Minerals Engineering 61 (2014) 40–46

Contents lists available at ScienceDirect

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

Effects of mixtures of grinding media of different shapes on milling

kinetics
Kalumba P. Simba ⇑, Michael H. Moys
School of Chemical & Metallurgical Engineering, University of the Witwatersrand, P.O. Box Private Bag X3, 2050 Johannesburg, South Africa

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

Article history: The effect of grinding media on milling kinetics has been generally studied using one media shape. How-
Received 27 October 2010 ever, very little work has been done on investigating mixtures of media shapes.
Accepted 10 March 2014 Combining different grinding mechanisms in term of contacts, the volume of grinding zones can be effi-
Available online 31 March 2014
ciently increased when there is an optimal mixture of two or more grinding media with different shapes
and, therefore the milling kinetics will be improved.
Keywords: A series of batch grinding tests was carried out using the same mass of spherical balls, Eclipsoids™ and
Grinding
cubes to break coarse, medium and fine sizes of quartz material. Then, mixtures of the same mass made
Grinding media shape
Mixture of grinding media
of spherical balls and cubes, spherical balls and Eclipsoids™ were successively considered. The breakage
Selection and breakage functions parameters were determined and used to evaluate the grinding performances of the mixtures of grinding
media under the same conditions.
It was found that mixtures of grinding media shapes can increase the breakage rate in a particular mill-
ing environment. But, spherical balls remain the most efficient grinding media. Finally, an optimal mix-
ture made of spherical balls and grinding media of different shapes, namely cubes and Eclipsoids that are
cheaper to manufacture can be used in the grinding process alternatively to 100% balls.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction
For decades milling has been the subject of intensive research.
Describing and understanding the process has been challenging
because of the tumbling mill environment itself. The problem with
tumbling ball mills is that they are extremely wasteful in terms of
energy consumed (Wills, 1992). The behaviour of the mill load
plays an important role in the overall grinding mechanism. It is
directly influenced by some major factors, among them the consti-
tution of the charge, the speed of rotation of the mill, and the type
of motion of individual pieces of medium in the mill (Gupta and
Yan, 2006). From an industrial point of view, the main operational
flexibility is the amount and composition of the charge. Ideally,
grinding media should have the largest possible surface area to
provide suitable contact with the material being ground and they
should be as heavy as possible to have sufficient energy required
for breaking the ore particles. Spherical balls or spheres (here re-
ferred as ‘‘balls’’) were found to be more efficient than other grind-
ing media shapes (Kelsall et al., 1973). Balls produced the finest
⇑ Corresponding author. Tel.: +27 11 717 7558; fax: +27 11 717 7591.
E-mail address: k.pascal.simba@gmail.com (Kalumba P. Simba).
products and used the least power for constant batch grinding time
(Norris, 1953). Alternative shapes to balls have been suggested to
reduce the grinding costs and increase the milling efficiency.
However, the spherical balls which are predominantly used in
ore grinding change shape through breakage and wear. They can
even break during the grinding process. The movement of these
irregularly shaped components through the mass of the charge is
believed to differ significantly from the behaviour of initial grind-
ing media shapes. Additionally these worn balls experience surface
and linear contacts with each other, while spherical ones have
only point contact interactions. The breakage is then done with a
mixture of grinding media of different shapes rather than with a
defined single shape of media.
In this research, we investigate the breakage rate for different
grinding media shapes and for the mixtures made of these shapes.
This requires a complete breakage characterization of the material
being described. The breakage parameters are determined for mix-
ture of grinding media of different shapes, and then compared to
the parameters of individual grinding media shapes. The overall
impact of the mixture of grinding media shapes is then evaluated
in terms of their grinding performances.
The grinding process is satisfactorily described using the popu-
lation balance model. In this model, milling is expressed in terms

