You are on page 1of 11

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

net/publication/264545541

A more holistic view of mill liner management

Conference Paper · October 2012

CITATIONS READS

5 622

7 authors, including:

Malcolm S. Powell Marko Hilden


The University of Queensland The University of Queensland
150 PUBLICATIONS   1,511 CITATIONS    22 PUBLICATIONS   125 CITATIONS   

SEE PROFILE SEE PROFILE

Nirmal S. Weerasekara Mohsen Yahyaei


The University of Queensland The University of Queensland
32 PUBLICATIONS   387 CITATIONS    38 PUBLICATIONS   209 CITATIONS   

SEE PROFILE SEE PROFILE

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

Variable Rates 2 AG/SAG mill model View project

CSRP Project 85 (2B10) Triple Pass HPGR Circuit Concept View project

All content following this page was uploaded by Mohsen Yahyaei on 15 July 2015.

The user has requested enhancement of the downloaded file.


HOME

A More Holistic View of Mill Liner


Management
M S Powell1, M M Hilden2, N Weerasekara3, M Yahyaei4, P Toor5,
J Franke6 and M Bird7

ABSTRACT
The combination of new tools with a more rigorous approach to understanding the interaction of
mill liners with overall mill performance and economics, forms the basis of an integrated approach
to liner management. Based on accurate liner wear measurements, some (limited) physical mill
inspections and surveys, accurate calibration of the mill load and analysis of process data, a whole
new approach to liner selection, relining intervals and mill control is proposed. Collection and
analysis of this information in operating mills illustrates the mill performance improvement
which can be attained through this careful measurement and modelling approach.
A more holistic approach to liner management is proposed, including a broadening of the
current work to cover the potential benefits of improved relining technologies and predictive liner
wear modelling.

INTRODUCTION Liner wear and grinding performance


There are two main functions for liners in a mill. The first is to The aggressive milling environment limits liner life. Mills
protect the mill shell from the aggressive milling environment. require regular liner replacement at a considerable material
The second is to deliver the input power to the charge through and labour cost in addition to the cost of lost production time
keying-in the charge to the rotary motion of the mill and lifting during shut-downs. An in-depth summary of this issue is
the charge up so that it can cascade or cataract down to provide presented in Powell et al (2006). The challenging aspect is that,
the energy for grinding. For mills with identical operating as liners wear, the shape of the lifters changes dramatically,
conditions, a different liner design, generally dominated by which in turn changes the energy transfer characteristics and
the profile of the lifter, will change the charge profile and thus has a significant influence on grinding efficiency (Powell
consequently the mechanism of grinding (Powell and Nurick, et al, 2011). The dual functions of the liners tend to conflict,
1996a, 1996b, 1996c). There have been numerous studies as designing the liner profile to maximise liner life inevitably
aimed at quantifying the effect of liner configurations on the reduces the power delivered by the mill and thus the overall
dynamics of the mill charge (McIvor, 1983; Vermeulen, 1985; grinding rate in the mill. It is therefore necessary to develop a
Powell, 1991; Mishra and Rajamani, 1993; Moys, 1993; Cilliers liner design which balances liner life and milling performance.
et al, 1994; Radziszewski, 1997; Van Nierop et al, 2001; Cleary,
2001; Rajamani, Joshi and Mishra, 2002; Djordjevic, Shi and Mill liner management tools
Morrison, 2004; Kalala, Bwalya and Moys, 2005a, 2005b; Kiangi Utilising an integrated approach in mill liner management
and Moys, 2006; Makokha and Moys, 2006; Makokha et al, 2007; identifies opportunities to improve mill control, specifically
Kalala, Breetzke and Moys, 2008; Yahyaei and Banisi, 2010). as it relates to changes in performance occurring over the liner
These studies indicate that there should be an optimum liner life. This approach requires accurate liner wear measurements,
configuration for a given milling duty. Due to the wear of liners some limited physical mill inspections and surveys, and an
they cannot constantly maintain this optimal profile, so the accurate calibration of the mill’s operating parameters with
challenge becomes one of designing liners to provide the best liner life, in order to develop an optimised operating strategy.
mill performance over the life of the liner. In the broadest sense Utilising plant data to quantify the throughput, power
this optimisation must include mill throughput and grind, liner consumption and product size generated by the mill as the
costs and relining time. liner wears, can help in the generation of a new liner design

1. FAusIMM, Professorial Research Fellow, Chair in Sustainable Comminution, Julius Kruttschnitt Mineral Research Centre, Sustainable Minerals Institute, The University of Queensland,
40 Isles Road, Indooroopilly Qld 4068. Email: malcolm.powell@uq.edu.au
2. Senior Research Fellow, Julius Kruttschnitt Mineral Research Centre, Sustainable Minerals Institute, The University of Queensland, 40 Isles Road, Indooroopilly Qld 4068. Email: m.hilden@uq.edu.au
3. Research Fellow, Julius Kruttschnitt Mineral Research Centre, Sustainable Minerals Institute, The University of Queensland, 40 Isles Road, Indooroopilly Qld 4068. Email: n.weerasekara@uq.edu.au
4. Postdoctoral Research Fellow, Julius Kruttschnitt Mineral Research Centre, Sustainable Minerals Institute, The University of Queensland, 40 Isles Road, Indooroopilly Qld 4068.
Email: m.yahyaei@uq.edu.au
5. MAusIMM, Analytic Metallurgist, Scanalyse Pty Ltd, Suite 1, Enterprise Unit 5, 7a De Laeter Way, Technology Park, Bentley WA 6102. Email: Paul.toor@scanalyse.com.au
6. Chief Science Officer, Scanalyse Pty Ltd, Suite 1, Enterprise Unit 5, 7a De Laeter Way, Technology Park, Bentley WA 6102. Email: Jochen.franke@scanalyse.com.au
7. Senior Metallurgist, Cadia Valley Operations, Newcrest Mining LTD, PMB, South Orange NSW 2800. Email: Matthew.bird@newcrest.com.au

