximum kinetic energy increases with increasing rotation speed.
D

Comparing Fig. 9(b) with Fig. 9(a), it can be seen that large kinetic energy cannot cause large transient wear rate
E

within the SAG mill at a certain rotation speed. Actually, the maximum transient wear rate appears when the
PT

kinetic energy increases rapidly, i.e. when the particles are accelerated. Therefore, the kinetic energy of the
CE

material cascading down the surface in the toe region is not the main reason for the occurring of the intense
AC

collisions between the particles and liners. The intense collisions mainly occur during the acceleration of the

particles adjacent to the bulk toe, which is the primary cause of the wear on the liners and is the reason for why

the maximum wear rate occurs at the top right of the lifers as shown in Fig. 6. Moreover, with increasing rotation

speed, the acceleration of the particles last longer. Therefore, the intense collisions also last longer and the

maximum wear rate increases.

As mentioned above, wear mainly occurs on the liners adjacent to the bulk toe. Fig. 10 shows the particle

12
 ACCEPTED MANUSCRIPT

size distribution in the toe region and an apparent separation can be observed. Specifically, more large particles

exist in the bulk toe when the rotation speed is smaller, i.e. the amount of separation increases with decreasing

rotation speed. Cleary has reported the effect of the rotation speed on the size segregation [2] and no detail

discussion will be given here for brevity. To investigate the particle size distribution quantitatively, the time

domain average volume fractions and kinetic energy fractions of the particles with different sizes in the toe

PT
regions within the mills at different rotation speeds are obtained as shown in Fig. 11. The toe region in this

RI
section is defined as the region with a width of 2 times the height of the lifters adjacent to the liners in the range

SC
of 110° to 180° as described in the inset of Fig. 11. Specifically, Fig. 11(a) shows that the volume fraction of the

small particles (0.015 ~ 0.025 m) increases with increasing rotation speed. In addition, the volume fraction of
NU
the large particles (0.125 ~ 0.175 m) decreases with increasing rotation speed, which agrees with the results
MA

shown in Fig. 10. Fig. 11(b) shows that the kinetic energy distribution with increasing particle size at different

rotation speeds in the toe region and the results are similar with that of the volume fraction: the kinetic energy
D

fraction of the small particles (0.015 ~ 0.025 m) increases with increasing rotation speed and that of the large
E

particles (0.125 ~ 0.175 m) decreases with increasing rotation speed. Note that the value of the kinetic energy
PT

fraction of the particles is not the same as the corresponding value of the volume fraction. When the rotation
CE

speed is 12 rpm, the kinetic energy fraction of the large particles (0.125 ~ 0.175 m) is still approximately 0.45,
AC

though the corresponding volume fraction is only approximately 0.35. Clearly, when the rotation speed is high

(12 rpm), the liners are almost covered with small particles before they enter into the toe region and undergo the

intense collisions with the large particles, which is the reason for why the maximum transient wear rate is

smaller than of the other two cases as shown in Fig. 9(a) and why the difference between the maximum wear

rates of the standard case (Ma10.5) and the 12 rpm case (Ma12) is smaller than the difference between the

maximum wear rates of the 9 rpm case (Ma9) and the standard case (Ma10.5) as shown in Fig. 6. Because the

13
 ACCEPTED MANUSCRIPT

cataracting materials have little influence on the second maximum wear rate (Fig. 8(a) shows that the second

maximum wear rate of the 0° case is not large, though the trajectories of the cataracting materials are high as

shown in Fig. 7(b)), the sliding of the small particles is the reason for why the second maximum wear rate of 12

rpm case is apparently larger than of the other two cases. Therefore, the size distribution in the toe region affects

the wear distribution on the liners.