http://dx.doi.org/10.1016/j.mineng.2014.03.006
0892-6875/Ó 2014 Elsevier Ltd. All rights reserved.
 K.P. Simba, M.H. Moys / Minerals Engineering 61 (2014) 40–46
of the selection function and the breakage function (Austin et al.,
1981). On a laboratory scale, the selection and breakage functions
are determined by batch grinding tests performed on single parti-
cle sizes for a ball charge consisting of a single size.
The selection function is the fractional rate at which a given size
of particle is broken into smaller particles. The breakage of a given
fraction of material usually obeys a first-order pattern (Austin
et al., 1984).
Rate of breakage of size i ¼ Si wi W
where wi is the weight fraction of material of size i, W is the total
material charge in the mill and Si is the specific rate of breakage
of size i.
Since the total mass W is constant, this equation becomes:
dwi ðtÞW
 ¼ S1 w1 ðtÞW
dt
S1 is the specific rate of breakage when the feed is ground for a set of
time t1, with units of time1. Then, if S1 does not vary with time, we
obtain after integration:
log w1 ðtÞ  log w1 ð0Þ ¼ S1 t=2:3
where t is the grinding time.
The formula used for the variation of the specific rate of break-
age Si with particle size is:
 a
xi 1
Si ðdÞ ¼ aðdÞ   QðxÞ with QðxÞ ¼ h iK ; K>0
x0
1 þ lxðdÞ
i
where a(d) and a are parameters dependent on the material ground
in a particular mill under defined operating conditions; xi is the par-
ticle size in mm; x0 is a reference size, usually 1 mm; l(d) defines
the particle size at which Q(x) is 0.5 and K is an index of how rap-
idly the rate of breakage falls away (Austin et al., 1982). The equa-
tion relating the value of the size xm (at which the rate of breakage
is a maximum for a given material) to the parameter l(d) is as
follows:
 1=K
K
l ¼ xm  1 ; on condition that K > a:
a were used to grind 11 different mono-sized feed fractions of quartz
The specific rate of breakage increases steadily with particle
size which reflects the decreasing strength of the particles as size
increases. This is attributed to the greater density of microflaws
in the interior of larger particles and to the greater likelihood that
a particular large particle will contain a flaw that will initiate frac-
ture under the prevailing stress conditions in a mill. The decrease
in particle strength does not lead to an indefinite increase in the
specific rate of breakage. As the particle size becomes significant
by comparison to the size of the smallest media particles, the pre-
vailing stress levels in the mill are insufficient to cause fracture and
the specific rate of breakage passes through a maximum and de-
creases with further increase in particle size (King, 2001).
The breakage function describes the size distribution of the
products of breakage. The general fitting model of the breakage
function is given by:
 c  b
xi1 xi1
Bi;j ¼ /j þ ð1  /j Þ ; 0  /j  1
x1 x1
where c, b and /j are all characteristic of the material being ground
and xi is the top size of the size interval indexed by i.
Bi,j represents the cumulative weight fraction of material broken
from size j which appears in size interval i. The subscript 1 refers to
the original material of size 1 at time t = 0, the subscript i refers to a
smaller size than the size j and the subscript j refers to the size
from which material that appears in size i are broken.
41
 d
xi
/j ¼ /1 ð7Þ
x1
where d characterizes the degree of non-normalization.
Some materials are subjected to abnormal breakage in labora-
tory mills where a departure from first-order kinetics occurs par-
ticularly for the larger particle sizes in the mill feed (Austin et al.,
1973). Several models were proposed to explain it (Austin et al.,
1977; Bilgili et al., 2006).
ð1Þ
The basic mill power drawn equation described by Hogg and
Fuerstenau (1972) is given by:
2pNs
P¼ ð8Þ
60
where N is the mill speed in rpm, s is the torque exerted by the mill
ð2Þ minus friction in the bearings.
2. Experimental
The small Wits laboratory mill was used to carry out the batch
ð3Þ
grinding tests on the quartz sample. This mill was fitted with eight
equally spaced lifters. The mill has 0.302 m diameter and 0.282 m
length, and driven by a 0.25 kW mono-phased variable speed mo-
tor. It was run at 75% of the critical speed with a ball filling J of 20%
and a powder filling U of 75%.
ð4Þ
2.1. Grinding media and test materials
Balls, cubes and Eclipsoids™ made of cast iron were used for the
batch tests. Eclipsoids™ are semi-prolate spheroid (stretched ellip-
soid of revolution). Their shape is similar to that of a half rugby
football (Fig. 1).
The spherical balls, cubes and one of the mixture used in this
study are presented in Fig. 2.
40 mm balls, 40  40 mm Eclipsoids and 32 mm cubes were
used. The total load mass was kept constant for all grinding media
shapes, as well as for the mixtures of grinding media for an average
ð5Þ porosity of 0.4 (Austin et al., 1984). The grinding media shapes
with a specific density of 2.64 g/cm3. The size fractions chosen for
the tests were ranging between 16,000 and 300 lm. These size
fractions were picked in order to get the value of the parameter
a and the value of xm, the size at which the maximum value of
the breakage rate S occurs.
The physical properties of the grinding media used are pre-
sented in Table 1.
2.2. Experimental methods
The one-size-fraction method (Austin et al., 1984) was used to
perform our batch grinding tests. Four grinding times were consid-
ered: 0.5, 1, 2 and 4 min. For every test, a blank sieving test was
done on the prepared feed material. The prepared material was
then loaded and ground in the laboratory mill for 30 s. After this
grinding period, the mill content was discharged. A full particle
ð6Þ size distribution was done on the collected product using nested
screens in decreasing order of sizepffiffiffifrom the top screen down to
the 75 lm screen at interval of 2. After all this, the material
was then recombined for batch grinding for 30 s to reach a total
grinding time of 1 min. The very same process was performed.
Then load was ground for 1 min to reach 2 min, and finally it
was ground for 2 min to reach 4 min. Each grinding process was
followed by a size analysis and the material was recombined at
the end of each process.
42 K.P. Simba, M.H. Moys / Minerals Engineering 61 (2014) 40–46