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 95


M S POWELL et al

with optimum mass and life characteristics and enhanced


cost-life benefits. DEM simulation outputs provide detailed
data of the force, duration and frequency of collisions on
finely discretised elements of the liner. Laboratory wear tests
can be used to investigate the impact on mill operations of
liner life and profile changes due to wear. The correlation of
DEM simulation results with the outputs of laboratory wear
experiments, could assist in developing a predictive wear
model which would provide a powerful tool for advanced
internal mill design based on an enhanced cost-life benefit.

CADIA CASE STUDY


The SAG mill in the Cadia low-grade circuit of the Newcrest FIG 1 - Scanned view of mill rendered by MillMapper®.
Cadia Valley Operations (CVO) provides a prime test case for
this methodology, given the commitment of the site personnel From this full 3D data, MillMapper® analysis software was
to optimise the performance of this huge 40 ft diameter, used to provide average data for each segment of the mill
20 MW mill. During the study, the mill operated in closed liner. This could be transferred to 2D profiles for wear-loss
circuit with two MP 1000 recycle crushers. The work was modelling. Wear profiles for a single wear cycle are shown
aimed at improving mill control over the life of the liner, and in Figure 2.
is reported in some detail by Bird, Powell and Hilden (2011). Utilising the liner profiles, the Scanalyse software calculates
The current review expands upon that work. the total liner mass of all components in the mill. Individual
Liner wear measurement models of the wear of each major component are then derived
so that the total liner mass at any time can be calculated based
Through scanning the liner profiles in 3D with MillMapper®
on the reline dates and treated tonnage. Examples of wear of
software, the progression of the liner wear profile has been
the most substantial liner sections are shown in Figure 3. The
mapped over the liner life, along the full length of the mill and
progression of the lifter height, lifter angle and plate thickness
on the feed and discharge ends. This data provides the basis
are also derived and described by regression equations. These
for a wear rate prediction model. That model can be used to
are presented in detail by Bird, Powell and Hilden (2011).
forecast the mass of the liners at any given time. The wear
accelerates near the end of the liner life, so rather than four Effect of wear on mill operation
scans spaced evenly over the life of the liner, an extra scan is
Plant control data from the Cadia 40 ft mill, corresponding with
required near the end of the liner life. In this work an extra
the liner wear scan data, was analysed at 2.5 minute intervals
scan was taken three weeks prior to the liner change-out. The
over a period of 19 months, in order to quantify performance
Cadia site made the commitment to adding extra liner scans in
trends as the mill liners wear. Plant data was preprocessed
order to maximise the benefit from the liner wear monitoring
to filter out data from stoppages, unstable operations and
program. Accordingly, five sets of profiles were measured,
instrumentation faults, and to smooth out noise. Averages
which together with the profile of a new liner, provided six sets
were calculated for the key operating parameters over weekly
of profiles spanning the liner life, offering an outstanding set
intervals following the major relines.
of liner profile data for review (Bird, Powell and Hilden, 2011).
Mill speed is varied throughout the liner life to protect the
The Cadia low-grade (LG) mill contains alternating high
liners from excessive wear. The drive power draw varies in
and low lifters and mini-lifters (positioned between the main
response to the changes in mill speed, filling, ball filling and
lifters). Between 18 November 2009 and 9 June 2011, Scanalyse
slurry filling (Figure 4).
MillMapper® reports (covering the four most recent sets of
liner scans) were used to analyse the liner wear with respect While the mill is being operated at a reduced speed at the
to lifter height, angle and liner mass and as a function of the start and end of the liner life, the mill feed rate set point must
cumulative tonnes of ore milled. The Scanalyse mill scans also be lower as shown in Figure 5a.
generated a three-dimensional model of the mill internals. An There is a reduction in mill feed just after ten weeks,
example of the 3D rendered profiles is shown in Figure 1. corresponding to the change in grates and the resultant

FIG 2 - Examples of liner profiles reconstructed from the scan data (Bird, Powell and Hilden, 2011).

96 11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012


A MORE HOLISTIC VIEW OF MILL LINER MANAGEMENT

250 160
ShellLifters 140
OuterFeedEnd
200
120

150 100

Tonnes
Tonnes

80
100 60
y=Ͳ2.7927x2 Ͳ 0.8394x+237.15
R²=0.9747 40
50 HighShellLifters y=Ͳ11.657x+140.92
y=Ͳ1.4965x2 Ͳ 0.9201x+188.71 20 R²=0.9696
R²=0.948 LowShellLifters
0 0
0 2 4 6 8 0 2 4 6 8
MtMilled MtMilled

FIG 3 - Liner mass loss and fitted wear functions.

FIG 4 - Weekly average mill speed and power draw.