PT
Based on the analysis above, we anticipate that the angle of the lifter face will have little influence on the

RI
size distribution in the toe region because of the results shown in Fig. 8(a): Mb30° ≈ Mb14° ≈ Mb0°. In addition, we

SC
also anticipate that the height of the lifter will have some effect on the size distribution in the toe region because

of the results shown in Fig. 8(b): Mb200 > Mb152 > Mb100. The results are indeed in line with the expectation
NU
(which are not listed here for brevity). However, the effect of the lifter height on the particle size distribution in
MA

the toe region is not as large as that of the rotation speed, which indicates that the segregation in the toe region is

not the only reason. Unlike the effect of the lifter shape on the second maximum wear rate (Mb), it is self-evident
D

that the maximum wear rate (Ma) will increase with decreasing lifter face angle and increasing lifter height.
E

3.6. Energy utilization and synthetic evaluation of the rotation speed and lifter shape
PT

As mentioned in the above section, the wear increases with increasing rotation speed in the range of 9 to 12
CE

rpm (N = 66 ~ 88%). The rotation rate can also affect the energy utilization, which is another essential
AC

characteristic of the SAG mill. Therefore, it is necessary to investigate the energy utilization of the SAG mill at

different rotation speeds.

Fig. 12 shows the collision frequency and energy dissipation rate during the impacts exerted on the liners in

1 s time interval. It can be seen that the most common collision energy exerted on the liners is approximately

5×10-7 J in all the three cases, whereas the maximum energy dissipation rate occurs when the collision energy is

approximately 1 J. Therefore, most of the collisions are gentle and can only leave slight wear on the liners and a

14
 ACCEPTED MANUSCRIPT

small number of the intense collisions are the main reason for the occurring of wear. Moreover, both the

frequency and energy dissipation rate of the intense collisions exerted on the liners increase with increasing

rotation speed. Fig. 13 shows the total collision frequency and total energy dissipation rate during the impacts

exerted on the ore particles. It can be seen that the most common collision energy exerted on the ore particles is

approximately 2×10-3 J in all the three cases, and the maximum energy dissipation rates of the 9, 10.5, 12 rpm

PT
cases respectively occur when the energy is approximately 0.02 J, 0.03 J and 0.1 J. Unlike the results of the

RI
liners, the differences of the three curves are not large.

SC
To investigate the effects of the rotation speed on the energy utilization, we have calculated the total

dissipation energy rates consumed by the liners and the ore particles as shown in Table 3 and the total wear rate
NU
on the liners as shown in Table 4. The results show that the rotation speed has a large effect on the total energy
MA

dissipation rate consumed by the liners, which has a positive relationship with the total wear rate. In addition,

the rotation speed also has influence on the useful breakage, which can be described by the total energy
D

dissipation rate consumed by the ore particles. Actually, EL/EO increases with increasing rotation speed (from
E

approximately 0.11 to 0.18), which indicates that the energy utilization becomes worse with increasing rotation
PT

speed.
CE

The effects of the lifter shape on the total energy dissipation rates consumed by the liners and the ore
AC

particles are given by Table 5 and the total wear rates on the liners are given by Table 6. The results indicate that

the lifter shape also has large effects on the total energy dissipation rate consumed by the liners, which has a

positive relationship with the total wear rate on the liners. In addition, the lifter shape also has influence on the

useful breakage. Obviously, EO of the 200 mm case is larger than of the standard case and the reason is probably

that higher lifters result in more frequent impact events as reported by Djordjevic et al. [70]. Intriguingly, EO of

the 100 mm case is also a little larger than of the standard case. In addition, EO of the 0° case is only a little larger

15
 ACCEPTED MANUSCRIPT

than of the standard case while EO of the 30° case is significantly larger. The reason for these results is likely that

the motion state between the particles adjacent to the liners is altered due to the effect of the lifter shapes.

Actually, it has been reported by Cleary that the power consumption and the lifter angle is not proportional [19].

It can be obtained from Table 5 that EL/EO of the 0° case is the largest of the five cases. Therefore, the energy

utilization become worse when the lifter face is very steep (the angle is 0°). Finally, it seems that the energy

PT
utilization of the 30° case is better than the other four cases, though the lifter is worn comparing with the

RI
standard case.

SC
4. Conclusions

The DEM and erosion model SIEM were used to predict wear within the SAG mill. The simulation results
NU
of the standard case (the rotation speed is 10.5 rpm; the angle of the lifer face is 14°; the lifter height is 152 mm)
MA

are quantitively validated against the corresponding experimental data [6], and the effects of the rotation speed

and lifter shape on wear within the SAG mill have been obtained and discussed. After the discussion, the
D

following conclusions are made:

E

(1) The accuracy of the approach is good for predicting wear within the SAG mill. The simulation results
PT

basically match the experiment.

