Materials Transactions, Vol. 46, No. 9 (2005) pp.

2036 to 2040
#2005 The Japan Institute of Metals

Prediction and Experimental Testing of Spherical Milling Media Wear Rate

Zhong Lianyun1;2;3; *1 , Wu Bolin1;3 , Zhang Lianmeng*2 , Fang Fang1;3 and He Xiaoyi3
1
Key Lab of Nonferrous Materials and New Processing Technology (Guangxi University, Guilin University of Technology),
Ministry of Education, Guilin 541004, P. R. China
2
State Key Lab of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology,
Wuhan 430070, P. R. China
3
Department of Materials and Chemical Engineering, Guilin Institute of Technology, Guilin 541004, P. R. China

This paper presents a conceptually simple derivation of grinding media mass wear rate model. An experiment of milling one media in ball
mill was designed to reveal the relation between media mass wear rate and the contacting points of the media with the other media and/or mill
wall and the pressure acting on the point. It was found that mass wear loss of spherical grinding media is proportion to the media contacting
points with other media and/or mill wall and the pressure acted on it. A physical model: f Am
1=3 t
1 b was established by theory derivation
and experiment testing. Two kinds of ceramic ball grinding media, 97 porcelain media and 75 porcelain media, were prepared. The mass wear
rate of the two kinds of media was determined by intergrinding with ordinary corundum milling media in ball mill. The experimental results
were equated using least square method. The mass wear rate equating results of both kinds of ceramic grinding media was almost the same as the
experimental results indicating the accuracy of the model. It is found that the higher the media performance was, the more accuracy the media
wear rate as predicted by the model. The possibility of using this model in practical operations was also discussed in this paper.
(Received December 17, 2004; Accepted July 6, 2005; Published September 15, 2005)
Keywords: ceramic ball milling media, mass wear rate model, theory developing, alumina ceramic ball, experiment testing

1.

Introduction

Since ball mill was discovered, it has become a most

common and non-fungible grinding methods in many
industries such as steel, ceramic, cement, glass and cosmetics
manufacturers. With respect to grinding media, world wide
steel consumption alone is estimated at over 600 000 tonnes
per annum. As such developing the ability of predicting mill
wear will allow a more accurate estimate of wear cost.1) But
to the authors knowledge, there has not a practicable
grinding media wear rate model been well established.
Typically, the ball charge motion prole exhibits three
comminution zones that can be characterized: (i) by ball
layers sliding over one another grinding material trapped
between them, the grinding zone, (ii) by balls rolling over one
another breaking material in low energy impact, the tumbling
zone, and (iii) by balls in ight re-entering into the ball
charge crushing material in high energy impact, the crushing
zone. However the contributions to total media wear of each
of these characteristics has not been well established. With
respect to abrasion, the aecting factors relevant to the
location of the media in the mill and the pressure acted on the
media is rather complex. These eect factors could be
roughly summed up as contacted points between media or
between media and mill wall, the pressure relevant to the
hardness, density and diameter of the media, the rotating
speed of the mill, the properties of the materials to be milled
and the viscosity of the liquid media etc. To present, there has
not an accurate and applicable media wear rate model been
well established for the prediction of media wear.
In fact, the prediction of the media wear rate is of
importance because it aects not only the cost of replacement
*1Undergraduate

Student, Wuhan University of Technology

*2Corresponding authors, E-mail: lmzhang@mail.whut.edu.cn;
ylzhong315.student@sina.com

media but also because the size distribution of the media in

the mill aects the grinding behavior in the mill.2) Which is
worth attentioning is that the materials worn down from the
media will become one of compositions of the materials to be
milled polluting the materials and aecting the properties of
the products to be produced.
Under similar assumptions, several authors have studied
the wear rate of spherical media and its eect on media
charge has generally been studied. It was assumed that the
mass wear rate of a piece of spherical media (mass per unit
time) is a power function of its radius r:3,4)
Fr k
4r 2 r