FIG 5 - (A) Weekly average run-of-mine and total semi-autogenous feed rate; (B) ratio of crusher feed to semi-autogenous feed.

decrease in production of pebbles. As the grates wear, the discharge rate, and the last week of the grate life corresponds
production of pebbles increases, as shown in Figure 5b, to peak pebble production.
dropping off dramatically as the grates are changed. The plot Grind size, expressed as a P80, is measured online using an
of the ratio of crusher feed to total SAG mill feed clearly Outotec PSI-200® particle size analyser. The average of the
illustrates the pebble production rate over the liner life. The grind size from the two cyclone clusters is shown in Figure 6.
period immediately following the reline of the grates (with The grind size is slightly finer when the mill is operated at a
the shell at mid liner-life) corresponds with the lowest pebble low throughput and low speed, and the average grind size

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 97


M S POWELL et al

FIG 6 - Weekly average cyclone P80 and calculated operating work index.

increases to approximately 200 μm for most of the liner life. that it is most unfavourable to mill performance to leave the
There is a marked decrease in grind size at the end of the liner for an extended period after it has been worn to a smooth
liner life. This should correspond with a shift from impact to profile.
abrasion breakage as the liner loses its profile and the charge At any time in the life of the liner, the total charge mass can be
experiences increasing slip on the liner. calculated from the load cell mass, corrected for the wear mass
The operating work index (OWi) is calculated from the lost since the mill reline. Using this data, it is possible to infer
online measured P80 and F80 from the Split-Online® feed size the total filling and proportion of balls in the charge, through
monitoring system, as well as the sum of the crusher, ball mill the application of a power model. A similar technique was
and SAG mill specific energies. Based on an assumption of employed by Kojovic, Pyecha and Corbin (2001) on the Red
a consistent average feed, a decrease in the OWi indicates Dog circuit, where the total load was calculated and the ball
improved operating efficiency; as a lower energy is required load was known with reasonable accuracy. The total and ball
to provide a given degree of reduction. The plot of Figure 6 filling values are inputs to the JKMRC power model developed
shows an improvement in comminution efficiency as the by Morrell (1993) which outputs the power-draw and charge
liner wears in from new to the replacement of the grates, mass. There is a unique ball and total filling which corresponds
with the OWi dropping from 27 to 25.2. The efficiency drops to a given charge mass and mill power draw. Thus, for an input
as the grate is changed, gradually improves, then steadily of the online load-cell reading (converted to charge mass) and
deteriorates over the last five weeks to a high of 28 kWh/t power draw values from the operating mill, the total and ball
when the liner is changed. This rapid increase at the end of filling values can be derived from the power model.
the liner life corresponds to the dramatic decrease in mill To test the above hypothesis, the historical control data for
throughput, from 2300 to 1700 t/h over the last five weeks. the Cadia 40 ft mill was analysed hourly over a 19-month
This considerable drop in throughput is not typical for the data period. For each time interval, the mill mass and
mill, but appears to be related to the liners being operated diameter were corrected for liner wear, giving the estimated
for a bit longer than usual in order to make the liner change- charge mass. The JKMRC power model was then calculated
out coincide with other plant down-time requirements. The iteratively to find the solution for total and ball filling at each
loss of liner profile is evident in the flat profile at change out, time interval. Figure 7 shows how the total load underwent a
illustrated by the 6.7 Mt line in Figure 2. The data indicates cyclic increase and decrease over the life of the liner.

FIG 7 - Load and estimated total and ball filling values based on the power model calculation.

98 11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012


A MORE HOLISTIC VIEW OF MILL LINER MANAGEMENT

Mill control based on liner wear the ball charge, correspondingly, drifting down. Overall, the
impact of inadequate knowledge of the mill filling costs the
With the ongoing liner profiles, the maximum trajectory
operation about ten per cent of throughput for about one
paths of the outermost balls and rocks can be calculated. The
third of the operating time.
MillTraj® program is used here, as it can run online with the
available control data; an approach not possible for discrete
element modelling (DEM) simulations. The design of the liner EXTENSION OF THE METHODOLOGY
and the operation of the mill need to ensure that there are no The results to date suggest that this technique may be usable
impacts above the toe region; otherwise the liners and balls as a mill-filling ‘soft-sensor’, permitting its use for process
will suffer severely accelerated wear and risk major breakage. control and optimisation. Additional refinements are being
The filling information presented in Figure 7 and the liner carried out to improve its robustness, so that this approach can
profile data are used as inputs to the simulation presented in be trialled as a real-time control tool in industrial applications.
Figure 8. The shifting toe position at the percentage of liner
life is labelled for each filling. The trajectories are given for
Slurry hold-up
the corresponding percentage of liner life. The mill speed is An estimate of the slurry holdup is required to give a more
selected to ensure impact in the middle of the toe region for accurate measure of the mass and power draw of the mill. For
each corresponding filling, using the lifter dimensions at that these simulations, slurry filling was assumed to be constant,
point in the liner life. corresponding to the filling measured during a crash-stop
inspection (a void fill fraction of 0.71). The variation in slurry
filling from 0.6 to one of the available mill charge voidage,
results in a 77 t variation in the mill load, with a corresponding
uncertainly in mill filling. This affects the power calculation
for estimating the ball load. The above variation in slurry
filling is equivalent to a ±2.2 percentage point variation in
total filling – which represents an undesirable uncertainty.
The effect on power draw can be dramatic if the mill goes into
a pooling state. A mill typically experiences a sudden slump
in power draw when a slurry pool develops inside the mill
(Morrell and Kojovic, 1996; Powell and Valery, 2006), with a
corresponding loss in grinding capacity – leading to reduced
throughput and/or a coarser grind. When utilising the power
model, if the influence of slurry pooling is not incorporated,
spurious results can be obtained. For the measured mill load,
FIG 8 - Charge trajectories limited by actual mill filings. the power draw is too low to resolve the ratio of balls to rocks.
These anomalous data can be easily edited out, but may also
For the historic progression in mill filling after installation of serve as a useful indicator of slurry transport issues. Based
a liner, the operating speed has to be limited to 72 per cent for on the assumption that the spurious power data was due to
the new liner; shifted up to 74 per cent after ten per cent of the slurry pooling, these events were plotted as a percentage of
liner life; then increased to 77 per cent after 20 per cent of the total operating time, as presented in Figure 9.
liner life. With an ability to confidently predict the mill filling
it can be controlled at a minimum of 30 per cent, allowing
the initial speed to be markedly increased to over 76 per
cent. The trajectory plots provide input on the safe operating
ranges for the mill. The MillTraj® program can be linked into the
operating data and wear model, to provide a live image of the
safe operating regime for any combination of mill speeds and
fillings.

