Flotation data for the design of process plants

Part 2 – case studies

R. C. Dunne*1, G. S. Lane2, G. D. Richmond3 and J. Dioses2
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

The Cadia Hill Gold Mine and the Hellyer copper, zinc and lead plant present different problems
with widely different target grind sizes, liberation characteristics. The contrast is between Hellyer’s
sequential flotation circuit versus Cadia’s ‘simple’ copper–gold flotation circuit. The common
theme is that pyrite forms the principal floatable concentrate diluent. Hellyer ore contains finely
disseminated chalcopyrite, sphalerite, galena and tetrahedrite. The flotation plant design was
based on extensive benchscale test work (including locked cycle tests) on drill core and many
months of operation of a 30 t h21 ‘pilot plant’ using the modified Cleveland Tin Mine processing
plant. The Cadia concentrator was designed based on an extensive benchscale variability test
work programme. Data from approximately a dozen locked cycle tests conducted on drill core
and 2 weeks of continuous pilot plant trials of samples obtained from an adit into the orebody
were used as the basis of design. The orebody is a low grade monzonite porphyry with
disseminated chalcopyrite/bornite/pyrite (0?17%Cu, 0?73 g t21 Au). Methods used to establish
the appropriated flotation circuits for the treatment of the Hellyer and Cadia ores, as well as the
interpretation of the test work data for these particular operations, are discussed. The predicted
versus actual operational plant is reviewed and the comparison was found to be satisfactory.
Keywords: Locked cycle tests, Residence time, Finely disseminated, Flotation circuit design, Hellyer Project, Cadia Project

This paper is part of a special issue on Metallurgical Plant Design and Operating Strategies

Introduction and zinc that failed to report to their respective primary

concentrates. Typical regrind particle 80% passing (P80)
The two case studies described in this paper outline the sizes were 38 mm for Cadia and 10–25 mm for Hellyer.
following stages of design: initial benchscale flotation Project optimisation work was aimed at understand-
test work to evaluate the orebody, pilot tests using ing how mineralogical variation impacted on the
mined ore bulk samples and final design and operation metallurgical performance of the Hellyer ore. Similar
of the selected flowsheet. work is continuing at Cadia.
In both case studies, the main characteristics of
variable mineralogy, flotation kinetics and orebody
representation posed some challenges. For Hellyer, the Case study one: a complex copper, lead
slow floating nature of the minerals, the fine particle size and zinc flotation circuit (the Hellyer
and the number of valuable minerals (lead, zinc, copper Project)
and silver) to be recovered meant that a complex
flowsheet resulted. At the Hellyer site, use was made The Hellyer Project was developed by Aberfoyle Limited
of an existing tin flotation plant and various flotation and proceeded from discovery in 1983 to production
cell types were tested in a large scale ‘pilot plant’. in 6 years.1 The 156106 t orebody consisted of fine
The Cadia Hill ore was tested at laboratory scale grained massive sulphide containing copper, lead, zinc
and silver minerals with gold in a predominantly pyrite
and validated in a pilot plant using four mined ore
matrix.
samples.
Sulphide mineralisation was intersected in 1983.
Both of the final flowsheet made use of regrind stages
Preliminary metallurgical test work commenced mid-
to allow for recovery of the fine minerals and in the case
1984 and an adit to provide access to the orebody was
of Hellyer, this allowed the recovery of each valuable
initiated in 1985. Laboratory batch test work providing
mineral to separate copper, lead and zinc concentrates,
two options for further evaluation was completed by the
with a final ‘catch all’ bulk concentrate to collect lead
end of 1985. Pilot plant test work on underground
samples commenced in June 1986 and detailed design for
1
Newmont Mining Corporation, Greenwood Village, CO, USA the Hellyer concentrator started in March 1987. Plant
2
Ausenco, South Brisbane, Qld, Australia construction began in January 1988 and the first
3
Minerals and Metals Group, Southbank, Vic., Australia flotation concentrate from the initial 106 t/year concen-
*Corresponding author, email Robert.Dunne@Newmont.com trator was produced in March 1989.2,3

ß 2010 Australasian Institute of Mining and Metallurgy

Published by Maney on behalf of the Institute and The AusIMM
Received 15 April 2002; accepted 1 September 2010 Mineral Processing and Extractive
DOI 10.1179/037195510X12816242170933 Metallurgy (Trans. Inst. Min Metall. C) 2010 VOL 119 NO 4 205
 Dunne et al. Flotation data for the design of process plants: Part 2

