Sunteți pe pagina 1din 7

Minerals Engineering 16 (2003) 807813

This article is also available online at:


www.elsevier.com/locate/mineng

Mechanistic modelling of mineral sizers


R. Heng a, K. Cheng a, D. Tuppurainen b, R.A. Bearman
a

b,*

, S. Oswald

Department of Mechanical Engineering, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia
b
Rio Tinto Technical Services, G.P.O. Box A42, Perth, WA 6837, Australia
c
Hamersley Iron, P.O. Box 21, Dampier, WA 6713, Australia
Received 4 March 2003; accepted 13 May 2003

Abstract
Comminution in mineral processing must be tailored to the nal product requirements. In many cases the requirement for
comminution is to generate the maximum amount of size reduction in the minimum number of equipment stages. There are however
certain applications where maximum size reduction, or nes generation, are not desirable.
In such cases, and where the material characteristics allow, the application of twin rolls, toothed mineral sizers is often advantageous. This paper examines the function of an MMD 625 mineral sizer in terms of its mechanical design parameters and its
interaction with the feed material. Using the understanding developed during the study a mechanistic predictive model of the sizer is
generated that allows analysis of product size distribution, wear and throughput.
 2003 Elsevier Ltd. All rights reserved.
Keywords: Comminution; Crushing; Modelling

1. Introduction
Key sectors of the minerals industry require comminution machinery that can crush material to a given size
without over-crushing. These include:
Coaldue to relatively fragile nature of the material
and the undesirability of nes.
Iron orein cases where nes are of a lower value, or
where ultranes should be minimized.
In such applications, equipment that is capable of
inputting high energy levels should be avoided. It should
also be noted that in the examples given above the
commodities are those regarded as bulk materials. As
bulk materials command a relatively low price, any
process equipment used in their treatment should be
capable of high throughputs particularly in relation to
the capital cost.
One device capable of high throughputs coupled with
restrained comminution is the mineral sizer. As the
name implies sizers combine a classication (sizing) and
*
Corresponding author. Tel.: +61-8-9327-2920; fax: +61-8-93272999.
E-mail address: ted.bearman@riotinto.com.au (R.A. Bearman).

0892-6875/$ - see front matter  2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0892-6875(03)00178-X

crushing (top size control) function due to their design.


The main feature of mineral sizers is that the crushing
takes place between toothed crushing rolls and/or stationery surfaces. Other advantages claimed include:
Ability to deal with high clay content, or sticky feed.
Design is self-screening, thus removing the requirement to pre-screen.
The focus of this paper is the MMD 625 mineral
sizer, where the 625 designation refers to the distance in
millimetres between the centres of the two crushing rolls.
MMD are the original and largest manufacturer of
mineral sizers with their range covering various duties
from a primary sizer (MMD 1500) down to the MMD
500. Due to the varying machine duties the design and
conguration of the MMD machines diers.
The machines designed for primary duty are the
largest throughput coarse comminution devices available. Typically the MMD 1500 machine is capable of
taking a feed size of 2 m down to a produce size of )350
mm at a rate of 10,000 tph. The MMD 625 is regarded
as a secondary machine and can be delivered in a variety
of length with either inward or outward rotation of the
crushing rolls. The main performance control features of
the MMD 625 are:

808

R. Heng et al. / Minerals Engineering 16 (2003) 807813

Length of crushing rollsthroughput;


Inward/outward rotationsize reduction top size
control and throughput;
Rolls speedthroughput;
Rolls separationtop size control, size reduction and
throughput;
Rolls/stationery surface separationsize reduction
and throughput;
Tooth congurationmanageable feed top size and
size reduction.
The MMD 625 sizers modelled in this paper are installed at the Hamersley Iron Yandicoogina operation in
Western Australia. The operation produces a )10 mm
pisolitic iron ore at a rate of 18 million tones per annum
from a three stage crushing and screening circuit. Two
MMD 625 machines act as secondary crushers receiving
an all in feed from a primary jaw crusher. Product from
the sizers is screened using three high G force banana
screens, with the undersize reporting to nal product
and the oversize being crushed using three tertiary
MP1000 cone crushers set at 14 mm closed side setting
in closed circuit with the screens.
The MMD 625 sizer is shown in Fig. 1 and has the
following conguration:
Outward rotationcrushing between toothed rolls
and stationary ngers;
Rolls speed: 60 rpm;
Rolls teethstationery teeth gap: 70 mm.
Tooth design comprises ve teeth on a segment with
48 segments on each roll (96 segments per machine), the
tooth design is shown in Fig. 2.
The machine conguration at the Yandicoogina operation was determined by MMD to meet the process
specication of 1400 tph (per machine) and a product
top size of 80 mm.