CE

(2) The rotation speed has significant influence on wear of the liners within the SAG mill. The wear rate
AC

increases with increasing rotation speed (from 9 to 12 rpm). In addition, the rotation speed also has

some effects on the wear distribution.

(3) The lifter shape also has some effects on wear of the liners within the SAG mill. The maximum wear

rate increases with decreasing lifter face angle (from 30° to 0°) and increasing lifter height (from 100 to

200 mm).

(4) The intense collisions between the particles and the lifters adjacent to the charge toe during the

16
 ACCEPTED MANUSCRIPT

acceleration of the particles (i.e. the beginning of the grinding process) is the primary cause of wear on

the liners. Therefore, the maximum wear rate occurs at the top left of the lifters. In addition, the rotation

speed can affect the size distribution of the ore particles in the toe region, which has important

influence on the wear distribution.

(5) The current approach is effective for studying wear within the SAG mill. Ongoing research involves

PT
applying the non-spherical DEM combined with the SIEM to model wear within the comminution

RI
devices and predicting the evolution of the lifter shape based on the current approach.

Acknowledgments
SC
NU
This research is funded by the National Natural Science Foundation of China (NSFC, Grant No. 21476193).
MA

References
D

[1] I. Finnie, Erosion of surfaces by solid particles, Wear, 3 (1960) 87-103.

E
PT

[2] P.W. Cleary, Predicting charge motion, power draw, segregation and wear in ball mills using discrete element methods,

Miner. Eng., 11 (1998) 1061-1080.

CE

[3] R.D. Morrison, P.W. Cleary, Towards a virtual comminution machine, Miner. Eng., 21 (2008) 770-781.
AC

[4] P.W. Cleary, P. Owen, D.I. Hoyer, S. Marshall, Prediction of mill liner shape evolution and changing operational

performance during the liner life cycle: Case study of a Hicom mill, Int. J. Numer. Meth. Eng., 81 (2010) 1157-1179.

[5] M. Akhondizadeh, M.F. Mahani, M. Rezaeizadeh, S.H. Mansouri, Prediction of tumbling mill liner wear: Abrasion

and impact effects, Proc. Inst. Mech. Eng. J J. Eng. Tribol., 230 (2016) 1310-1320.

[6] S. Banisi, M. Hadizadeh, 3-D liner wear profile measurement and analysis in industrial SAG mills, Miner. Eng., 20

(2007) 132-139.

17
 ACCEPTED MANUSCRIPT

[7] D. Boemer, J.-P. Ponthot, A generic wear prediction procedure based on the discrete element method for ball mill

liners in the cement industry, Miner. Eng., 109 (2017) 55-79.

[8] P.A. Cundall, O.D.L. Strack, A discrete numerical model for granular assemblies, Geotechnique, 29 (1979) 47-65.

[9] H.P. Zhu, Z.Y. Zhou, R.Y. Yang, A.B. Yu, Discrete particle simulation of particulate systems: theoretical

developments, Chem. Eng. Sci., 62 (2007) 3378-3396.

PT
[10] C.-Y. Wu, Discrete Element Modelling of Particulate Media, Royal Society of Chemistry, Cambridge, 2012.

RI
[11] P.W. Cleary, Large scale industrial DEM modelling, Eng. Computation., 21 (2004) 169-204.

SC
[12] P.W. Cleary, Industrial particle flow modelling using discrete element method, Eng. Computation., 26 (2009)

698-743.
NU
[13] P.W. Cleary, R.C.Z. Cohen, S.M. Harrison, M.D. Sinnott, M. Prakash, S. Mead, Prediction of industrial, biophysical
MA

and extreme geophysical flows using particle methods, Eng. Computation., 30 (2013) 157-196.

[14] P.W. Cleary, Ball motion, axial segregation and power consumption in a full scale two chamber cement mill, Miner.
D

Eng., 22 (2009) 809-820.