Mineralogy implemented at the new Hellyer concentrator to be built

The mineralogy of the Hellyer ore is complex and the beside the mine. However, mine development and
orebody was classified in several categories on the basis concentrator design and construction would take a
of texture, typically associated with distance from the number of years. It was decided to convert the nearby
vent zone. The sulphide mineralisation was unaltered Luina tin concentrator in the interim to produce bulk
leading to the very fine grain size and high level of concentrate at a rate equivalent to the future Hellyer
sulphide intergrowth and association. Grain size gen- concentrator bulk production rate. This had the
erally increased and pyrite associated decreased towards advantage of retaining concentrator workforce from
the top of the orebody, and the barite cap. Gold was Luina, as the tin reserves were exhausted, as well as
principally associated with arsenopyrite and pyrite with establishing a foothold in the variable bulk concentrate
market. It also allowed further metallurgical develop-
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

minor electrum. Thus, arsenopyrite was predominantly

associated with pyrite with minor associations with ment of a difficult ore.
sphalerite and galena. Chalcopyrite was strongly asso- The key conclusions from the benchscale test work
ciated with sphalerite and often included in sphalerite were that the minerals were slow floating and long
grains. Silver occurred as tetrahedrite, in solid solution residence times were required to maximise selectivity and
in galena and very fine grains in pyrite. The typical recovery and a regrind of the lead and zinc rougher
concentrates to finer than 20 mm was required to
occurrence of minerals is shown in Table 1.
optimise performance.
Principal conclusions from the early mineralogy4 were
The Hellyer ore benchscale process flowsheet was
that a fine grind was required for sulphide liberation,
similar to that of other copper, lead and zinc concen-
although even with fine grinding it would be difficult to
trators. However, the metallurgy was considered suffi-
achieve high concentrate grades and recoveries.
ciently difficult due the fine particle liberation sizes and
Misplaced minerals would lead to loss of payable metals
the slow flotation response to warrant pilot plant
and gold would be upgraded to a refractory arsenopyr-
operation. An attempt to use the common North
ite/pyrite flotation tailings. Mineralogical variation
American practice of combined copper–silver–lead
within the orebody would probably result in only minor
roughing with subsequent separation of copper–silver
metallurgical differences.
and lead was not successful. In particular a satisfactory
Benchscale flotation test work lead concentrate grade could not be achieved.
Benchscale batch flotation test work and mineralogical Locked cycle batch flotation test work
investigations were conducted at the Aberfoyle labora- Locked cycle flotation tests on the four product
tory in Tasmania and Amdel in Adelaide. Preliminary sequential circuit confirmed that retaining production
roughing tests were performed on drill core composites. of bulk concentrate permitted higher grade separate lead
This work showed that very fine grinding, of approxi- and zinc concentrates. This enhanced economic returns
mately 40 mm, was required to obtain reasonable and greatly increased circuit stability. The locked
recoveries. Unfortunately, cleaning of the rougher flotation tests also confirmed the slow flotation rates
concentrate failed to produce acceptable concentrate and the deleterious effects of recirculating loads. It was
grades. Further testing was carried out at a laboratory in found beneficial to keep the lead and zinc roughers in
Canada where an acceptable metallurgical reagent open circuit to minimise retention times and maintain
regime was developed. A key step to this was the use circuit stability. The flowsheet development philosophy
of a laboratory mild steel mill for grinding rather than a was for a simple, conventional reagent regime in order
conventional stainless steel mill. An extensive flotation to speed development. Reagent optimisation would be a
test work programme followed at the Aberfoyle continuing part of plant operations. Depressants such as
laboratory in Tasmania. At the completion of this sulphur dioxide were discarded for the easier to handle
programme two flowsheets producing either one bulk or sodium metabisulphite.
four separate concentrates, were proposed for flotation High circulating loads in locked cycle test work for
locked cycle test work. lead and zinc cleaning resulted in a requirement to
The bulk flotation flowsheet was aimed at generating produce a bulk lead/zinc concentrate from the lead and
a combined or bulk lead, zinc, silver, gold and copper zinc cleaner tailings to enable the production of saleable
product for processing in Imperial Smelter Furnace grade concentrates. The reagent scheme required was
smelters. The four product flowsheet produced indivi- typical of other base metal concentrators. Typical locked
dual products of copper–silver, lead, zinc and bulk cycle test data showing recoveries and grades are
concentrates. The purpose of the two flowsheets related provided in Figs. 1 and 2.
to the proposed development of the Hellyer orebody. Even though there were only six cycles, the flowsheet
The four product flowsheet was designed to be for the processes was complex and time consuming. For
the six cycles there are 27 rougher floats, 6 scavenger
Table 1 Hellyer mineralogy floats, 48 cleaner floats and 18 regrind steps. One cycle
takes 280 min to complete and there are 312 reagent
Mineral Occurrence, %
addition steps. The documented predicted grades and
Pyrite 40–70 recoveries from the locked test are shown below in
Sphalerite 15–25 Table 2.
Galena 6–12
Arsenopyrite 1–3 Pilot plant
Chalcopyrite 0.8–2 Aberfoyle operated the Cleveland Tin Mine and
Tetrahedrite 0.1 processing plant at Luina, 50 km North West of
Non-sulphides 5–25
Hellyer. In the mid-1980s, the dramatic fall in the tin

