You are on page 1of 13

PAPER 8

Horses for Courses – Tailoring Front End Design to Project


Requirements
Paul Staples, Manager Development1
Greg Lane, General Manager Technical Solutions2
Paul Messenger, Manager Studies and Process3
1
Ausenco Ltd
Level 2 44 St Georges Tce
Perth, WA 6000
Australia
PH: 61 (08) 9223 1911
E-mail: paul.staples@ausenco.com
2
Ausenco Ltd
8/2404 Logan Road
Eight Mile Plains, Brisbane, QLD 4114
PH: 61 (07) 3112 8221
E-mail: greg.lane@ausenco.com
3
Ausenco Canada Ltd
Suite 605
375 Water St
Vancouver BC V6B 5C6
PH: 1 (604) 453-4800
E-mail: paul.messenger@ausenco.com

Key Words: Grinding circuits, rod mills, milling in acid, HPGR

40th Annual Meeting of the January 22 to 24, 2008


Canadian Mineral Processors Ottawa, Ontario, Canada

117
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

ABSTRACT

Every project is different. While “conventional” SAG and ball mill circuits will suit many projects,
for others there are alternatives, which may produce improved outcomes. This paper examines
particular examples of recent front end designs which either went against conventional thinking or
which employed “out of the box” thinking to realize the desired results.

One such example is the design strategy employed for a small capacity gold plant in Victoria,
Australia. The selection of HPGR technology for small capacity projects is not generally favoured,
given the relatively high cost of small HPGR machines. However, the unusual size distribution of
the gold and its liberation characteristics were expected to lead to improved gold recovery offsetting
the higher capital cost.

Another example is the first large scale application of milling in acid for copper recovery, at a
project in south east Asia. The stainless steel, single stage SAG mill using ceramic grinding media
was more expensive than a mill constructed of conventional materials. However, the steps of
scrubbing, slurrying, grinding and leaching could all be achieved in a single unit operation and the
need for thickening and repulping with acid after milling was eliminated.

Rod mills were selected for both primary and secondary milling duties for another south east Asian
project despite their somewhat outdated reputation. A review of previous industry experience
confirmed that rod mills could outperform conventional rod/ball or SAG/ball circuits in cases where
the size distribution of the ground product is critical to downstream recovery.

These examples are discussed with particular focus on the key drivers in each case.

INTRODUCTION

The dominant comminution circuit in the international minerals processing industry is comprised
of a primary crusher followed by crushed ore stockpile, reclaim system and grinding by semi-
autogenous (SAG) and ball mills. Circuits comprising multistage crushing, rod milling and ball
milling have moved from being a “traditional” approach to, in most cases, a “historic” approach
to size reduction, driven in the main by the lower capital and operating costs of the SAG mill
based circuit.

However, the SAG (or AG) mill based circuit is not always the most technically suitable
outcome, and does not necessarily provide the best commercial outcome for a project. This paper
discusses some examples of recent projects undertaken by Ausenco where the conventional
approach to grinding circuit design did not provide the optimum outcome and either new or old
technology provided better downstream metallurgical response, improved project schedule or
lower costs. For each project, the critical drivers for process route selection are unique. The
drivers typically consist of:
• Ore characteristics
• Project capacity

118
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

• Mine design
• Metallurgical performance as a function of grinding circuit product size
• Metal prices
• Consumable costs
• Labour costs for operations and maintenance
• Operating and maintenance material costs
• Capital costs
• Trends in circuit design (“me too” syndrome)
• Site layout considerations
• Environmental considerations

The following generalisations and “rules of thumb” assist us in circuit selection:


• SAG mill based circuits generally have lower capital costs.
• Multistage crushing circuits followed by rod and/or ball mills tend to be more
comminution energy efficient when milling competent ores, but consideration of the
ancillary equipment power consumption for dust control, conveyors and screens reduces
the difference.
• Autogenous circuits can have a low operating cost, provided that the ore characteristics
are suitable, reasonably consistent and controllable. They also tend to have the highest
capital cost and be the most difficult to operate.
• Some circuits containing rod mills have been documented to produce a sharper size
distribution, less fines for a given P80, and thus can benefit metallurgical response when
fines are difficult to recover.
• The comminution component of multistage crushing circuits that incorporate high
pressure grinding rolls (HPGR) are more energy efficient than the comminution
component of SAG mill based circuits, but consideration of the auxiliary drives for these
circuits reduces the margin.