Key outcomes
The total load and mill filling starts low after a liner change,
then steadily increase as the liner wears. Lacking any hard
data, the impact of the change in liner mass has not been
adequately accounted for, leading to a far lower mill filling
than was thought to be the case. This in turn necessitated a
decrease in mill speed, as the charge was impacting directly FIG 9 - Weekly average percent of slurry pool issue.
on the liners, as indicated by excessive mill noise (measured
by a single fixed microphone). Together these elements result There are many instances when the mill has a low filling
in a marked loss in power and throughput over the first six of 25 per cent after installation of the new liner. As the mill
weeks of operation with new liners, as shown in Figure 4. The fills these instances drop off to negligible values. They pick up
load again decreases when the grate is changed, as the change again after replacement of the grates and then at the end of the
in liner mass was not compensated for in the load set point. liner life. With a higher mill filling, the mill can hold a greater
Over the life of the liner, the data indicates a steady decrease in volume of slurry and has more of the grate area exposed to
the ball filling (from 17 to 13 per cent) corresponding with the the charge, so it can handle a higher volumetric slurry flow
mass loss due to liner wear, and an increase in the total filling prior to the onset of pooling. This appears to be the issue after
(from 26 to 35 per cent). Thus, in running on an unknown relining. The mill filling is again lower for a period after the
influence of liner wear, and without a measure of ball filling, grate change and there is a corresponding increase in pooling.
the selected load set-point results in the load creeping up and It is also possible that the absolute open area is also influencing

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 99


M S POWELL et al

the slurry transport, with the discharge area increasing as the In order to make the outputs of the grindcurve meaningful
grate wears. When crash-stopping the mill, pooling has not a good quality sample of the mill discharge is required. This
been observed, so this appears to be a transient state, driven presents quite a challenge for a solids flow rate of 2500 t/h
by fluctuations in feed composition, and more likely occurring and a slurry flow of about 1700 m3/h from a trommel
only at lower mill fillings. undersize which falls directly into a sump. A new technique
Having a model that better correlates slurry discharge was developed for the Cadia mill and the process is due for
capacity with grate condition would improve the predictive implementation over the next liner life.
capability of the filling model. The switch to a pooling
condition could also serve as an alarm to the operator to LINERS COST-LIFE BENEFIT
modify the feedrate, water addition and the amount of
Scanalyse, in collaboration with JKMRC and Newcrest
crushed recycle product, and to potentially switch to coarser Mining, conducted an extensive 12 month study on the effect
feeders to help release the pooling state of the mill. of liner wear on the mill performance of the 32 ft SAG mill
at Cadia Valley Operations, from July 2010 to June 2011. The
Power model methodology of determining the optimum liner shape was
The power model can utilise the slurry level information, but previously presented at Metplant 2011 (Toor et al, 2011a), and
this needs to be provided by an exterior model, as discussed expanded on at the SAG11 conference (Toor et al, 2011b, Toor
above. When the mill shifts into pooling the model cannot et al, 2012). The results presented indicated that liners with
cope in its current form. It does accommodate the reduction of a worn shape and reduced mass but shorter life, outperform
power for overflow mills, so can be modified to the gradually their longer life counterparts in all key production parameters
increasing pool depth, if that information is available. Since
of throughput, energy consumption and product size. This
the balls are added at a frequent and constant rate, it is known
paper quantifies the benefit of having shorter liner life cycles
that the ball load cannot fluctuate wildly. The use of a long-
for improved performance, by calculating the increase in
term moving average of ball load to smooth out the calculation
throughput and reduction in power, based on historical
of filling is being investigated as an enhancement.
data for this mill. It also presents the total overall gain by
Another improvement to the power model would be accounting for the increased effort required for extra relines.
to make an allowance for the change in power draw as This provides a holistic view of the net advantages when
the liner profile changes. The change in mill diameter is designing liners for performance and not specifically life span.
accommodated through the wear model, but the liner profile
information is not utilised. This requires an improvement to Figure 10 shows how throughput increases as the liner
the power model, based principally on the change in fraction wears, drops at half-life when the discharge grates (1/2 set)
of the cataracting charge. This enhancement is in the current and the feed outer liner are relined, then increase as the shell
research portfolio of the JKMRC. liners wear further, resulting in a 12 per cent increase in
throughput at the end of the liner life when compared to new.
Grindcurves related to liner profile It should be noted that the throughput post grate reline is
Process data provides a useful overview for correlating mill still approximately three per cent higher, on average, than the
performance with the liner profile. However, it is difficult throughput prior to the grate reline, indicating the increase in
to tease out the optimal operating condition that should be throughput is not only due to an increased open area of the
pursued as the liner profile changes with wear. Grindcurves, grates, but is also due to the wear in the shell liners. Aside from
as per the procedure of Powell, van der Westhuizen and throughput improvements over the shell liner life, specific
Mainza (2009); and Powell, Perkins and Mainza (2011); energy (kWh/t) decreases by a significant 12 per cent over the
provide a technique for rapidly assessing the ideal mill same period. Importantly, this decrease is observed, even with
filling for a given operating condition. This will vary as the an increase of mill speed to adjust for a reduction in lift due
liner wears, but the magnitude of this effect has not yet been to worn lifters. Finally, grinding surveys conducted during
investigated in a structured manner. the course of the project indicate that the SAG product is finer