Eclipsoids Eclipsoids Eclipsoids

Fig. 1. Scaw Metals Eclipsoids™ used for the batch tests.

Spherical balls Cubes Mixture of grinding

media shapes

Fig. 2. Photographs of spherical balls, cubes and a mixture of grinding media shapes used in this study.

Table 1
Physical properties of the grinding media.

Grinding media Balls Eclipsoids Cubes Mix B–Ea Mix B–C 1a Mix B–C 2a
Mass (kg) 0.256 0.256 0.260 – – –
Specific gravity (g/cm3) 7.64 7.64 7.93 – – –
Surface area (cm2) 50.27 62.84 61.44 – – –
Volume (cm3) 33.51 33.51 32.77 – – –
Grinding media number 72 72 71 36–36 36–35 54–18
Mass charge (kg) 18.432 18.432 18.460 18.432 18.432 18.504
Total surface area (cm2) 3619.58 4524.48 4362.24 4072.03 3987.60 3820.61
Total volume (cm3) 2413.06 2413.06 2326.53 2413.06 2368.03 2399.62
a
Mix B–E is the mixture made of 50% balls and 50% Eclipsoids, Mix B–C 1 is the mixture made of 50% balls and 50% cubes and Mix B-C 2 is the mixture made of 75% balls
and 25% cubes.

3. Results and discussion
The first order plots for different feed sizes of quartz ground by
balls, cubes, Eclipsoids and the mixtures made of these media
shapes were plotted using the weight fraction remaining on the
top size. They are presented in Figs. 3–8.
These plots can be described by first-order grinding kinetics
with coefficients of determination R2 between 0.972 and 0.999
for all the single shapes, but the coefficients of determination are
between 0.883 and 0.999 for the mixtures of grinding media used.
It can be seen that the first-order law is a good approximation for
smaller sizes used, but it does not apply very satisfactorily for the
largest size (16,000 + 13,200 lm and 13,200 + 9500 lm) espe-
cially for the mixtures. In this abnormal region, there are large
lumps too strong to be broken in the mill or particles too big for