Assessing the process risk / benefit relationship is typically the most difficult issue and assessors
bring to the table their experiences, biases, and often expected outcomes. Consequently, the
eventual outcome is the subject of qualitative and semi-quantitative assessments. Often the
margin between various circuit options is less than the error margins involved in the assessment
and “gut feel” plays a large role in process selection.

The following sections illustrate some examples of projects where the project drivers led to the
selection of relatively unusual circuits or equipment, including:
• Using a HPGR to prepare gravity circuit feed
• Grinding in an acidic environment
• Using a two stage rod mill circuit

119
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

APPLICATION OF HPGR FOR GOLD LIBERATION

General

The gold mineralization for this particular property is characterised by association with quartz
veining and faults in ribbon shaped reef clusters. The gold is nuggety and coarse, of the order of
0.5 mm to 5 mm in size (Figure 1) (McLean, 2007 and Gardula, 2006).

Figures 1 and 2; Coarse Gold Mineralisation and Gold from Circuit Gravity Traps
(McLean, 2007 and Gardula, 2006)

Bench scale and extensive pilot plant testwork programs were conducted as part of the feasibility
study. The main conclusions of the testwork were that the gold grades were from 8 to 14 g/t, a
range consistent with the typical head grades predicted for the first few years of operation, and
that gravity recovery was key to achieving acceptable gold recoveries.

The ore was moderately hard and also had a moderate abrasion index. It was shown that
liberation of the gold particles occurred at particle sizes of 1 to 2 mm and that the majority of the
gold particles were in this size range and above.

Overall, the ores tested are notable for their relatively high grade, good response to gravity
recovery and coarse liberation size. Otherwise the ores were metallurgically clean, with only
minor amounts of arsenic.

Discussion

The principal objective of the plant design was to maximize gold recovery in a cost effective
manner. Maximising gravity gold recovery was expected to maximise overall gold recovery as
coarse, well liberated gold, not recovered by the gravity system, would go to the milling and
flotation circuit where gold losses to tailings are more likely. This meant that fine crushing was

120
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

required to feed the gravity circuit. The testwork indicated that gravity recovery was related to
the percent passing 4 mm in the feed.

The options examined were to:


1. Tertiary crush to minus 10 mm followed by ball milling prior to gravity
2. Four stage crush to minus 4 mm with conventional crushers
3. Use an HPGR to tertiary crush to minus 4mm
4. Use alternative high compression crushers

Tertiary crushing to nominally minus 10 mm would not achieve sufficient liberation and gravity
separation would need to be included as part of the ball milling circuit. The main concern with
this circuit was that optimum plant metallurgical performance would not be achieved due to
“overgrinding” and flattening or smearing of coarse gold grains in the ball mill. Closed circuit
ball milling results in a large lock-up of gold value in the milling circuit and also leads to less
efficient gravity and classification processes.

A four stage crushing circuit would produce a particle size distribution closer to the optimum for
gold liberation and feed to the gravity circuit, but with a complex flowsheet and large plant
footprint. The latter was a critical concern, because of a limited amount of space available for the
plant site. Circuits of this type which treat similar high quartz ores also tend to experience high
wear and result in high maintenance and operating costs to maintain acceptable availability (Barr
and Acton, 1998).

A HPGR unit was considered for the tertiary crushing stage following conventional two stage
crushing (Lane, 2005, and Pyke, 2006). Batch testing was conducted on 1300 kg of ore to obtain
power performance measures and wear information. The testwork showed that 10-15% more
minus 2 mm product could be obtained using an open circuit HPGR than the four stage crushing
option and even more if a closed circuit option was selected. Alternate high pressure crushers
were considered but this option was not pursued due to concerns with equipment suitability.

A series of comparative case studies was carried out and it was determined that the HPGR option
had a higher capital cost than the four stage crushing option and a 1% increase in gold recovery
was required to achieve payback in 3 years, based on a $400/oz gold price.

Small HPGR units are capital intensive, which often precludes their use on small tonnage
operations. However, the specific requirement to crush to less than 4 mm in this application
made competing technologies equally capital or maintenance intensive.