2. MMD sizer model


The MMD 625 sizes and crushes material in a variety
of ways due to the geometry of the teeth and the interaction with the feed material. The model consists of

Fig. 1. Plan view of MMD 625 Sizer at Hamersley Iron Yandicoogina.

Fig. 2. Tooth conguration for MMD 625.

Feed
Distribution

Particle
Movement

Tooth and F inger


Wear Profile

Product Size
Distribution

Redistribution

Outer
Screening

Breakage

Specific
Energy

Fig. 3. Flow chart for the fundamental model.

eight modules (see Fig. 3), where each module corresponds to a distinct stage within the machine. By having
this design, the modules can be edited or replaced
without aecting the functionality of the whole program.
The aim of the modelling is to determine:

product size distribution,


throughput,
power consumption,
rate and pattern of tooth wear.

The rate and pattern of tooth wear being of particular


importance as this in turn impacts the other process
parameters that are being calculated (Fig. 3).
The key information passed between each module is
the percentage weight of feed of a specic particle size at
a particular tooth position. This information can be
represented in a 2D matrix as shown in Fig. 4.

Fig. 4. Matrix of percentage weight, particle size, tooth position.

R. Heng et al. / Minerals Engineering 16 (2003) 807813

For the previous example, particles in the )5000 mm


to +4204 mm size fraction, at the 3rd tooth position
represent 0.05% by weight of the original feed. 5000 mm
has been assigned the value of the maximum feed size
for this model. This is deemed to be in excess of the runof-mine feed that a primary crushing device could be
expected to handle. By default the top row of % weights
will always contain zero, as no feed can be in excess of
5000 mm in size.
The size fractions dier by a factor of p412. Compared
to the standard sieve sizes that dier by p12 to allow a
greater resolution to be obtained for the particle sizes. In
the matrix, the number of teeth, n, is a parameter that
can be entered into the model by the user.
2.1. Feed distribution module
Given the design of the MMD sizer the feed distribution along the length of the machine and the position
of the falling curtain of feed with respect to the rolls is
particularly important as this controls the maximum
throughput of the machine and the wear pattern of the
teeth.
The feed system into the MMD sizers at Yandicoogina consists of a conveyor belt that empties into a
trouser leg chute that splits the feed to the two parallel
sizers. Due to constraints with the feed system the distribution along the length of the sizers is biased towards
the centre section of the rolls and the curtain of feed
tends to fall to the outside of the rolls. As these factors
have a major impact on the process performance of the
sizers a method of modelling this feed distribution is
required. The feed distribution modules aim is to simulate the way feed enters the sizer. To cater for the variety of feed distributions, the model employs three types
of feed proles: uniform, triangular or parabolic. The
a and b parameters correspond to the tooth position of the feed boundaries. This allows uneven feed to
be modelled, and will enable the user to determine the
eect of dierent feed arrangements (Fig. 5).

809

material that is smaller than the rolls separation falls


through without breakage, but the larger material is
carried by the teeth to the outside where it is broken
between the teeth and the stationary ngers.
The function of the particle movement module is
twofold:
To determine the proportion of particles that will
pass through the centre of the sizer, without undergoing breakage and concomitantly the amount reporting for breakage.
Model the movement of particles from the centre of
the sizer to the outer edge.
Trapezoids were selected as having the optimal shape
for calculating the probability of passing easily. The
probability of a particle passing through the gap is determined using the ratio of passable area to total area. In
Fig. 6 below the rock will pass through the light-coloured hole if its centroid falls within the darkened
trapezoid.
Probability of passage
Passable area

Total area
Average length  Particle diameter  ? Height  Particle diameter

Total area

The probability of selection for passage described


above is analogous to the selection function used in
many other types of breakage modelling. In this case the
more mechanistic approach to the issue of selection was
taken due to the action of the MMD and the existence of
a selection function for each tooth. It should also be
noted that the selection function will change dramatically with wear and the non-uniform wear rate and
pattern.