FIG 10 - Throughput and specific energy as a function of shell liner life for Cadia 32 ft mill.

100 11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012
A MORE HOLISTIC VIEW OF MILL LINER MANAGEMENT

as the liner wears. Thus all available data suggests that the
three primary performance indicators of throughput, energy
consumption and product size, all improved considerably as
the liners wear in the 32 ft mill studied.

Designing liners for performance


The observations suggest that liners with a reduced thickness
and lower lifter profile outperform their larger counterparts.
This section highlights the potential benefits of having shorter
life liners, which deliver more tonnes at a lower power and
produce a finer grind. Moreover, it is envisaged that the
increase in downtime necessitated by more frequent relines
can be overcome by the increase in mill performance. FIG 11 - Effect of reline shutdown frequency and duration on throughput.
It is possible to predict the throughput of liners with reduced
mass by analysing existing plant data. Taking a half-worn Figure 11 illustrates that reducing the additional downtime
liner as the starting profile of a new liner, the throughput over increases the throughput benefit. As a half set of grates are
the life of this liner is equal to the average throughput from replaced every three months, only an additional 14 hour
the second half of the original liner life. Table 1 summarises relining time is required to replace the shell liner. Using this
the increase in mill performance with shorter liner life cycles extra 14 hours of reline time per six month cycle, a comparison
and additional relines. can be made between the two designs based on overall
TABLE 1 tonnes processed at the required power. The comparison is
summarised in Table 2. It shows that by using the proposed
Summary of improved performance with shorter liner life cycles.
design over a six month period, a throughput increase of
35 k/t is predicted (1.25 per cent) with a corresponding power
No of relines/ Throughput Power (kW) Specific energy
reduction of 365 MWh (1.37 per cent).
six month cycle (t/h) (kWh/t)
1 707 6 684 9.47 Cost benefit analysis
2 719 6 638 9.25 A broad estimate can be made of the potential economic
3 733 6 548 8.93 benefits of utilising liners designed for performance and not
simply life span. Table 3 presents a summary of the expected
4 738 6 505 8.81 benefits for the mill studied, based on an assumed mill
6 736 6 463 8.78 availability of 96 per cent.
8 736 6 463 8.78 To estimate the cost of relining, a Figure of $3/kg of steel
was used for the material cost, plus ten per cent on the
Using simplified examples of each additional reline taking material cost for labour costs. The volume of the new and half-
a possible 20, 30, 40, 50 or 60 hours, Figure 11 quantifies the life shell liners in the proposed design change was obtained
combined effect of shorter but more productive liner life cycles from MillMapper®. Table 4 summaries the estimated cost of
and the additional downtime their more frequent installation relining.
necessitates. As is evident from the graph, the current setup
of relining every six months is not optimal, regardless of how TABLE 3
long a reline takes within the range of typical durations. At Summary of expected benefits through using proposed shell liner design.
present the shell reline of the 32 ft mill takes approximately
30 to 40 hours (average 34 h). Interestingly, the throughput Gold Copper
curves for the 30 h and 40 h relines indicate that thinner
optimised liners, requiring relines three times in each six Current throughput (t/h) 707 707
month cycle (equating to a shell reline every two months) Annual production 6 620 kg 2 729 025 t
would deliver the highest tonnage. Annual Au production (oz) 212 854 N/A
However, if relines are taking approximately 60 hours,
Price A$1 600 /oz A$6 000 /t
there is little advantage to relining more frequently with
thinner liners. This highlights the need for efficient relining to Annual revenue A$340 570 000 A$16 374 000
ensure optimal mill performance, especially when using the Additional annual revenue (1.25 per cent increase) A$4 257 000 A$204 700
presented philosophy of thinner liners which require more
Total A$4.46 M
frequent replacement.

TABLE 2
Comparison of throughput and power for current and proposed shell liner design.

Current design Proposed design Proposed design (two Difference over six Difference (per cent)
(three months) sets over six months) months
Throughput (t/h) 707 719 719 12 1.70
Liner life (days) 165 82.2 164.4 0.6 0.35
Total throughput (kt) 2 801 1 418 2 836 35 1.25
Total power (MWh) 26 607 13 120 26 241 -365 -1.37

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 101
M S POWELL et al

TABLE 4 balls on the specific liner material. Although lacking absolute


Estimated cost of relining. predictive capability, these concepts can assist in modelling
the evolution of liner profiles with time.
Current shell Proposed shell The MillMapper® liner profile data can provide a drop-out
liners liners of the geometries in a form which can be directly imported
Liners volume (m3) 22.9 18.3 into the DEM software. Utilising the progressive profiles, a
wear model can be fitted, as presented by Weerasekara et al
Liners mass (kg) 178 620 142 740 (2011). The progression of the liner from new to full-worn
Cost of material @ $3/kg A$535 860 A$428 220 is illustrated in Figure 12, showing the excellent quality
Cost of labour@ 10% A$53 586 A$42 822 of the data. Even qualitatively, the DEM outputs paint an
informative picture of the life of a liner and demonstrate how
Total cost of relining A$589 446 A$471 042 wear rate is influenced by the liner profile. Once a wear model
Number of relining per year 2 4 is fitted with the required wear rate calibration, it can be used
Annual cost of relining A$1 178 892 A$1 884 168 to predict the wear of new liner designs.