Fig. 4. First order plots for dry grinding of quartz with 40  40 mm Eclipsoids
Fig. 3. First order plots for dry grinding of quartz with 40 mm balls charge. charge.
 K.P. Simba, M.H. Moys / Minerals Engineering 61 (2014) 40–46
Fig. 5. First order plots for dry grinding of quartz with 32 mm cubes charge.
Fig. 6. First order plots for dry grinding of quartz with the mixture of 50% balls and
50% Eclipsoids charge.
Fig. 7. First order plots for dry grinding of quartz with the mixture of 50% balls and
50% cubes charge.
the energy of the tumbling mill to be used efficiently in causing
fracture (Austin et al., 1984). In this region, the breakage rate de-
creases with time in a non-linear manner (Bilgili and Scarlett,
2005). The energy applied is just sufficient to load comparatively
few regions of the quartz particle to the fracture point, and only
a few particles result (fracture by cleavage). And relatively small
pieces are cleaved off the particle, leaving a particle of essentially
the original size (Spottiswood and Kelly, 1990). The progeny size
is comparatively close to the original particle size. The fine
particles produced may then cause the slowing down phenomena
43
Fig. 8. First order plots for dry grinding of quartz with the mixture of 75% balls and
25% cubes charge.
observed and the slow movement of air through beds of fine parti-
cles. This blows away particles or absorbs impact like a hydraulic
shock absorber (Yekeler, 2007). The bed of cohesive fine particle
may develop almost liquid-like properties flowing away particles
from the grinding media – grinding media collision region leading
to insufficient stress to be transmitted to individual particles for
fracture to occur. The quartz material used exhibits abnormal
breakage, and is assumed to comprise at least two fractions: a
fast-breaking fraction and a slow-breaking one.
The variations in the specific rates of breakage at different feed
particle sizes for balls, cubes, Eclipsoids and the mixtures are
shown in Table 2.
The rate of breakage varies with size. The specific rate of break-
age increases up to a maximum feed size and decreases rapidly
above this size fraction for all grinding media charges (refer to
description). The particle sizes at which the rate of breakage is
maximum for each media shape used are presented in Table 3.
Therefore, above the maximum xm, particles are either too
strong or too big to be properly nipped and fractured by the grind-
ing media used and have a slow rate of breakage.
Figs. 9–11 show the variation of the specific rate of breakage
with size using these different grinding media shapes.
These curves show that balls are the most efficient. The break-
age rates of balls are higher than the ones of the other grinding
media in the normal grinding region. But Eclipsoids are breaking
the coarser particles faster than balls. The specific rate of breakage
of Eclipsoids for coarser particles is higher than the one of balls
(see Table 2). Fig. 9 also reveals that cubes are the least efficient
of the three single grinding media shapes considered.
The mixture of 50% balls and 50% Eclipsoids presents a higher
breakage rate in the abnormal region compared to balls and Eclip-
soids alone. This is illustrated in Fig. 10.
Fig. 11 shows that the three mixtures used have similar break-
age rate behaviour in the medium and the fine particle size region,
but they behave differently in the coarser region for the mill con-
ditions used. And for the different mixtures made of balls (the most
efficient) and cubes (the least efficient), the 75%–25% mixture has a
higher breakage rate for coarser particles, compared to the 50%–
50% mixture. This is an indication of how important the proportion
of a grinding media shape is in the mixture.
The specific rate of breakage parameters were estimated by using
the non-linear regression technique fitting the Si to Eq. (4) and are
presented in Table 4. Basically, this technique finds the best combi-
nation of fitting parameters of a model by minimizing the square of
the differences between the experimental values Pexpt(t) and the
predicted ones Pmodel(t) (Katubilwa and Moys, 2009).
The parameter K was fixed and kept constant because we do
not have enough information to determine it accurately. We used
44 K.P. Simba, M.H. Moys / Minerals Engineering 61 (2014) 40–46

Table 2
Specific rate of breakage for different particle sizes.

Particle size, xi (lm) Specific rate of breakage Si (min1)

Balls Eclipsoids Cubes Mix B–E Mix B–C 1 Mix B–C 2
16,000 + 13,200 – – – 0.950 0.449 0.729
13,200 + 9500 0.946 0.982 0.644 – – –
9500 + 6700 – – – – 0.913 –
6700 + 4750 – – – 1.620 – –
4750 + 3350 1.437 1.342 0.953 – – –
3350 + 2360 – – – 0.999 – 0.954
2360 + 1700 – – – – 0.740 –
1700 + 1180 0.487 0.469 0.266 – – –
600 + 425 – – 0.079 – – 0.141
425 + 300 0.107 0.094 – 0.086 0.093 –

Table 3
Particle sizes at which the selection function are maxima.

Media Balls Eclipsoids Cubes Mix B– Mix B–C Mix B-C

shape E 1 2
xm (lm) 7080 7297 7626 8382 6115 7575

Fig. 11. Variation of the specific rate of breakage with size for mixtures of grinding
media used.