Outcome

Once commissioned, the circuit met gravity gold recovery expectations and gold liberation was
excellent. Issues arose with the wear rate of the HPGR wear surfaces due to wet high silica ore,
but these were being addressed prior to the plant shutting down for other reasons. The plant is
currently on care and maintenance while further resources and reserve development is completed.

121
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

Figure 3: HPGR Schematic Flowsheet for Enhanced Gravity Gold Recovery

Other process outcomes were that the HPGR single pass product size was < 4 mm, energy
consumption was as predicted by the testwork and flake production was good (Gardula, 2006).

Figure 4: HPGR Product (Gardula, 2006)

122
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

MILLING IN ACID

General

The Sepon Copper plant in Laos is designed to produce 60,000 tpa of cathode copper from a
predominantly chalcocite ore-body (Weeks, 2006).

Run of mine ore is crushed and milled in acidic process liquor. Copper is leached from the
milled ore in a series of agitated tanks. Residue from the atmospheric leach is thickened and the
pregnant leach solution (PLS) that overflows from the thickener is processed through solvent
extraction (SX) to recover copper. Copper is electrowon from solvent extraction electrolyte to
produce LME Grade A copper cathodes.

The solid residue stream is washed in a counter current decantation (CCD) circuit and pyrite is
floated from the residue, prior to disposal to a tailings storage facility. The pyrite concentrate
undergoes pressure oxidation to produce sulphuric acid and iron, needed for copper leaching.

The ore for this project is very wet and contains large amounts of highly weathered clays. Just
over half the logged core had a clay content between 40 and 60%, Another quarter had clay
content between 60 and 80% (McLean, 2005). The copper values are contained in chalcocite
clays and various copper carbonates and oxides. Overall these are very amenable to high copper
recovery by acid leaching.

Discussion

Issues that were considered while designing the plant included the difficult material handling
characteristics of the clays as they were wet and quite sticky. This called for a simple front end
design.

During the feasibility stage, testwork done on the scrubbing characteristics of the material was
promising, leading to consideration of a unique grinding circuit which operates in a highly acidic
environment and combines the processes of scrubbing, grinding and leaching into a single unit
operation. By direct feeding the single stage SAG mill from a mineral sizer with no stockpile, the
solution also avoided any more handling than necessary.

The proposed circuit introduced a number of difficult design and operational issues. The mill
would be fed highly acidic rafinate to slurry and dilute the feed to the grinding circuit. This
meant that not only the entire mill but also the grinding media itself would have to be specified
to accommodate the acidic environment (pH <1).

The power modelling and sizing of the mill were completed using conventional comminution
measures such as the Bond work indices and an allowance was made for the scrubbing energy
required. The ores typically were of low competency with Bond ball mill work indices in the
range of 6-8 kWh/t. Some harder components exist with work indices of 10-11 kWh/t and above.

123
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

The mill selected was a shell supported design, which allowed more flexibility for material
selection, a better overall design for acid protection and reduced stresses on the shell. The shell
is two piece with integral welded flat heads to reduce bolted joints and has two bearing riding
rings on the circumferential flanges. The ring gear is bolted to the shell at the centre flange. The
bearings are located in such a position as to limit the risk of spillage and leaks from the feed
chute.

The shell and heads are fabricated from mild steel, with a 6 mm SAF 2205 grade stainless steel
liner welded to the internal surface for protection from the acid environment in the mill. Another
6 mm liner made of rubber sits over the stainless steel and protects it from liner damage. The bolt
holes are also all lined with 6 mm stainless sleeves which are welded to the liner. The mill liners
(plates and lifters) are rubber and are fixed by 316 stainless bolts. The trommel screen, drop box
and feed chute are all also fabricated in 2205 stainless steel.

Other design aspects specific to this mill duty were incorporated including mill discharge acidic
fume extraction systems, feed chute and mill discharge sealing systems and ceramic feed chute
lining system.

Figure 5: Acid Grinding at Sepon (McLean, 2005)

The media selection process was a difficult one. Initially, tests were undertaken to assess the
competency of ceramic balls and flint pebbles in terms of both breakage and ability to retain
shape. These were also tested for their acid resistance and corrosion rates. The breakage rate of

124
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

the ceramic media was far superior to the flint pebbles while the corrosion rate was similar for
both media.

Ceramic media are relatively expensive, the consumption rates were higher than expected during
start-up and their low SG impacted on grinding performance. Improvements to the mill control
and crushing systems later significantly reduced media consumption due to breakage, but trials
ultimately led the operation to use high chrome steel media.