E
PT

[15] P.W. Cleary, R.D. Morrison, Particle methods for modelling in mineral processing, Int. J. Comput. Fluid D., 23

(2009) 137-146.
CE

[16] H.P. Zhu, Z.Y. Zhou, R.Y. Yang, A.B. Yu, Discrete particle simulation of particulate systems: a review of major
AC

applications and findings, Chem. Eng. Sci., 63 (2008) 5728-5770.

[17] B.K. Mishra, R.K. Rajamani, The discrete element method for the simulation of ball mills, Appl. Math. Model., 16

(1992) 598-604.

[18] P.W. Cleary, Modelling comminution devices using DEM, Int. J. Numer. Anal. Met., 25 (2001b) 83-105.

[19] P.W. Cleary, Charge behaviour and power consumption in ball mills: sensitivity to mill operating conditions, liner

geometry and charge composition, Int. J. Miner. Process., 63 (2001a) 79-114.

18
 ACCEPTED MANUSCRIPT

[20] P.W. Cleary, M.L. Sawley, DEM modelling of industrial granular flows: 3D case studies and the effect of particle

shape on hopper discharge, Appl. Math. Model., 26 (2002) 89-111.

[21] P.W. Cleary, Axial transport in dry ball mills, Appl. Math. Model., 30 (2006) 1343-1355.

[22] R.K. Rajamani, B.K. Mishra, Dynamics of ball and rock charge in sag mills, In: Proc. SAG (1996), Mining and

Mineral Process Engineering, University of British Columbia, Vancouver, B.C., Canada, pp. 700-712.

PT
[23] P.W. Cleary, Recent advances in DEM modelling of tumbling mills, Miner. Eng., 14 (2001c) 1295-1319.

RI
[24] R. Morrison, P.W. Cleary, W. Valery, Comparing power and performance trends from DEM and JK modelling, In:

SC
Proc. SAG (2001), Mining and Mineral Process Engineering, University of British Columbia, Vancouver, B.C., Canada,

pp. 284-300.
NU
[25] P.W. Cleary, R.D. Morrison, Prediction of 3D slurry flow within the grinding chamber and discharge from a pilot
MA

scale SAG mill, Miner. Eng., 39 (2012) 184-195.

[26] N.S. Weerasekara, L.X. Liu, M.S. Powell, Estimating energy in grinding using DEM modelling, Miner. Eng., 85
D

(2016) 23-33.
E
PT

[27] G.W. Delaney, P.W. Cleary, R.D. Morrison, S. Cummins, B. Loveday, Predicting breakage and the evolution of rock

size and shape distributions in AG and SAG mills using DEM, Miner. Eng., 50 (2013) 132-139.
CE

[28] N.S. Weerasekara, M.S. Powell, P.W. Cleary, L.M. Tavares, M. Evertsson, R.D. Morrison, J. Quist, R.M. Carvalho,
AC

The contribution of DEM to the science of comminution, Powder Technol., 248 (2013) 3-24.

[29] M.A. Van Nierop, G. Glover, A.L. Hinde, M.H. Moys, A discrete element method investigation of the charge motion

and power draw of an experimental two-dimensional mill, Int. J. Miner. Process., 61 (2001) 77-92.

[30] P.W. Cleary, R. Morrisson, S. Morrell, Comparison of DEM and experiment for a scale model SAG mill, Int. J.

Miner. Process., 68 (2003) 129-165.

[31] O. Hlungwani, J. Rikhotso, H. Dong, M.H. Moys, Further validation of DEM modeling of milling: effects of liner

19
 ACCEPTED MANUSCRIPT

profile and mill speed, Miner. Eng., 16 (2003) 993-998.

[32] P.W. Cleary, M.D. Sinnott, R.D. Morrison, DEM prediction of particle flows in grinding processes, Int. J. Numer.

Meth. Fl., 58 (2008) 319-353.

[33] F. Shi, Comparison of grinding media—Cylpebs versus balls, Miner. Eng., 17 (2004) 1259-1268.

[34] N.S. Lameck, M.H. Moys, Effects of media shape on milling kinetics, Miner. Eng., 19 (2006) 1377-1379.

PT
[35] B.K. Mishra, C. Thornton, Impact breakage of particle agglomerates, Int. J. Miner. Process., 61 (2001) 225-239.

RI
[36] L.M. Tavares, R.P. King, Single-particle fracture under impact loading, Int. J. Miner. Process., 54 (1998) 1-28.