Where  is a linear wear rate (length per unit time),
is the

media density and
is a constant. When
0, the mass
wear rate is proportional to the surface area (4r 2 ) of the
piece and the rate of decrease of the radius is constant, which
is known as the Bond wear law. When
1, the mass wear
rate is proportional to the volume (4r 3 =3) of the piece,
which is known as the Davis wear law. However, according
to the abrasive wear mechanism dened as the removal of
surface material by rubbing or grinding down surfaces, the
abrasive wear should be in proportion to the media eective
surface area (the used surface).
Rabinowicz1) described the grinding zone abrasion as a
function of applied force F, distance slide x, abrasion grain
angle , hardness Hr and density
based on the assumption
of ball charge motion that the abrasive wear mechanism and
the impact wear mechanism were associated with the energy
dissipated in the grinding zone where ball layers slide over
one another and the energy dissipated in impact in the
tumbling and crushing zones, respectively:
mabrasion

tan
tan
Fx

Eabrasion
Hr
Hr

Where mabrasion represents loss per ball [mg/ball] where

Prediction and Experimental Testing of Spherical Milling Media Wear Rate

the energy Eabrasion is dissipated in grinding on one ball.

Increasing energy dissipated on a ball increasing material
loss independently of ball size. For the same energy
dissipated in abrasion on two balls of dierent diameter,
material loss would be the same, which for a large ball would
represent a smaller diameter change than for a small ball.
Essentially, abrasive wear is in proportion to ball surface
area. Though it is correct in theory, this equation isnt
applicable in practice because it is impossible to measure the
energy dissipated on the balls during milling.
In this paper we present a dierent and conceptually
simpler method to relate the media mass to wear law, and
demonstrate the application techniques by using alumina
ceramic (Al2 O3 ) ball media as an example. An abrasive wear
rate model is established with which experimental data is
well consistent. Two kinds of media were prepared and
compared for discussion.

2037

on the contacting points. So these factors should be

considered in the wear rate formula. Here a parameter b is
used to represent all this eect factors. Thus equation (4)
becomes:
f r

a
b
rt

For convenience in practical operations, eq. (5) was

expressed in another form by converting media radius to
mass based on the following equation:
m


4r 3
3

which is:

r


1=3
3
lm
1=3
m
4

where
2.

Theoretical Derivation of Physical Model

l

The abrasive wear mechanism was dened as the removal

of surface material by rubbing or grinding down surfaces.
Though abrasive wear is in proportion to ball surface area.5)
Consider the mass wear rate of a piece of spherical media
is dened as:
f m

m
tm0

4r 2
a

3
4
r t=3 rt


1=3

So eq. (5) becomes:

f

a
a
Am
1=3
b

b
rt
tlm1=3
t

where
3
A

where f m is the mass wear rate of the media,
m is mass

loss of the media milled for time t, and m0 is the original mass
of the media.
It is obviously that the material was ground o the media
from the locations where it contacted or rubbed with other
materials, media or mill pot. Thus the media mass wear loss
is in proportion to the media contacting or eective surface
area (4kr 2 ), where k is the media eective surface
coecient (the ratio of contacting surface to the media
whole surface) and r is the radius of the media. Consider the
mass of a piece of spherical media of radius r is m

4r 3 =3, where
is the density of the media. Thus
equation (3) becomes:
f r pk

3
4


4
where a pk 4
=3
, p is the proportional coecient of the
media eective surface to mass loss. It is obviously that
coecient a is only relevant to media related factors
including medias density and geometrical factors. a determines the contacting points of the media with other media
and the pressure of the media acting on the other media or the
materials. In other words a contributes to media wear rate
mainly through contacting points.
Excepted for the media relevant factors, as we all know,
environmental conditions, such as media mixing, the characters of the milled materials, the volume, speed and charge
of mill, compaction pressure of other medias acted on the
media and viscosity of liquid media etc, put a signicant
eect on the media wear rate by determining the contacting
points number between the other milling balls and the media,
the impact action of the milling balls and the impact pressure

a
l

Equation (5) shows that spherical media wear rate is a

reciprocal function of its radius r
1 and eq. (8) shows that
spherical media wear rate is a reciprocal function of its mass
m
1=3 . It is obviously that the coecients a, A and b in
eqs. (5) and (8) are the same for the same kind of grinding
media milling at the same grinding conditions. However at
dierent grinding conditions they are dierent for each kind
of media and milling. Actually the coecients a, A and b
should be rst corroborated by testing certain amount of
media samples mass wear rates. Then the corroborated
coecient was used as a constant to predict the media mass
wear loss of the same kind of grinding media milling at the
same conditions at any time. Using eqs. (5) and (8), it can be
well explained the well-known fact that the media wear rate
increases when its radius or mass decreases. In other words,
the spherical balls mass wear rate is in inverse proportion to
the media radius r
1 , not to the media surface area or volume.
In order to examine the accuracy of the derivation eqs. (5)
and (8), two kinds of grinding media were prepared.
3.