Height

2.2. Particle movement module


Particle motion within the sizer is a critical aspect in
understanding how material is captured and broken.
The action of the machine is that material falls in a
curtain between the two outward rotating rolls and

Average Length
Fig. 6. Possible area that a particle may pass.

Fig. 5. The uniform, triangular, parabolic feed proles.

810

R. Heng et al. / Minerals Engineering 16 (2003) 807813

2.3. Outer screening module


1
Probability of Selection
for Passage

0.5

10
0
110
Gap (mm)

50

30

70

90

90

70

Particle Size (mm)

Fig. 7. Particle passage selection function for single segment containing ve oset teeth.

Fig. 7 shows the selection function for a single


toothed segment containing ve oset teeth (as per Fig.
2). At an instant when a single tooth is at its minimum
setting (70 mm) the last of the ve teeth on the segment
will be at 110 mm, this variation is illustrated in Fig. 7
against the particle size reporting for selection.
The variation in selection function along the length of
ve toothed segments is shown in Fig. 8.
The particle movement section deals with the change
in position of the rocks as they move from the middle of
the sizer to the breakage side. Since it is assumed that
the sizer is under steady state conditions, any vertical
trajectories can be ignored.
From video footage it was determined that the
movement of the rocks is random. Thus it is appropriate
to use the random normal distribution function to calculate the probability of rock trajectories. The only
parameter that is required in such a case is the standard
deviation of normal distribution. From the video, it was
observed that no rock moved more than nine teeth positions in either direction. Therefore if we assume that 18
teeth positions is equivalent to 3d, the standard deviation for the entire prole is d  3.

Probability of Selection
for Passage

0.5

90

70
90
70
50
30
10

80

110

100

110
90
70
100
80
110
90
70

2.4. Specic energy module


Any rock undergoing breakage will have a particular
amount of energy applied to it. This module determines
the amount of energy that will be applied to a rock,
according to the machine geometry. Once the specic
energy is determined, the sizes and proportions of
breakage products can be calculated.
Kicks theory (1885) states that the energy consumed
in size reduction is proportional to the reduction in
volume of the particles undergoing breakage. Hukki
(1975) demonstrated that Kicks work is most applicable
for crushing applications where the particles are larger
than 1 cm in diameter. In order to obtain a value for the
constant of proportionality, drop weight tests by Briggs
(1997) have been analysed. By tting a log curve to the
data, a value of 0.321 kW h/t has been determined.
Thus Kicks law can be used in the following form:


d
E 0:321 ln
gap
where E is the specic energy (kW h/t), d is the initial
particle size (mm), gap is the gap size between the ngers
and the teeth (mm).
It is important to note that the geometry of the machine and the rock strength parameters (captured within
the coecient) are the sole determining factors for the
specic energy. It must be noted that another advantage
of using the Kick equation is that the coecient also
takes into account the eciency of energy transfer.

Gap (mm)

Just as the gaps in the middle of the sizer will act to


screen undersized material, a similar eect occurs on the
outer edge of the sizer. In this instance the crucial factor
is the geometry of the ngers. Before and after each
breakage event, undersized particles may fall through
the gaps in the stationary ngers.
The action of the outer section where the teeth and
stationary ngers interact is critical in determining the
size reduction behaviour of the sizer. The stationary
ngers are located on in-ll boxes so that their position relative to the teeth can be adjusted to account for
wear. On the MMD 625 there are ve ngers per in-ll
box. The sizing function of this outer area is particularly
important as there is no method of adjusting the gap
between the rolls to account for wear.
As with the particle movement module the selection of
material for breakage is a key element and the calculation of the function is undertaken in the same manner.

Particle
Size
(mm)

Fig. 8. Variation in selection function over ve segments.