SC
[37] L.M. Tavares, R.M. Carvalho, Modeling breakage rates of coarse particles in ball mills, Miner. Eng., 22 (2009)

650-659.
NU
[38] M. Yahyaei, N.S. Weerasekara, M.S. Powell, Characterisation of abrasion breakage using multi size pilot mills, In:
MA

13th European Symposium on Comminution and Classification (2013), Brunswick, pp. 229–232.

[39] C.T. Jayasundara, R.Y. Yang, A.B. Yu, D. Curry, Discrete particle simulation of particle flow in the IsaMill process,
D

Ind. Eng. Chem. Res., 45 (2006) 6349-6359.

E
PT

[40] C.T. Jayasundara, R.Y. Yang, A.B. Yu, D. Curry, Discrete particle simulation of particle flow in IsaMill—effect of

grinding medium properties, Chem. Eng. J., 135 (2008) 103-112.

CE

[41] C.T. Jayasundara, R.Y. Yang, A.B. Yu, J. Rubenstein, Effects of disc rotation speed and media loading on particle
AC

flow and grinding performance in a horizontal stirred mill, Int. J. Miner. Process., 96 (2010) 27-35.

[42] M. Sinnott, P.W. Cleary, R. Morrison, Analysis of stirred mill performance using DEM simulation: Part 1–media

motion, energy consumption and collisional environment, Miner. Eng., 19 (2006) 1537-1550.

[43] P.W. Cleary, M. Sinnott, R. Morrison, Analysis of stirred mill performance using DEM simulation: Part 2–Coherent

flow structures, liner stress and wear, mixing and transport, Miner. Eng., 19 (2006) 1551-1572.

[44] R.D. Morrison, P.W. Cleary, M.D. Sinnott, Using DEM to compare the energy efficiency of pilot scale ball and

20
 ACCEPTED MANUSCRIPT

tower mills, Miner. Eng., 22 (2009) 665-672.

[45] P.W. Cleary, D. Hoyer, Centrifugal mill charge motion and power draw: comparison of DEM predictions with

experiment, Int. J. Miner. Process., 59 (2000) 131-148.

[46] R. Gers, E. Climent, D. Legendre, D. Anne-Archard, C. Frances, Numerical modelling of grinding in a stirred media

mill: hydrodynamics and collision characteristics, Chem. Eng. Sci., 65 (2010) 2052-2064.

PT
[47] J.G.A. Bitter, A study of erosion phenomena part I, Wear, 6 (1963) 5-21.

RI
[48] J.G.A. Bitter, A study of erosion phenomena: Part II, Wear, 6 (1963) 169-190.

SC
[49] R.W. Lyczkowski, J.X. Bouillard, State-of-the-art review of erosion modeling in fluid/solids systems, Prog. Energy

Combust. Sci., 28 (2002) 543-602.

NU
[50] K.W. Chu, S.B. Kuang, A.B. Yu, A. Vince, G.D. Barnett, P.J. Barnett, Prediction of wear and its effect on the
MA

multiphase flow and separation performance of dense medium cyclone, Miner. Eng., 56 (2014) 91-101.

[51] H. Ashrafizadeh, F. Ashrafizadeh, A numerical 3D simulation for prediction of wear caused by solid particle impact,
D

Wear, 276 (2012) 75-84.

E
PT

[52] Y. Zhao, L. Xu, J. Zheng, CFD–DEM simulation of tube erosion in a fluidized bed, AlChE J., 63 (2017) 418-437.

[53] L. Xu, Q. Zhang, J. Zheng, Y. Zhao, Numerical prediction of erosion in elbow based on CFD-DEM simulation,
CE

Powder Technol., 302 (2016) 236-246.

AC

[54] B.K. Mishra, A review of computer simulation of tumbling mills by the discrete element method: Part II—Practical

applications, Int. J. Miner. Process., 71 (2003) 95-112.

[55] J.T. Kalala, M.M. Bwalya, M.H. Moys, Discrete element method (DEM) modelling of evolving mill liner profiles

due to wear. Part I: DEM validation, Miner. Eng., 18 (2005) 1386-1391.