Annual increase of relining cost A$0.705 M

From Tables 3 and 4 it can be seen that the estimated


increase in annual revenue from the proposed design change
is in the order of $4 M. It should also be noted that the new
design is expected to reduce energy consumption and/or
specific energy – further increasing the net revenue achieved.
Based on the survey data, it could also be reasoned that the
new design would also reduce the P80 of the product, thereby
increasing the flotation recovery. These two additional
benefits of reduced specific energy consumption and possible FIG 12 - Slices of liner profiles from new to fully-worn in EDEM.
increase in recovery, add further commercial weight to the
proposed liner design with reduced mass. A good wear model can then be applied to modify the liner
design and to test the outcomes prior to manufacturing a liner
There is clearly further opportunity to optimise the liner with a new profile. Currently the accuracy of the predictions
profile if a shorter liner life is targeted. Optimisation would is insufficient for confident liner design, but this research is
aim at minimising any wasted shell plate thickness evident at ongoing.
the end of the liner life, plus changing lifter width to reduce
metal wastage there. The excessive width is a function of the Laboratory testing of liner wear
structure required for the very high new lifter bar and is not As noted in the DEM wear modelling section, the models
required for a lower lifter design. Improvements along these have to be calibrated to production data. It would be most
lines would further decrease the liner cost, and lighter liners useful to have a test that could calibrate the wear constant for
are faster to replace, with the added potential of decreasing a given ore and liner combination.
the number of liner elements in line with the capacity of the
In general the factors that affect liner wear can be categorised
liner handler.
into operating parameters and material characteristics
(Rezaeizadeh et al, 2010a). Various approaches have been
PREDICTIVE LINER WEAR MODEL proposed to predict the lifter wear (Radziszewski, 1997;
A predictive wear model for liners would be most useful in Cleary, 1998; Kalala, Bwalya and Moys, 2005a; Teeri et al, 2006;
optimising liner design. This is not merely for liner life cycle Kalala, Breetzke and Moys, 2008). These approaches provide
prediction, but more importantly for providing an assessment an insight into liner wear, but a full relationship between
of mill performance over the life of the liner. Although the mill operating parameters, lifter profile and materials, with
status of DEM mill modelling is inadequate at providing fully wear rate is missing. Laboratory testing of liner wear could
predictive simulations of mill performance, it can supply provide reliable data for developing a predictive wear model
qualitative comparisons of performance related to liner in conjunction with DEM simulation results. Using liner
design. Combining DEM outputs, trajectory analyses, power wear tests makes it possible to compare the wear of different
modelling and historic data can provide a useful mechanism compositions of materials as well as the wear of various liner
for estimating the impact of progressive liner profile on designs.
mill performance. Thus combining a capability to assess the
A pilot test is being trialled to allow the rapid comparison
progression of the liner profile over time with performance
data will provide a powerful tool for improved liner design of multiple materials in near field conditions. The test method
and selection. based on methodology proposed by Powell and Cornelius
(1992) and trialled at the JKMRC by Rezaeizadeh et al (2010a),
Discrete element method wear modelling has been adapted to suit different lifter and backing plate
Some work has been conducted on utilising the DEM outputs designs and materials. The pilot mill and sample mounting
of contact force and frequency to model the wear of mill are illustrated in Figure 13.
liners, Cleary (1998); Kalala, Bwalya and Moys (2005a, 2005b);
Powell et al (2006); McBride and Powell (2006); Makokha Utilising a liner wear model
and Moys (2006); Rezaeizadeh et al (2010b); each of which To develop a predictive wear model, laboratory wear tests
contributes to the predictive wear modelling of liners. These can provide experimental data, while DEM simulations offer
models all require calibration to industrial data to fit the a detailed view of forces acting on the liner surface. This could
wear rate as a function of the contacting materials – ore and be done through linking the results of experimental tests with

102 11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012
A MORE HOLISTIC VIEW OF MILL LINER MANAGEMENT

been used to identify opportunities to improve the mill control;


particularly as it relates to managing changes in performance
over the liner life. Quantifying the considerable impact that
changes in liner mass and profile (over liner life) have on
the total mass of charge and other operating parameters, can
aid in the development of a strategy for improved real-time
process control and optimisation.
Since grinding performance is strongly integrated with
liner design, monitoring process data over the liner life
provides a useful overview of the correlation between mill
performance and the liner profile. This approach can lead to
the development of liner designs with an optimum liner mass
and life span, thereby improving the grinding performance
and enhancing the cost-life benefit of liners. A possible profit-
based approach is presented for optimising liner design and
changing relining intervals to maintain mill productivity.
Combining the outputs of careful liner wear monitoring,
FIG 13 - Pilot mill and lifter bar for testing sample blocks. trajectory analysis, predictive wear modelling, power
modelling, DEM simulations, laboratory wear testing
the outputs of DEM simulations. The collision energy spectra
and a new relining methodology can provide a powerful
and collision intensity distribution on the surface of liners
management system for predicting the impact of liner design
can be derived from DEM simulations. A semi theoretical
on mill performance. It has been demonstrated that annual
model can then be applied to correlate the DEM simulations
production improvements of three to five per cent can be
with measured wear rates in laboratory tests. The predictive achieved through: eliminating inappropriate drift in mill
wear model provides a means for optimising liner design. filling resulting from liner wear; minimising step changes in
A combination of the predictive wear model with grinding filling with each liner replacement; improving control of the
models can potentially be applied in the liner design stage to ball load; maintaining tighter and more consistent control;
balance the liner life and mill performance over life of the liner. and improving mill filling set-points.