Table 4
Breakage rate parameters obtained from the laboratory tests.

Grinding a l a K SSE R2
media

Fig. 9. Variation of the specific rate of breakage with size for balls, cubes and their Balls 0.272 8.81 1.14 3.70 0.012445000 0.996343153
mixtures. Cubes 0.151 9.49 1.14 3.70 0.009938289 0.991499046
Eclipsoids 0.258 9.08 1.14 3.70 0.011180000 0.986700132
Mix B–E 0.234 10.43 1.14 3.70 0.023279655 0.964345277
Mix B–C 1 0.257 7.61 1.14 3.70 0.035727100 0.954280896
Mix B–C 2 0.249 9.43 1.14 3.70 0.000370930 0.999511907

Fig. 10. Variation of the specific rate of breakage with size for balls, Eclipsoids and
their mixture.
the Austin’s value for K (Austin et al., 1982). a, which is character-
istic of the material, was satisfactorily determined and kept con-
stant. The parameter a is a positive number normally in the
range 0.5–1.5. It is characteristic of the material and does not vary
with rotational speed, ball load, ball size or mill hold-up over the
normal recommended test ranges (Austin and Brame, 1983) for
dry milling, but the value of a will vary with mill conditions. The
value of a that was found to satisfactorily characterize the break-
age rate is 1.14.
Then, a and l which vary with the mill conditions were deter-
mined using the same technique. The parameter a is directly pro-
portional to the specific rate of breakage. It is expressed in
time1. The parameter l is the particle size at which the correction
factor that allows for the slower mean breakage rate of the larger
sizes is 0.5. This parameter l is related to xm since both are depen-
dent on where the curve specific rate of breakage vs. particle size
starts to bend. The values of l were found to be directly propor-
tional to the corresponding values of xm for each grinding media.
The values of the parameters a and l are presented in Fig. 12.
Balls present the highest a-value parameter while cubes have
the lowest one. Consequently, they have the highest rate of break-
age while cubes have the lowest rate of breakage among all the
grinding media used. Balls also present the lowest value of param-
eter l among the single media shapes considered. This shows that
 K.P. Simba, M.H. Moys / Minerals Engineering 61 (2014) 40–46 45

which is then used as a new feed for finer grinding (Cuhadaroglu

Fig. 12. a-Values and l-values for the different grinding media used.
Fig. 13. Variation of the specific rate of breakage with size for all the grinding
media used and their mixtures.
Table 5
Breakage function parameters obtained from the laboratory tests.
Breakage function parameters b c
5.80 1.01
balls have the highest breakage rate and steadier load behaviour
during the grinding process.
On the other hand, the mixture of balls and Eclipsoids has the
highest value of parameter l among all the single media shapes
and the mixtures of grinding media used. This highest value of
the parameter l is a sign of an increase in the degree of cataracting
of the load. There is an increase of the cataracting effect, but lower
impact forces to cause breakage. And this makes this mixture capa-
ble of higher breakage rate of coarser particles in the abnormal re-
gion. In this region, the material is so much weaker in tension than
in compression, and the fracture occurs predominantly by cleavage
(Spottiswood and Kelly, 1990). Also, the geometry of balls and
Eclipsoids in presence offers point, line and area contacts. This
causes the feed material to be firstly exposed to coarse grinding
et al., 2008). Balls and Eclipsoids are subjected to conflicting behav-
iour resulting in the decrease of the breakage rate in the normal
grinding region. The parameter a decreases from 0.258 for Eclip-
soids alone to 0.234 for the mixture of balls and Eclipsoids.
Also, the addition of balls to cubes increases the breakage rate
parameter of cubes from 0.151 to 0.257 and to 0.249 respectively
for the 50%–50% and for the 75%–25% mixtures made of these grind-
ing media. The parameter l decreases from 9.49 for cubes alone to
7.61 and to 9.43 respectively for the 50%–50% and for the 75%–25%
mixtures made of these grinding media. The decrease of the catar-
acting effect is favourable to an increase of the breakage rate. This
suggests that the mixtures of balls and cubes used are more effi-
cient in term of the disappearance of the initial particle size consid-
ered for breakage compared to cubes alone. The relatively low
values of a for these mixtures compared to balls alone seem to indi-
cate that cubes have a negative effect on the balls in term of break-
age rate. Hence, the proportion of each shape in the mixture of
grinding media of different shapes needs to be investigated.
The variations of the specific rate of breakage with size for all
the grinding media used and their mixture are presented in Fig. 13.
Fig. 13 shows that balls, Eclipsoids and their mixture, as well as
the mixtures made of balls and cubes have similar breakage rate
for finer and medium particles when using the same mass of
charge. They break differently the coarser particles. It can also be
seen that cubes are the least efficient to break finer and medium
particles, but break faster coarser particles than its mixture made
of 50% cubes and 50% balls.
The Bi,j values are assumed to be normalizable (d = 0). Therefore,
the fraction appearing at sizes less than the initial feed size is inde-
pendent of the initial feed size. The B values were found to be
insensitive to the precise mill conditions, at least in the normal
operating range of milling conditions, but independent on the
grinding media shape used (Ipek, 2007). The breakage function
parameters for all our grinding media shapes were found to be:
(Table 5).
The specific energy used in all our tests was assumed to be the
/ same, as the mill power draw by the mill is basically determined by
0.71
the mass charge (Bond, 1960; Morrell, 1993). Similarly, Lameck
(2005) and Lameck et al. (2006) found that the two different grind-
ing media shapes he used drew the same amount of power at all
charges levels studied in the speed range within which most mills
are operated. This ‘general agreement’ has been verified for the
grinding media shapes used which offers point contact action, line
contact action and flat facets contacts during the grinding process.
Table 6 below presents the values of the average power drawn
by the grinding media used.
Since the mean power (79.73 W) falls within the range
(mean ± 2 standard deviations), we are 95% confident that the
powers measured for all the loads are identical.
This ‘general agreement’ that the power drawn is function of
the mass of the charge has been confirmed to be true for all shapes,
despite the fact that the grinding media shapes investigated are
supposed to present different load behaviour during the grinding
process due to their geometry.