206 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4
 Dunne et al. Flotation data for the design of process plants: Part 2
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

1 Hellyer recoveries

2 Hellyer concentrate grades

price resulted in closure of the tin mine that paralleled Cleveland plant. Major ore textures were tested with
the development of Hellyer. Aberfoyle decided to use the few significant differences in performance observed.
Cleveland facility to pilot the bulk and sequential At the successful completion of the pilot plant trial,
flotation flowsheets developed at benchscale. The the Cleveland plant was upgraded to 30 t h21 and
decision to conduct the 6 to 10 t h21 pilot trials at operated producing a bulk concentrate while the Hellyer
Cleveland over the typical 0?25 to 1 t h21 pilot plant at plant was designed and constructed. During this period,
facilities like Amdel was taken due to its proximity to flotation cells and impeller types were evaluated focusing
Hellyer and because this plant became available. on the particularly slow flotation rates and low air
The principal difference between the benchscale test requirements. In addition, column flotation and
work and the pilot plant operation was the increased Jameson cells were evaluated in roughing and cleaning
quantity of less than 7 mm fines produced in the pilot duties due to their increasing application in other
plant. At benchscale only 22% of the mass was less than flotation plants.5 This work led to the installation of
7 mm for a grind of 40 mm. At pilot scale this increased 50 m3 rougher and 30 m3 cleaner tank cells fitted with
to as much as 50% less than 7 mm and was typically 30– low power input flotation impellers in the majority of
35% less than 7 mm in piloting and during later Hellyer flotation circuit. A small flotation column fitted with a
plant operation. The greatest impact of this change in Jetflote (Jameson cell) downcomer feed system was
size distribution on flotation was seen in the increase in installed for copper concentrate cleaning.6
fine galena (relatively soft and high relative density) and
the slower lead flotation response. Reagent additions Metallurgical recovery predictions
increased marginally due to the higher fines content. The Predicted metallurgical recoveries for Hellyer were
slow flotation rates required low pulling rates (aeration determined from operating results obtained from the
rate) in flotation to maximise the selectivity of the pilot plant. Grade–recoveries from all stable periods of
valuable minerals over pyrite. Pilot plant flotation times operation of two or more shifts were plotted and
were typically at least double that observed in the recoveries selected for the circuits were determined from
benchscale test work. A total of 15 000 t ore was treated detailed surveys conducted during the pilot plant
through the pilot plant before an upgrade of the operations.
It had been anticipated from laboratory work that
recoveries would vary with mineralogical texture and a
Table 2 Predicted grades and recoveries from locked
number of these textures were tested in the pilot plant.
test
However, the metallurgical performance for all textures
Copper Lead Zinc tested was similar so a constant grade–recovery was
attributed to the whole of orebody. It was recognised
Grade, % 10.3 63.6 49 that performance would vary with head grade. As the
Recovery, % 48 31 62
head grade in the mine schedule was reasonably

Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4 207
 Dunne et al. Flotation data for the design of process plants: Part 2

constant for the first 5 years no allowance was made for concentrate contained approximately 35% zinc and 15%
head grade variation in the predicted grade–recovery lead.
relationship. It was anticipated that improvements in The pilot plant residence times were typically twice the
metallurgy over time would offset any deterioration of benchscale residence times. For the Hellyer plant
performance as head grade decreased near the end of the operation, the residence time was three times that
life of Hellyer. required at bench scale. The slow flotation in the plant
was linked to high levels of surface passivation (typically
Design criteria and description of the Hellyer metal hydroxides) but this was not known until after the
flotation circuit Hellyer concentrator had been commissioned. The
The Hellyer plant has a complex flotation flowsheet with passivation resulted in particles that were slow to further
oxidise as noted in the early benchscale test work.
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