FURTHER AREAS TO BE INCORPORATED ACKNOWLEDGEMENTS


More aspects can be brought to bear on mill lining economics The authors would like to acknowledge the project funding
and mill performance over life of liner. A few that show and permission to publish provided by Newcrest Cadia Valley
potential are outlined here. Operations. The help of site personnel and the metallurgy
team in collecting data is greatly appreciated.
Interaction with relining methodology
Any change in liner design should be coordinated with the REFERENCES
entire relining process, not just in terms of ensuring that Bird, M, Powell, M S and Hilden, M, 2011. Adapting mill control
the liner handler can cope with the liner segments, but in to account for liner wear on the Cadia 40ft mill, in Proceedings
optimising the relining time. If, as has been alluded to in the International Autogenous and Semiautogenous Grinding Technology
liner cost benefit section, it pays to have a shorter liner life 2011, pp 1-9, Vancouver.
with more frequent relines, then the efficiency of the relining Cilliers, J J, Austin, L G, Leger, P and Deneys, A, 1994. A method
process becomes even more crucial. Taking advantage of of investigating rod motion in a laboratory rod mill, Minerals
advanced relining equipment to leverage these potential Engineering, 7(5/6):533-549.
benefits was presented by Russell (2006). Using thinner liners Cleary, P W, 1998. Predicting charge motion, power draw, segregation,
allows parts to be made larger, thus decreasing the number of wear and particle breakage in ball mills using discrete element
components to be installed during a reline. The concept was methods, Minerals Engineering, 11:1061-1080.
expanded on by Rubie (2011), to incorporate a sophisticated
Cleary, P W, 2001. Charge behaviour and power consumption in ball
time and motion analysis of the relining procedure. mills: Sensitivity to mill operation conditions, liner geometry and
charge composition, International Journal of Mineral Processing,
Discharge modelling 63:79-114.
Rather than considering the grates and ports to be a separate Djordjevic, N, Shi, F N and Morrison, R, 2004. Determination of lifter
issue, it is proposed that the discharge grate be optimised design, speed and filling effects in AG mills by 3D DEM, Minerals
as an integral unit with the shell liner. The data on mill Engineering, 17:1135-1142.
operation clearly indicates the strong influence that the grate
Kalala, J T, Bwalya, M M and Moys, M H, 2005a. Discrete element
has on mill performance (Figure 4). It is proposed that an method (DEM) modelling of evolving mill liner profiles due to
integrated understanding of the influence of liners and grates wear, Part I: DEM validation, Minerals Engineering, 18:1386-1391.
would provide performance benefits. They could be set up
Kalala, J T, Bwalya, M and Moys, M H, 2005b. Discrete element method
to compensate for the change in mill performance that each
(DEM) modelling of evolving liner profile due to wear, Part II:
induces with wear. Clearly good discharge and grinding Industrial case study, Minerals Engineering, 18:1392-1397.
models are required to seriously consider this approach.
These models are under development, so their integration Kalala, T J, Breetzke, M and Moys, M H, 2008. Study of the influence of
liner wear on the load behaviour of an industrial dry tumbling mill
into liner design has yet to be investigated.
using the discrete element method (DEM), International Journal of
Mineral Processing, 86:33-39.
CONCLUSIONS Kiangi, K K and Moys, M H, 2006. Measurement of the load behaviour
Integrated liner wear monitoring, along with accurate in a dry pilot mill using an inductive proximity probe, Minerals
calibration of mill operating parameters with liner life, has Engineering, 19:1348-1356.