Table 6
Power drawn by different grinding media shapes.

Balls Eclipsoids Cubes Mix B–E Mix B–C 1 Mix B–C 2

Power (W) 78.89 80.36 77.66 82.48 78.95 80.01
Coefficient of variation (%) 1.05 0.80 2.59 3.46 0.97 0.36
Mean  (1 + 2CVa) 80.54 81.64 81.68 88.18 80.48 80.58
Mean  (1  2CVa) 72.24 79.08 73.64 76.78 77.42 79.44
a
CV: coefficient of variation.
46 K.P. Simba, M.H. Moys / Minerals Engineering 61 (2014) 40–46
4. Conclusions
In this research, comparative tests have been conducted using
balls, Eclipsoids, cubes and the mixtures made of these different
grinding media shapes in a laboratory mill. The total weight of
the grinding media was kept constant. These grinding media shape
have different surface areas and contact mechanisms in grinding
action, due to differences in their geometry.
The selection function has been evaluated for coarser, medium
and fine sizes of quartz while the breakage function is assumed to
be normalizable. The breakage rate parameters were satisfactorily
determined. Considered alone, balls were revealed to be the most
efficient, cubes being the least efficient. Eclipsoids present a similar
breakage rate to balls, but they break the coarser particles faster
than balls.
All the mixtures of grinding media of different shapes presented
similar breakage rate to balls and Eclipsoids for medium and fines
particles for the same feed mass. The mixture of 50% balls and 50%
Eclipsoids which presents coefficients of variation greater than 5%
compared to balls alone in terms of both breakage rate parameters
a and l is considered not recommendable as an alternative to balls
in the mill conditions used. This mixture has a higher breakage rate
in the abnormal region. The mixture of 50% balls and 50% cubes is a
good alternative to balls and need to be used with lifters designed
specifically and at rotational speed that account for the increase of
the degree of cataracting represented by the parameter l. Finally,
the mixture of 75% balls and 25% cubes presents parameters a
and l with coefficients of variation of 5.98% and 4.98% respectively
when compared to balls. This calls for further investigations.
In general, the addition of balls to cubes showed a significant in-
crease in the rate of breakage of the mixture constituted when
compared to cubes alone. But, the proportion of cubes in the mix-
ture needs to be investigated in order to reduce the conflicting ef-
fects on the breakage rate of these different shapes in the mixture.
Therefore, an optimal mixture of grinding media of different
shapes can be used as an alternative to 100% balls in the grinding
process.
Further studies of the breakage rate and the breakage distribu-
tion under more different laboratory conditions and studies of the
load behaviour of the mixture of grinding media of different shapes
need to be done.
Acknowledgement
The Mineral Processing research group in the School of Chemical
and Metallurgical Engineering at the University of the Witwaters-
rand is acknowledged for its cooperation in conducting this work.
References