Experimental Procedure

Two kinds of ceramic ball grinding media were prepared in

this experiment to verify the validity of eqs. (5) and (8). One
kind of ceramic ball containing 97.3% alumina (97 porcelain)
and the other containing 75% alumina (75 porcelain). The 97
porcelain balls were prepared by using self-prepared submicron-size non-agglomerated alumina powder as raw
materials. The alumina powder was mixed with CaO
MgOAl2 O3 SiO2 quaternary system uxes consisting of
talc, clay and dolomite etc. in ball mill for 8 h. Then the

Z. Lianyun, W. Bolin, Z. Lianmeng, F. Fang and H. Xiaoyi

4.

Results and Discussion

The experimental results of 97 porcelains and 75

porcelains mass wear rate were shown in Table 1 and
Fig. 1. The coecients A and b of equation (8) were
corroborated using least square method. Equations (9) and
(10) were the adapting equations of 97 porcelain and 75
porcelain, respectively. The equating curves were shown in
Fig. 1, and the calculated results by the equating equation
were shown in Table 2.
97 porcelain:
f 2:622  10
2 m
1=3
1:616  10
3

75 porcelain:
f 2:258  10
1 m
1=3
2:293  10
2

10

In experiment, both kinds of ceramic ball milling media

were high performance grinding media with extremely low
Table 1 Media mass wear rate of 97 porcelain and 75 porcelain.
97 porcelain

75 porcelain

Media mass, Media mass wear rate Media mass Media mass wear rate
m/g
/%h
1
m/g
/%h
1
10.0847

1:06  10
2

20.2001

3

7:43  10

12.03

7:27  10
2

39.4207
78.9182

6:09  10
3
4:50  10
3

23.437
47.601

5:33  10
2
3:94  10
2

95.302

2:65  10
2

6.341

9:07  10
2

0.6
-1

mixture was dried at 100 C for 24 h. The ceramics was

shaped by cold isostatic pressing technique under 300 MPa
compressive pressure for 3 minute and sintered at electric
furnace at 1455 C in air for 3 h. The whole heating time was
controlled less than 8 h, followed by cooling in the furnace.
The 75 porcelain balls were prepared by using aluminosilicate refractory waste carried from steel company as raw
materials. The waste was mixed with CaOMgOAl2 O3
SiO2 quaternary system uxes consisting of calcium carbonate, clay and dolomite etc. in ball mill for 8 h. The mixture
was dried at 100 C for 24 h. The ceramic ball was shaped by
cold isostatic pressing technique under 65 MPa compressive
pressure for 3 min and sintered in electric furnace at 1325 C
in air for 2 h, followed by cooling in furnace, the whole
heating time being controlled less than 7 h.
The prepared alumina ceramic balls milling media were
rst interground with ordinary corundum milling media in
ball mill with 40% grinding charge. The media was mixed in
natural gradation with 0.8% (mass) media of diameter less
than 1.0 cm, 8% (mass) media of diameter between 1.0 cm
and 2.0 cm, 73.1% (mass) media of diameter between 2.0 and
3.0 cm, and 18.1% (mass) media of diameter larger than
3.0 cm. The balls were secondary ground without any other
grinding media one by one in the same mill at the same speed.
The mass wear rate was calculated using equation (3). The
experimental results were adapted using least square method.
In both experiments the mill speed was 1200 rounds per
minute and the liquid media was water. The water was added
to the mill with half ball out of water. The mill pot was of
corundum.

mass wear rate, /%.h

2038

fitting curve of 75 porcelain wear rate

0.4

experiment results of 75 porcelain wear

rate
fitting curve of 97 porcelain wear rate

0.2

0
0

20

40
60
mass of ceramic ball, m/g

80

100

Fig. 1 Relation of ceramic ball grinding media mass wear rated and the
media mass.