2.5. Breakage module


Single particle tests to determine the comminution
behaviour of rock can be separated into pendulum and

drop weight based tests. The twin pendulum test relies


on the particle being broken between an input pendulum
released from a known height and a rebound pendulum.
The drop weight test diers in that the particles are
placed on a hard surface and struck by a falling weight.
Both these approaches have been used extensively in the
eld of comminution.
The JKMRC has specialized in the application of
these test methods since the work of Narayanan and
Whiten (1988) highlighted their use in the eld of
comminution modelling. Recent developments at the
JKMRC have seen the twin pendulum being replaced by
the drop weight apparatus.
The drop weight apparatus is seen to have several
advantages including:
extended input energy range compared to the twin
pendulum devices,
shorter time span of operation compared to the pendulum test,
extended particle size range,
ability to conduct particle bed breakage studies.
The standard drop weight device is tted with a 20 kg
mass, which can be extended to 50 kg. The eective
range of drop heights is 0.051.0 m, which represents a
wide energy range from 0.01 to 50 kW h/t (based on 10
50 mm particles).
Following sample preparation the mean mass of each
set of particles to be broken is calculated. Based on the
required specic input energy for each test, the height
from which the drop weight is to be released is determined using the relationship below:
m  Eis
hi
0:0272  Md
where hi is the initial height of the drop weight above the
anvil (cm), m is the mean mass of each set of particles
(kg), Md is the mass of the drop weight (kg), Eis is specic input energy (kW h/t).
Typically 10 mm is added to the calculated drop
height for each test. This ensures that the required nal
specic comminution energy is obtained, since after
breaking a particle the drop weight is brought to rest at
a height above the anvil. The average oset can be
measured for each sample of particles broken, in which
case the applied energy is
0:0272  Md  hi  hf
Eis
m
where hf is the average height at which the drop weight
comes to rest above the anvil.
The results from the drop weight tests provide an
energy/input size/product size relationship. This relationship is analyzed using a set of curves to describe the
size distribution produced from breakage events of increasing size reduction or energy input.

Cumulative Percent Passing

R. Heng et al. / Minerals Engineering 16 (2003) 807813

811

100
t75
80

t50
t25

60

t10

40

t4

20

t2

0
0

10

20
30
t10 (%)

40

50

Fig. 9. Size distribution parameter t versus t10.

The descriptor employed in this approach is the t


parameter. Each product size distribution curve is normalized with respect to the input size to give percentage
passing gures for various fractions. Therefore t10 is the
percentage passing 1/10th of the original feed size and t2
is the percentage passing 1/2 of the original feed size.
The t10 parameter is the most often quoted parameter. Typically in a crusher t10 is 1020%, whereas in a
tumbling mill values in the range 2050% are expected.
To make use of this description of ore breakage the
marker points t2, t4, t25, t50 and t75 are stored in a
matrix form against t10. This same data can be represented graphically as shown in Fig. 9.
Fig. 9 is a powerful graph as each vertical line (or
value of t10) represents an entire cumulative percent
passing mass size distribution.
The t10 value is related to the specic comminution
energy by the equation
t10 A1  ebEcs
where t10 is the percentage passing 1/10th of the initial
mean size, Ecs is the specic comminution energy (kW h/
t), A, b are the ore impact breakage parameters.
The A parameter represents the theoretical limiting
value of t10, whilst b is the slope of the t10 versus Ecs
graph.
Using this approach to comminution the key factors
are t10, Ecs , A and b. The relation of these parameters to
other measures of rock strength are examined later.
2.6. Tooth and nger wear
The teeth and ngers of the sizer will wear signicantly over time causing an increase in the gap size.
The mechanism of wear in the sizer was examined to
determine an appropriate simplication that could be
used in the model. Initial thoughts on the type of wear
that is experienced by the sizer teeth were weighted towards the idea of gouging wear. To investigate this issue
a series of micrographs were taken of the surface.
The results of the examination contradicted the initial
thoughts, in that the percentage of gouging wear seen on
the surface was less than 510%, with a majority of the

812

R. Heng et al. / Minerals Engineering 16 (2003) 807813

wear being attributable to hard particle abrasion. The


nding that high impact gouging wear is not a major
mechanism also allowed the issue of work hardening to
be ignored for modelling purposes. Work hardening of
the surface occurs as the crystal structure of manganese
steel alters under repeated, high force impacts. The
change in structure causes signicant changes in surface
hardness. The lack of the impact, gouging wear mechanism precludes work hardening, with this being conrmed by hardness testing of the manganese steel
surface.
With these ndings regarding the wear mechanism
and absence of work hardening the assumption was
made to treat the wear as a function of ow rate and
dierential velocity between rolls and ore material.
In order to estimate wear, the teeth and ngers were
modelled as triangular prisms. It was not viable to create
an exact model of each tooth due to the aects of
boltholes, chamfers, hard facing and the irregular wear
pattern. By determining the wear prole of the teeth, a
spectrum of product size can be derived as the geometry
of the machine changes.