Austin, L.G., Shoji, K., Everett, M.D., 1973. An explanation of abnormal breakage of
large particle sizes in laboratory mills. Powder Technol. 7, 3–7.
Austin, L.G., Trimarchi, T., Weymont, N.P., 1977. An Analysis of some cases of non-
first-order breakage rate. Powder Technol. 17, 109–113.
Austin, L.G., Bagga, P., Celik, M., 1981. Breakage properties of some materials in a
laboratory ball mills. Powder Technol. 28, 235–243.
Austin, L.G., Shoji, K., Bell, D., 1982. Rate equations for non-linear breakage in mills
due to material effects. Powder Technol. 31 (1), 127–133.
Austin, L.G., Brame, K., 1983. A comparison of Bond method for sizing wet tumbling
ball mills with a size-mass balance simulation model. Powder Technol. 34 (2),
261–274.
Austin, L.G., Klimpel, R.R., Luckie, P.T., 1984. Processing Engineering of Size
Reduction: Ball Milling. AIME/SME, New York, USA.
Bilgili, E., Scarlett, B., 2005. Population balance of non-linear effects in milling
processes. Powder Technol. 153, 59–71.
Bilgili, E., Yepes, J., Scarlett, B., 2006. Formulation of non-liner framework for
population balance modelling of batch grinding: beyond first-order kinetics.
Chem. Eng. Sci. 61, 33–44.
Bond, F.C., 1960. Crushing and grinding calculations, Br. Chem. Eng. 6, 378–391,
543–548.
Cuhadaroglu, D., Samanli, S., Kizgut, S., 2008. The effect of grinding media shape on
the specific rate of breakage. Part. Part. Syst. Charact. 25, 465–473.
Gupta, A., Yan, D.S., 2006. Mineral Processing Design and Operation: An
introduction. Elsevier, ISBN 978-0-4444-51636-7.
Hogg, R., Fuerstenau, D.W., 1972. Power Relationships for Tumbling Mills, vol. 252.
SME/AIME, pp. 418–423.
Ipek, H., 2007. Effects of grinding media shapes on breakage parameters. Part. Part.
Syst. Charact. 24, 229–235.
Katubilwa, F.M., Moys, M.H., 2009. Effect of ball size distribution on milling rate.
Miner. Eng. 22, 1283–1288.
Kelsall, D.F., Stewart, P.S.D., Weller, K.R., 1973. Continuous grinding in a small wet
ball mill. Part V. A study of the influence of media shape. Powder Technol. 8, 77–
83.
King, R.P., 2001. Modeling and Simulation of Mineral Processing Systems.
Butterworth-Heinemann, ISBN 0-7506-4884-8.
Lameck, N.S., 2005. Effects of Grinding Media Shapes on Ball Mill Performance. MSc,
Dissertation, University of The Witwatersrand, Johannesburg.
Lameck, N.S., Kiangi, K.K., Moys, M.H., 2006. Effects of grinding media
shapes on load behaviour and mill power in a dry ball mill. Miner. Eng. 19,
1357–1361.
Morrell, S., 1993. The Prediction of Power Draw in wet Tumbling Mills. PhD Thesis,
University of Queensland, Australia.
Norris, G.C., 1953. Some grinding tests with spheres and others shapes. Trans. Inst.
Min. Metall. 63, 197.
Spottiswood, D.J., Kelly, E.G., 1990. The breakage function; what is it really? Miner.
Eng. 3 (5), 405–414.
Wills, B.A., 1992. Mineral Processing Technology: An Introduction to the Practical
Aspects of Ore Treatment and Mineral Recovery, Fifth ed. Pergamon.
Yekeler, M., 2007. Breakage and morphological parameters determined by
laboratory tests. Handb. Powder Technol. 12, 438–485.