Table 2 Calculated results of 97 porcelain and 75 porcelain media mass

wear rate using eqs. (9) and (10), respectively.
97 porcelain

75 porcelain

Media mass, Media mass wear rate, Media mass, Media mass wear rate
m/g
/%h
1
m/g
/%h
1
10.0847

1:06  10
2

20.2001

7:43  10

3

12.03

7:56  10
2

6:09  10

3

23.437

5:60  10
2

4:50  10

3

47.601

3:94  10
2

95.302

2:65  10
2

39.4207
78.9182

6.341

9:91  10
2

wear rate. It deserves to be mentioned that the 97 porcelains

performance was so high that its wear rate was only 1/10 of
the compared samples when interground in the same ball
mill for the same time. The compared sample was carried
from Chinese factory, which was one of the most excellent
high alumina milling media of the world.
Figure 1 shows media mass wear rate decreased when
media mass increased. From Fig. 1 it is obviously that the
results of both kinds of ceramic balls media wear rate adapted
using the established model were well in agreement with the
experiment results (the results were shown in Table 1 and
Table 2), which provides the evidence for the accuracy of the
model for predicting the mass wear loss and/or mass wear
rate of spherical ball in practical operations. During production, the coecients A and b should be ascertained through
sampling test rstly before using this model to predict the
grinding media mass wear rate and/or mass wear, then they
could be used as a constant in the model for the same kind of
ceramic ball milling media at the same milling conditions.
Thus the milling medias wear extent at any time could be
easily calculated by this model, the composition of the
materials being milled or the degree of the milling materials
polluted by the media could be accurately calculated. Using
this model, the media could be fed in time and the cost of the
media could be easily estimated.
It was mentioned above that b was a parameter relating to
media location, motivation and the compaction or grinding
pressure acted on media etc. But the accurate eect of these
factors on wear rate was dicult to determine because of
their variable and indeterminacy during milling. No law has
been studied and established to correlate these factors with

Prediction and Experimental Testing of Spherical Milling Media Wear Rate

mass wear rate, /%. h -1

fitting curve of 75 porcelain media mass

wear rate grinding by one media
experiment results of 75 porcelain mass
wear rate grinding by one meida

0.008
0.007
0.006
0.005
0.004
0.003
0.002
0.001
0

fitting curve of 97 porcelain media mass

wear rate grinding by one media
experiment results of 97 porcelain mass
wear rate grinding by one maisa

20

40

60

80
100
ceramic balls mass, m/g

Fig. 2 Relations between ceramic ball mass wear rate and its mass
grinding only by one media.

media wear rate. In order to study the eect of contacting

points between media and the pressure acted on the ceramic
ball on media wear rate, Only a milling ball was put in the
ball mill and milled for some time, a simplest but most
accurate experiment was designed to simplify and simulate
media motion in mill. Then another one was subjected the
same milling process in the same mill under the same
conditions. It was obviously that the complex milling
characteristics was simplied as one media motion in the
mill. The media was milled only by one contacting point
between media and ball mill pot under the pressure of media
weight. In other words, this experiment tested the media wear
rate of wearing only by one contacting point under its weight.
The experiment results was calculated by eq. (3). The
experimental results of mass wear rate were adapted. The
experimental results and the adapting curves calculated using
least square method (eqs. (11) and (12) was 97 porcelain
medias and 75 porcelain medias adapted eqation respectively) were shown in Fig. 2.
97 porcelain:
y 3:225  10
6 m
1=3 1:894  10
4