3. Results and discussion


Fig. 10 shows the tooth prole predicted by the model
after 790,000 t of ore has passed through the sizer.
Overall the general trend is represented in the model.
However sections of low wear tend to be underestimated. High-wear sections show good agreement. The
dierence between observed and modelled data for the
low wear areas (towards the ends of the sizers) may be
attributable to three main reasons:
Eect of feed rate on feed distribution along the
length of the sizer.
Preference for larger particles to migrate towards the
ends of the sizer.
Mixture of wear mechanisms.

The feed rate to the sizers varies considerably over


time as the circuit prior to the sizers is a straight-through
design where there are no bins or buers to smooth the
owrate. Without these means to smooth the owrate
variation is introduced mainly from truck tips at the
primary crusher. At the higher owrates the curtain of
feed into the sizers extends further towards the ends of
the machines, thus causing higher wear rates.
The action of the sizers also tends to direct large
particles along the length of the sizer to be crushed.
Given that there are also a variations in the sizer feed
size due to variations in primary crusher setting (due to
wear) the percentage of larger feed is not always consistent through the life of the sizer teeth.
Another observation that may impact the wear towards the ends of the sizers relates to the identication
of greater gouging wear in these areas. Although not
representing sucient events to cause work hardening
the increased incidence will accelerate wear as the
gouging and stressing of the surface allows the hard
particle abrasion to be more ecient in the removal of
material. It is believed that the greater incidence of
gouging wear may be related to the discussion above
regarding the migration of larger material towards the
end, which is more dicult to grip and break.
A further observation from Fig. 10 is the serrated
nature of the measured data. Every fth tooth, displays
signicantly less wear than adjacent teeth. Opposing
ngers are positioned in sets of ve. It is believed that
the gap between nger segments results in lower wear
for the facing tooth. Further work on the model would
be required to take into account the gap sizes in the wear
calculations.
Fig. 11 displays the cumulative product size distribution predicted by the model. The initial gap between
the tip of the ngers and the root of the teeth was set
to 70 mm, and this has been replicated in the model.
The machine had processed 150,000 t of material, prior
to the reading. This has been accounted for in the
model.

100
100

80
80

% Passing

70
60
50
40

60

40

30
Measured Data

20

Model Results

10

ACTUAL

20

FEED

Tooth Position
Fig. 10. Comparison of model results to measured data.

29

27

25

23

21

19

17

15

13

11

MODEL

Tooth Height (mm)

90

0
1

10

100

1000

Size (mm)

Fig. 11. Comparison of model and actual size distribution data.

R. Heng et al. / Minerals Engineering 16 (2003) 807813

813

The model assumes normal feed conditions, uses


machine geometry values derived from the appropriate
engineering drawings.
The model shows good correlation with the measured
data, although it tends to overestimate the size reduction
in the lower end of the size distribution curve. The divergence is minor and well within the measurement accuracy of the sampling. Should the divergence be due to
deciencies in the model it is suggested that the main
causes could be either the selection function used, or the
shape of the feed material particles.

of the work is to apply similar techniques to the primary


MMD mineral sizers at the Robe River Pannawonica
operation.

4. Conclusion

References

Product size distribution and crushing component


wear are key parameters in determining the performance
of a crushing device. This paper has described the key
modules used in the modelling of the MMD 625 Mineral
Sizer. Using machine geometry, ore breakage characteristics and representations of selection and energy
input a model has been derived that successfully predicts
product size distribution and tooth wear. The next stage

Briggs, C.A., 1997. A Fundamental Model of a Cone Crusher, Ph.D.


Thesis, University of Queensland.
Hukki, R.T., 1975. The principles of comminution: an analytical
summary. Engineering Minerals Journal.
Kick, F., 1885. Das Gesetz der Proportionalen Widerstande und seine
Anwendung, Leipzig.
Narayanan, S.S., Whiten, W.J., 1988. Determination of comminution
characteristics from single particle breakage tests and its application to ball mill scale-up. In: Trans. IMM, vol. 97, Section C, pp.
C115C124.

Acknowledgements
The authors would like to thank operations sta at
Hamersley Iron Yandicoogina (John Smoothy), MMD
(Alan Potts, Ali Benbia) and Transmin (Ross Nunn,
Evan Douglas).

S-ar putea să vă placă și