[56] J.T. Kalala, M. Bwalya, M.H. Moys, Discrete element method (DEM) modelling of evolving mill liner profiles due

to wear. Part II. Industrial case study, Miner. Eng., 18 (2005) 1392-1397.

21
 ACCEPTED MANUSCRIPT

[57] P.W. Cleary, P. Owen, Effect of liner design on performance of a HICOM® mill over the predicted liner life cycle,

Int. J. Miner. Process., 134 (2015) 11-22.

[58] M. Rezaeizadeh, M. Fooladi, M.S. Powell, S.H. Mansouri, N.S. Weerasekara, A new predictive model of lifter bar

wear in mills, Miner. Eng., 23 (2010) 1174-1181.

[59] P. Radziszewski, R. Varadi, T. Chenje, L. Santella, A. Sciannamblo, Tumbling mill steel media abrasion wear test

PT
development, Miner. Eng., 18 (2005) 333-341.

RI
[60] M.S. Powell, N.S. Weerasekara, S. Cole, R.D. LaRoche, J. Favier, DEM modelling of liner evolution and its

SC
influence on grinding rate in ball mills, Miner. Eng., 24 (2011) 341-351.

[61] M. Lindqvist, C.M. Evertsson, Liner wear in jaw crushers, Miner. Eng., 16 (2003) 1-12.
NU
[62] J.F. Archard, Contact and rubbing of flat surfaces, J. Appl. Phys., 24 (1953) 981-988.
MA

[63] T. Teeria, V.-T. Kuokkala, P. Siitonenh, P. Kivikyto-Reponenh, J. Liimatainenb, Impact wear in mineral crushing, In:

Proceedings of the Estonian Academy of Sciences, Engineering, Estonian Academy Publishers 12 (2006), 408-418.
D

[64] M. Yahyaei, S. Banisi, M. Hadizadeh, Modification of SAG mill liner shape based on 3-D liner wear profile
E
PT

measurements, Int. J. Miner. Process., 91 (2009) 111-115.

[65] J.M. Ting, B.T. Corkum, Computational laboratory for discrete element geomechanics, J. Comput. Civil Eng., 6
CE

(1992) 129-146.
AC

[66] I. Finnie, J. Wolak, Y. Kabil, Erosion of metals by solid particles, J. Mater., 2 (1967) 682-700.

[67] P.W. Cleary, M. Sinnott, R. Morrison, Prediction of slurry transport in SAG mills using SPH fluid flow in a dynamic

DEM based porous media, Miner. Eng., 19 (2006) 1517-1527.

[68] P. Owen, P.W. Cleary, The relationship between charge shape characteristics and fill level and lifter height for a SAG

mill, Miner. Eng., 83 (2015) 19-32.

[69] V.L. Popov, Contact Mechanics and Friction, Springer, Berlin, 2010.

22
 ACCEPTED MANUSCRIPT

[70] N. Djordjevic, F.N. Shi, R. Morrison, Determination of lifter design, speed and filling effects in AG mills by 3D

DEM, Miner. Eng., 17 (2004) 1135-1142.

PT
RI
SC
NU
MA
E D
PT
CE
AC

23
 ACCEPTED MANUSCRIPT

Figure Captions
Fig. 1. Virtual image of the whole 3-D SAG mill.

Fig. 2. Steady state charge behavior at the rotation speed of 10.5 rpm (N = 77%) when the angle of the lifter face is 14°

and the height of the lifters is 152 mm: (a) distribution of the materials, balls in black and ore particles in yellow, (b)

distribution of the particle velocities, blue: 0 m/s, red: 12 m/s.

PT
Fig. 3. The predicted wear contour of the liners when the angle of the lifter face is 14° and the height of the lifters is 152

RI
mm: (a) the whole SAG mill, (b) the local details.

SC
Fig. 4. Predicted wear profiles of the liners compare with the experimental data from Banisi and Hadizadeh [6]. The

angle of the lifter face is 14° and the height of the lifters is 152 mm.
NU
Fig. 5. Steady state charge behavior for different rotation rates: (a) 9 rpm (N = 66%) and (b) 12 rpm (N = 88%) when the
MA

angle of the lifter face is 14° and the height of the lifters is 152 mm (blue: 0 m/s, red: 12 m/s).