11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012 103
M S POWELL et al

Kojovic, T, Pyecha, J R and Corbin, J R, 2001. Use of online mill Powell, M S, van der Westhuizen, A P and Mainza, A N, 2009.
charge simulators at the Red Dog grinding circuit, in Proceedings Applying grindcurves to mill operation and optimisation, Minerals
International Autogenous and Semiautogenous Grinding Technology, Engineering, 22:625-632.
pp 11-23.
Powell, M S, Weerasekara, N S, Cole, S, LaRoche, R D and Farier, J,
Makokha, A B and Moys, M H, 2006. Towards optimising ball milling 2011. DEM modelling of liner evolution and its influence on
capacity: Effect of lifter design, Minerals Engineering, 19:1439-1445. grinding rate in ball mills, Minerals Engineering, 24(3-4):341-351.
Makokha, A B, Moys, M H, Bwalya, M M and Ki, K, 2007. A new Radziszewski, P, 1997. Ball charge dynamics and liner wear simulation,
approach to optimising the life and performance of worn liners in in Proceedings 29th Annual Meeting of CMP, pp 91-107, Ottawa,
ball mills: Experimental study and DEM simulation, International Canada.
Journal of Mineral Processing, 84:221-227.
Rajamani, R K, Joshi, A D and Mishra, B K, 2002. Simulation of
McBride, A T and Powell, M S, 2006. A structured approach to modelling industrial SAG mill charge motion in 3D space, SME-AIME Annual
SAG mill liner wear – Numerical modeling of liner evolution, in Meeting, Phoenix, Arizona (SME).
Proceedings International Autogenous and Semiautogenous Grinding
Rezaeizadeh, M, Fooladi, M, Powell, M S, Mansouri, S H and
Technology, p 120-132, Vancouver.
Weerasekara, N S, 2010b. A new predictive model of lifter bar wear
McIvor, R E, 1983. Effect of speed and liner configuration on ball mill in mills, Minerals Engineering, 23:1174-1181.
performance, Mining Engineering, pp 617-622.
Rezaeizadeh, M, Fooladi, M, Powell, M S and Weerasekara, N S, 2010a.
Mishra, B K and Rajamani, R K, 1993. Numerical simulation of charge An experimental investigation of the effect of operating parameters
motion in ball mills – Lifter bar effect, Minerals and Metallurgical on the wear of lifters in tumbling mills, Minerals Engineering,
Processing, May:86-90. 23:558-562.
Morrell, S, 1993. The prediction of power draw in wet tumbling Rubie, P, 2011. Simulation as a tool to enable world’s best mill relining
mills, PHD thesis (unpublished), The University of Queensland, practice, in Proceedings International Autogenous and Semiautogenous
Brisbane. Grinding Technology, pp 1-10.
Morrell, S and Kojovic, T, 1996. The influence of slurry transport on Russell, J, 2006. Advanced grinding mill relining for process
the power draw of autogenous and semi-autogenous mills, in metallurgists and management, in Proceedings International
Proceedings International Autogenous and Semiautogenous Grinding Autogenous and Semiautogenous Grinding Technology, III:11-22.
Technology, pp 378-389, Vancouver.
Teeri, T, Kuokkala, V T, Siitonen, P, Reponen, P K and Liimatainen,
Moys, M H, 1993. A model of mill power as affected by mill speed, load J, 2006. Impact wear in mineral crushing, Science and Engineering,
volume, and liner design, Journal of the Southern African Mining and 12:408-418.
Metallurgy, 93(6):135-141.
Toor, P, Franke, J, Powell, M S, Bird, M and Waters, T, 2012. Designing
Powell, M S, 1991. The effect of liner design on the motion of the outer liners for performance not life, in Proceedings 8th International
grinding elements in a rotary mill, International Journal of Mineral Comminution Symposium.
Processing, 31:163-193.
Toor, P, Franke, J, Powell, M S and Perkins, T, 2011b. Quantifying
Powell, M S and Cornelius, F S, 1992. A test for the realistic simulation the influence of liner shape and mill filling for performance
of coarse abrasive wear, Prep Tribology (SAIT). optimization, in Proceedings International Autogenous and
Powell, M S and Nurick, G N, 1996a. A study of charge motion in Semiautogenous Grinding Technology, pp 1-9.
rotary mills, part 1 – Extension of the theory, Minerals Engineering, Toor, P, Perkins, T, Powell, M and Franke, J, 2011a. The influence of
9(2):259-268. liner wear on milling efficiency, in Proceedings Metplant 2011,
Powell, M S and Nurick, G N, 1996b. A study of charge motion in pp 193-212 (The Australasian Institute of Mining and Metallurgy:
rotary mills, part 2 – Experimental work, Minerals Engineering, Melbourne).
9(3):343-350. Van Nierop, M A, Glover, G, Hinde, A L and Moys, M H, 2001. A
Powell, M S and Nurick, G N, 1996c. A study of charge motion in discrete element method investigation of the charge motion and
rotary mills, part 3 – Analysis of results, Minerals Engineering, power draw of an experimental two-dimensional mill, International
9(4):pp 399-418. Journal of Mineral Processing, 61:77-92.
Powell, M S, Perkins, T and Mainza, A N, 2011. Grindcurves applied to Vermeulen, L A, 1985. The lifting action of lifter bars in rotary mill,
a range of SAG and AG mills, in Proceedings International Autogenous Journal of the Southern African Institute of Mining and Metallurgy,
and Semiautogenous Grinding Technology, p 1-20. 85:51-63.
Powell, M S, Smit, I, Rdziszewski, P, Cleary, P, Rattray, B, Eriksson, Weerasekara, N, Powell, M S, Franke, J and Favier, J, 2011. The
K and Schaeffer, L, 2006. The selection and design of mill liners, evolution of integrated mill liner profile and performance through
in Advances in Comminution, pp 331-376 (Society for Mining, 3D DEM modelling simulation as a tool to enable world’s best
Metallurgy, and Exploration, Inc: Colorado). mill relining practice, in Proceedings International Autogenous and
Semiautogenous Grinding Technology, p 1-15.
Powell, M S and Valery, W, 2006. Slurry pooling and transport
issues in SAG mills, in Proceedings International Autogenous and Yahyaei, M and Banisi, S, 2010. Spreadsheet-based modeling of
Semiautogenous Grinding Technology, p 133-152, Vancouver. liner wear impact on charge motion in tumbling mills, Minerals
Engineering, 23:1213-1219.

104 11TH MILL OPERATORS’ CONFERENCE 2012 / HOBART, TAS, 29 - 31 OCTOBER 2012

View publication stats

You might also like