11

2039

points between media. The milling conditions would also

directly inuence the impact action of the milling balls and
further inuence the pressure acted on the contacting points.
So the constant of the motion b put a signicant inuence on
media mass wear rate.
Comparing the wear rate of the two kinds of media shows
that the adapting results of 97 porcelain were closer to the
experimental results than 75 porcelain media (see Fig. 1 and
Fig. 2) indicating that the more excellent the media performance is, the less wear rate error the established model causes.
This would assure an accuracy of the established model. All
the experimental results of 97 porcelain media were coincide
with the adapting results. The experimental results of 75
porcelain media deviated from the adapting line was because
the experimental accuracy and the performance of the media
were not so excellent. The experimental results, although
speculative, do conrm that the model is accurate and
applicable in practical oparation. Further, associating costs to
the percentage wear changes along with associated costs of
dierent wear media will inevitably allow the determination
whether a change in charge media is cost eective or not.
It should be noted that the parameters a, A and b in wear
rate model (8) or (5) diered from dierent kinds of media
and dierent milling conditions. They are only valid to the
same kind of media milled at the same conditions. As Howat
and Vermeulen6) had shown, the media wear rate depends on
the size of the material being ground, being lower when the
product from the mill is ner. Besides, the media properties,
the liquid media viscosity, the mill speed and charge, the
diameter of the charged media etc., will aect the wear rate
of the media, which inevitably will aect the coecient a, A
and b of the wear rate model eqs. (5) and (8). To use the
present model, a, A and b must be determined by sample test
rstly. However, it is obviously that a, A and b are easy to be
determined as in this experiment did, it wouldnt become a
problem for its usage in practical operation.
5.

Conclusions

75 porcelain:
y 6:707  10
4 m
1=3 2:143  10
3

12

It is obviously in Fig. 2 that the mass wear rate was almost

equal when the spherical media was worn only by one
contacting point under its weight, the adapting curves was
almost straight lines especially the adapting line of 97
porcelain. The experiment and adapting results indicated that
the medias wear rate was equal when wearing only by one
point although their mass was dierent. Media wear loss was
proportion to its mass. These results revealed the fact that
media wear rate is proportion to both the contacting points
between media and between media and mill pot and the
pressure on the points. It is well know that large media has
more contacting points with the other media at the same
conditions. Though geometrical coecient a and A in the
model mainly determine contacting points between a media
with the other media. However the media contacting points
would be dierent when the milling conditions, including
media mixing, mill speed and volume, the location that the
media was in the mill, the viscosity of the liquid media etc. as
mentioned above, would directly inuence the contacting

(1) A conceptually simple wear rate model of spherical

media was well established by theoretical derivation.
The model is f Am
1=3 t
1 b, the coecient A and
b should be determined by sample experiment.
(2) Two kinds of alumina ceramic ball grinding media 97
porcelain and 75 porcelain were prepared as the sample,
of verifying the accuracy of the established wear rate
model. Both kinds of ceramic ball grinding media are
revealed to be excellent media.
(3) The two kinds of alumina ceramic ball grinding media
were interground with ordinary high alumina grinding
media in ball mill. The mass wear rate experimental
results were adapted using least square method by the
established wear rate model. The tting results of both
kinds of media well coincided with the experimental
results providing evidence for the accuracy of the
model.
(4) When milled only by one point under its weight, the
medias mass wear rates were almost equal and the
tting lines was a straight line for 97 porcelain
indicating that spherical media mass wear rate is in

2040

Z. Lianyun, W. Bolin, Z. Lianmeng, F. Fang and H. Xiaoyi

proportion to the contacting points between media and

between the media and mill wall and to the pressure on
the points. When the contacting point and the pressure
are the same, the media wear rate are equal, the wear
loss was proportion to the media mass.
(5) The established model is convenient to be used in
practical operation by dening the coecients by
sample testing. The higher the media performance,
the more accurate it will be.
Acknowledgments
This work was mainly supported by National Natural
Science Fund of P. R. China, under free application Grant

No. 50272016 and Guangxi Acadamic Degrees Committee

2003 Degree Authorization Oce Academic Construction
Fun, and partly by Guangxi Province Education Department
Natural Science Fund of P. R. China.
REFERENCES
1) P. Radziszewski: Miner. Eng. 13(8-9) (2000) 949961.
2) L. G. Austin, K. Shoji and P. T. Luchie: Powder Technol. 14 (1976) 71
79.
3) F. C. Bond: Mining Eng. 10 (1958) 592595.
4) E. W. Davis: Trans. AIME 61 (1919) 250296.
5) Y. Rao, Nararajan: Mineral Processing & Extractive Metallurgy Review
7 (1991) 137173.
6) D. D. Howat and L. A. Vermeulen: Powder Technol. 55 (1988) 231240.