Fig. 6. The predicted wear profiles of the liners for rotation rates between 9 rpm and 12 rpm. The angle of the lifter face
D

is 14° and the height of the lifters is 152 mm.

E
PT

Fig. 7. Steady state charge behavior for lifers with face angles: (a) 30° and (b) 0° when the lifter height is 152 mm and for

lifters with heights: (c) 100 mm and (d) 200 mm when the angle of the lifter face is 14°. The rotation speed is 10.5 rpm
CE

(blue: 0 m/s, red: 12 m/s).

AC

Fig. 8. The predicted wear profiles of the liners using different lifter shapes: (a) different lifter face angles between 0°

and 30° when the lifter height is 152 mm; (b) different lifter heights between 100 mm and 200 mm when the angle of the

lifter face is 14°. The rotation speed is 10.5 rpm.

Fig. 9. (a) Predicted transient wear rate profiles of a single lifter and (b) kinetic energy of the particles adjacent to the

lifter along circumference of the shell within the SAG mill at different rotation speeds. The angle of the lifter face is 14°

and the height of the lifters is 152 mm.

24
 ACCEPTED MANUSCRIPT

Fig. 10. The snapshot of the particle size distribution within the SAG mill at different rotation speeds: (a) 9 rpm (N =

66%), (b) 10.5 rpm (N = 77%), (c) 12 rpm (N = 88%) (blue: 1.5 cm, red: 17.5 cm). The angle of the lifter face is 14° and

the height of the lifters is 152 mm.

Fig. 11. (a) The volume fraction and (b) kinetic energy fraction of the particles in the toe regions at different rotation

speeds versus particle size. The angle of the lifter face is 14° and the height of the lifters is 152 mm.

PT
Fig. 12. Comparison of (a) collision frequency and (b) energy dissipation spectra of the impacts exerted on liners at

RI
different rotation speeds. The angle of the lifter face is 14° and the height of the lifters is 152 mm.

SC
Fig. 13. Comparison of (a) collision frequency and (b) energy dissipation spectra of the impacts exerted on ore particles

at different rotation speeds. The angle of the lifter face is 14° and the height of the lifters is 152 mm.
NU
MA
E D
PT
CE
AC

25
 ACCEPTED MANUSCRIPT

Table Captions
Table 1 Ore particle size distribution used in the simulation.

Table 2 Simulation parameters.

Table 3 Total energy dissipation rates consumed by the liners and the ore particles at different rotation speeds. The angle

of the lifter face is 14° and the height of the lifters is 152 mm.

PT
Table 4 Total wear rates at different rotation speeds. The angle of the lifter face is 14° and the height of the lifters is 152

RI
mm.

SC
Table 5 Total energy dissipation rates consumed by the liners and the ore particles when the rotation speed is 10.5 rpm.

The angle of the lifter face is 14° (standard case), 30°, 0° when the height of the lifters is 152 mm and the height of the
NU
lifter is 152 mm (standard case), 100 mm, 200 mm when the angle of the lifter face is 14°.
MA

Table 6 Total wear rates when the rotation speed is 10.5 rpm. The angle of the lifter face is 14° (standard case), 30°, 0°

when the height of the lifters is 152 mm and the height of the lifter is 152 mm (standard case), 100 mm, 200 mm when
D

the angle of the lifter face is 14°.

E

Nomencalture
PT
CE
AC

26
 ACCEPTED MANUSCRIPT

Table 1

Ore particle size (cm) Volume fraction of the ore particles (%)

1.5 ~ 2.5 25

2.5 ~ 3.5 18

3.5 ~ 4.5 15

PT
4.5 ~ 7.5 15

RI
7.5 ~ 12.5 15

SC
12.5 ~ 17.5 12
NU
MA
E D
PT
CE
AC

27
 ACCEPTED MANUSCRIPT

Table 2

Parameters Value

Parameters of the liners

Vickers hardness HV370

Length of the liners (m) 0.5

PT
Height of the lifters H (mm) 152, 100, 200

RI
Angle of the lifer face (°) 14, 30, 0

SC
No. of the lifters 60

Parameters of the particles

NU
Shape sphere
MA

Density (kg/m3) 4500 (ore), 7800 (ball)