Figure 6: Ceramic Media in the Sepon Mill Prior to Start Up (McLean, 2005)

Outcome

Following a typical path of optimization, the mill is currently exceeding design throughput with
no issues relating to acid damage.

Ausenco is currently involved with another project where a similar milling in acid philosophy
will be employed. This time an acid and impact resistant epoxy coating will be used.

ROD MILLS FOR NUI PHAO

Introduction

Stephenson (2007) discusses the two stage rod milling circuit proposed for a planned

125
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

polymetallic project in south east Asia. The main driver for the selection of a rod / rod circuit
was its ability to deliver a narrow product size distribution. Data on primary rod mills followed
by secondary rod mills is relatively scarce and it could be further stated that use of rod mills is
nowadays viewed by some as being outdated.

Rod mills have been around since the early 20th century and for many years were a mainstay of
many milling operations. Most commonly, they were used to prepare crushed feed for ball
milling, but their size was limited by the length of rods that could be manufactured and safely
operated. Mill design is constrained by rod length and efforts to minimise the chances of rod
tangling. The largest rod mills are approximately 6 m in length, have an L/D ratio of around 1.4
and operate at about 68% critical speed.

The mineralization for this deposit consists of tungsten, gold, bismuth, copper and fluorine,
occurring as scheelite, wolframite, native gold, chalcopyrite, native bismuth, bismuthinite and
fluorite. The scheelite liberates well below 200 microns but is friable and prone to overgrinding.
The ores appear to be of medium competency and have rod mill work indices ranging from 10 to
14 kWh/t.

Discussion

The primary grinding objective for this project is to liberate scheelite and fluorite and minimize
over-grinding, in order to maximize gravity and flotation recovery. The difference in grinding
circuit product size distribution based on grinding circuit flowsheet has been examined in
numerous studies. In those studies typically the focus has been on circuits such as:
• HPGR / ball mill circuits
• Stage crush / rod / ball mill circuits
• SAG and ball mill circuits

Despite the established metallurgical observation that open circuit rod milling produces a
narrower size distribution than open circuit ball milling, the conclusion of these studies was that
for most circuits with a closed circuit ball mill as the final grinding stage, product size
distributions are more or less equal. However, a comparison of rod/ball and rod/rod size
distributions from the Canada Tungsten Mining Corporation Cantung mine indicated that the
rod/rod circuit produced significantly less fines by mass and tungsten distribution (Bolu, 1984).
This project was the only one of its type found during the study and highlights a significant
advantage of a rod/rod circuit over a rod/ball circuit in terms of product size distribution
(Figure 7).

The Cantung circuit was modelled to determine the appropriate efficiency factors to apply to the
design of a two stage rod milling circuit. The results of this analysis supported some anecdotal
information, which indicated that the conventional rod milling efficiency factor EF6 (ratio of
reduction) used in the standard Bond-Rowland power calculation did not apply to the Cantung
rod mills.

126
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

100
Cum. % Passing

Rod/Ball - WO3

"Bond linear"
Rod/Rod - WO3
Rod/Ball - Mass
Rod/Rod - Mass

10
10 100 1000
Size (micron)

Figure 7: Size Distribution Data from Cantung

The power modeling showed this factor was close to unity. This is supported by Barrett who
reports that the Cantung primary rod mill power requirements and product size distribution are
exactly as would be predicted from standard calculations provided the high EF6 factor is ignored
(Barrett, 1982). Rowland also has stated that for ores with low work indices or which have
natural breakage characteristics that produce more fines, the ratio of reduction factor does not
always apply (Rowland, 1985).

Outcome

Final mill selections were three 4.57 m diameter by 6.4 m long mills (one primary and two
secondary) with installed motor power of 1,700 kW. The selected rod mill work index for design
was 13.1 kWh/t, slightly above the average of 12.3 kWh/t tested, but which is expected to
represent the hardest blend of ore that will typically be fed to the circuit (Stephenson, 2007).

Ultimately, selection of the mills was made based on input from the project owner, various
consultants, mill suppliers and the power analysis described here. Achievement of the design
throughput and product grind size for the project will depend on a number of factors including
the ability to provide the circuit with continuous feed of the required size, efficient operation of
the circuit in terms of mill densities, rod sizes, classification efficiency, screen availability and
circulating load, as well as on the availability of reasonable design contingency in the case that
lower milling efficiencies are encountered in either the primary or secondary circuit.