Vickers hardness HV160 (ore), HV370 (ball)

D

Coefficient of restitution between ore particles (-) 0.3

E
PT

Coefficient of restitution between ore and steel (-) 0.5

Coefficient of restitution between ball and liner (-) 0.8

CE

Coefficient of friction fs (-) 0.5

AC

Normal spring constant kn (N/m) 2.8×106

Tangential spring constant kt (N/m) 8×105

Parameters of the mill

Rotational speed (rpm) 9, 10.5, 12

Slice length (m) 0.5 (along the cylindrical belly)

Mill filling by volume (%) 35

28
 ACCEPTED MANUSCRIPT

Ball filling by volume (%) 15

PT
RI
SC
NU
MA
D
E
PT
CE
AC

29
 ACCEPTED MANUSCRIPT

Table 3

Rotation Speed (rpm) Energy Dissipation Rate Energy Dissipation Rate Consumed by

Consumed by Liners EL (J/s) Ore Particles EO (J/s)

9.0 3.07103 2.77105

10.5 4.24103 3.38105

PT
12.0 7.82103 4.15105

RI
SC
NU
MA
E D
PT
CE
AC

30
 ACCEPTED MANUSCRIPT

Table 4

Rotation Speed (rpm) Wear Rate (m3/h)

9.0 4.9810-4

10.5 6.2610-4

12.0 9.1910-4

PT
RI
SC
NU
MA
ED
PT
CE
AC

31
 ACCEPTED MANUSCRIPT

Table 5

Angle and Height of the Energy Dissipation Rate Energy Dissipation Rate Consumed

Lifter Consumed by Liners EL (J/s) by Ore particles EO (J/s)

Standard Case (14°, 152 mm) 4.24103 3.38105

30°, 152 mm 3.45103 3.49105

PT
0°, 152 mm 7.17103 3.39105

RI
100 mm, 14° 4.35103 3.43105

SC
200 mm, 14° 4.99103 3.47105
NU
MA
E D
PT
CE
AC

32
 ACCEPTED MANUSCRIPT

Table 6

Rotation Speed (rpm) Wear Rate (m3/h)

Standard Case (14°, 152 mm) 6.2610-4

30°, 152 mm 5.5510-4

0°, 152 mm 8.1210-4

PT
100 mm, 14° 5.8310-4

RI
200 mm, 14° 6.9510-4

SC
NU
MA
DE
PT
CE
AC

33
 ACCEPTED MANUSCRIPT

Nomencalture

Variable Description Unit

mi mass of particle i kg

vi velocity of particle i m/s

Fc,n,ij normal contact force between particle i and j N

PT
Fc,t,ij tangential contact force between particle i and j N

RI
g gravitational acceleration m/s2

SC
Ii moment of inertia of particle i kg·m2

ωi angular velocity of particle i rad/s

NU
Tc,ij torque between particle i and j N·m
MA

δn,ij normal displacement between particle i and j m

δt,ij tangential displacement between particle i and j m

D

vn,ij normal relative velocity between particle i and j m/s

E
PT

vt,ij tangential relative velocity between particle i and j m/s

kn normal spring stiffness N/m

CE

kt tangential spring stiffness N/m

AC

ηn normal damping stiffness kg/s

ηt tangential damping stiffness kg/s

fs sliding friction coefficient -

r particle radius m

W volume loss of target surface m3

vi,t tangential velocity of particle i m/s

34
 ACCEPTED MANUSCRIPT

EShear shear impact energy J

p plastic flow pressure Pa

PT
RI
SC
NU
MA
E D
PT
CE
AC

35
 ACCEPTED MANUSCRIPT

Highlights

1. The wear in SAG mills is obtained by DEM simulation and the results are validated.

2. The SIEM is applied to predict wear in SAG mills based on DEM.

3. The effects of rotation speed on wear in SAG mills are obtained.

4. The effects of lifer shape on wear in SAG mills are obtained.

PT
5. The main reason for wear on the liners is the collisions under the toe region.

RI
SC
NU
MA
E D
PT
CE
AC

36
Graphics Abstract
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11
Figure 12
Figure 13