Operability of the circuit will also be critical to its success or failure. Adoption of the 70%
critical speed should mitigate to some degree, the risk of rod tangling with such large and long
rod mills.

127
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

CONCLUSIONS

The three examples discussed above demonstrate how the application of unusual comminution
circuits can provide benefits to projects in terms of improved downstream metallurgical
performance or more cost effective and efficient design.
1. In the first example, gravity gold recovery, and consequently overall gold recovery, is
enhanced by selecting equipment to reduce the ore to the optimum particle size for gold
liberation, rather than selecting equipment for the lowest capital cost.
2. The use of appropriate materials to facilitate milling in acid can permit simultaneous
scrubbing, grinding and leaching in a single unit operation.
3. In the third example, two stage rod milling is selected to provide close control of product
size distribution in order to maximise recovery.

Every mineral project is different. The following factors all contribute to the individuality of
every project and influence the requirements of a comminution circuit:
• Ore characteristics
• Project capacity
• Mine design
• Metallurgical performance as a function of grinding circuit product size
• Metal prices at the time
• Costs of delivered consumables
• Labour costs for operations and maintenance
• Power and other utility costs
• Capital costs
• Site layout considerations
• Environmental considerations

These factors and often the personal experiences of members of the design team all lead to the
final selection of a comminution circuit for a project.

It is important to identify alternative comminution circuits that are likely to achieve the desired
capacity and size reduction, rather than focus only on “conventional” circuits throughout project
development. All the factors that influence the requirements of the comminution circuit must
then be recognised and applied to the selection process in order to arrive at the optimum circuit
for a particular project.

A “conventional” circuit may or may not be the optimum comminution circuit for a given
project; it is “Horses for Courses.”

128
Proceedings of the 40th Annual Canadian Mineral Processors Conference – 2008

REFERENCES

Barr, S., Acton, T., 1998, Heap Leaching at Newmont’s Muruntau Operation, Mining Engineering,
Jan.

Barrett, D., “Factors Which Influence the Selection of Comminution Circuits”, Design and
Installation of Comminution Circuits, Eds. Mular, A. and Jergensen, G. (II), SME-AIME, 1982

Bolu, M., Bouma, P., and Paterson, J., “Rod Mill Grinding for Improved Gravity Recovery at the
Canada Tungsten Scheelite Concentrator” SME-AIME Annual Meeting, Feb 1984

Gardula, A., Presentation for Application of HPGR in Processing Gold Ore in Australia, SAG 2006
Conference, University of British Columbia, Vancouver, September.

Johansen, G., English. D., Lane, G., Hayward, T., Gardula, A., Bendigo Gold Mine Opts to
Maximise Coarse Gold Recovery, AusIMM Bulletin, March/April, 2005

Lane, G., 2005, High Pressure Grinding Rolls at Bendigo, Randol Innovative Metallurgy Forum,
August 2005, Perth, Australia

Lane, G., 2007, Some Observations Regarding SAG Milling, Ninth Mill Operator’s Conference,
AusIMM, Fremantle.

Lane, G., Siddall, B., 2002, Mineral Processing and Hydrometallurgy Plant Design – World’s Best
Practice, AusIMM Conference, Sydney, April.

Pyke, P., Johansen, G., English, D., Lane, G., Hayward, T., and Gardula A., 2006, Application of
HPGR in Processing Gold Ore in Australia, SAG 2006 Conference, Vancouver, September.

McLean, E., 2005, Milling in Acid, IIR 6th Crushing and Grinding Conference, Perth, Australia

McLean, E., Hayward, T., English D., Smith, S., 2007, Process Design Aspects at Bendigo
Mining’s Gold Operations, Ninth Mill Operators Conference, AusIMM, Perth

Rowland, C. A., “Section 3 - Rod Mills”, SME Minerals Processing Handbook, Eds. Weiss, N.L.,
SME – AIME, New York, 1985

Stephenson, D., 2007, Back to the Future - When Rod Mills are Required, IIR 8th Crushing and
Grinding Conference, Brisbane, Australia

Weekes, B., Jakowetz, A., Optimisation of the Sepon Copper Milling Circuit, MetPlant 2006, Perth,
Australia.

129

You might also like