four distinct final concentrates being produced and

multiple cleaning and regrinding stages. The Hellyer schematic flowsheet is shown in Fig. 3
Flotation feed was conditioned with reagents and then (lead regrind not included).
fed to the head of the first copper–silver rougher cells. Some relevant Hellyer flotation operating data are
Concentrate generated from the first three Maxwell shown in Table 3.
MX14 (50 m3) roughers was fed to a 10 m Plant operation and metallurgical development
(height)60?9 m (diameter) column cell to produce a During the operation of the Hellyer concentrator the
final silver rich copper concentrate containing 12–16% variability of grade–recovery relationship with ore type
copper. Column cell tailings were returned to the became apparent. Metal ratios, particularly zinc/iron,
conditioning tank ahead of the copper–silver rougher were a strong determinant on recoveries. It was then
cells. recognised that the initial trialling of textures in the pilot
The tailing from the copper–silver roughers was plant was flawed because by coincidence they had all
conditioned with lime to increase the pH before adding been tested at a near identical metal ratio.8 Correlations
cyanide prior to the lead roughing circuit. Flotation were then developed initially between plant performance
collector was stage added to each of the seven Maxwell and QEM*SEM phase specific surface areas measure-
MX14 lead rougher cells. The lead rougher concentrate ments and later with metal ratios. This then allowed the
fed a 335 kW Kubota Tower Mill and the product was prediction of metallurgical performance for each mining
further cleaned in a three stage cleaning circuit using block in the orebody.
Maxwell MX12 (32 m3) cells. The first cleaner tailing The significant impact of mineral grain size and
reports to the bulk flotation circuit. associations on metallurgical performance was clearly
The tailing from the lead roughers was conditioned demonstrated in the latter years of Hellyer. Statistical
with copper sulphate and lime, before zinc rougher analysis of optical mineralogical data showed a close
flotation. Collector was again stage added to each of the correlation between mineral associations and zinc
eight MX14 Maxwell zinc rougher and scavenger cells. recovery. This is illustrated in Fig. 4.
Additional copper sulphate and lime was added midway It is also important to realise the significance of
through scavenging. The zinc scavenger tailing, contain- simpler measurements such as metal ratios. These are
ing less than 1% zinc, was the final tailings. The zinc more readily available than mineralogical data and can
scavenger concentrate reported to the zinc regrind be very useful. At Hellyer the average zinc/iron ratio for
circuit. This stream reported to a Kubota Tower Mill the orebody was 0?6, but the initial pilot plant work was
in closed circuit with a second set of cyclones. This performed on ores with zinc/iron ratios of 0?7–0?8.
stream was later preclassified before regrind due to These ores floated well and led to the erroneous
differences in flotation performance caused by the conclusion that there was little difference between ore
regrind environment.7 The regrinding improved libera- types. The later treatment in the Hellyer concentrator of
tion of sphalerite from the pyrites gangue. The reground ores of zinc/iron ratio equal to 0?4 showed zinc
material was reconditioned with copper sulphate, lime recoveries could be as much as 20% lower due to the
and collector. This stream reported to the first of three increased sphalerite/pyrite associations.
zinc cleaning stages. Lime was added to the first and Subsequent investigations9 found that flotation resi-
third zinc cleaning stages to raise the pH. The zinc dence times were strongly affected by mineral oxidation.
concentrate produced from the third cleaner contained Galvanic actions occurred between the galena and the
50% zinc with a recovery of 63–68% (but as low as 45% pyrite producing lead ‘hydroxides’ that coated all
during initial operation). minerals. These lead ‘hydroxides’ reduced the flotation
Before the bulk rougher stage, the combined lead and rate of galena and increased the flotation rate of
zinc first cleaner tailings were reground in the bulk sphalerite and pyrite. Selectivity was adversely affected
regrind circuit consisting of cyclones enclosed circuit and differential separation only possible at very low
with a 335 kW Kubota Tower Mill. This regrind step flotation rates. This effect was exacerbated by increased
was necessary to reduce the number of composite fines, so it was more evident in the continuous grinding
sphalerite–pyrite and galena–pyrite particles to enable circuits than at benchscale.
more selective flotation. The product size was about 20– The mineral oxidation occurred steadily over the first
25 mm and was reconditioned with copper sulphate. 30–60 min flotation before reaching a plateau. The
Lime was added to the bulk roughers and collector Cleveland 30 t h21 operation which floated lead imme-
added to each of the six bulk rougher cells. The bulk diately after grinding experienced less oxidation than the
rougher concentrate reported to the first of two bulk Hellyer circuit which used 30 min copper flotation
cleaning stages. The bulk scavenger tailing reported with before the lead circuit. This partly explains the increased
the zinc scavenger tailing to the final tailings sump and residence time needed in the Hellyer circuit over the
was then pumped to the tailings dam. The final bulk 30 t h21 Cleveland plant.

208 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4
 Dunne et al. Flotation data for the design of process plants: Part 2
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

3 Hellyer concentrator flowsheet flotation circuit (circa 1990)

Flotation selectivity and recovery were enhanced in In retrospect, the selection of low power intensity
the laboratory and plant through the use of surface flotation cells may also have been detrimental to
cleaning agents (e.g. ethylenediaminetetraacetic acid) or flotation performance, particularly if flotation condi-
through the use of high shear rates in high intensity tions in grinding and pre-flotation are optimised.
conditioning in flotation conditioning. Improved plant The impact of some of the key process changes, as
performance was achieved through a combination of discussed in the previous section, on metallurgical
adding lime to the grinding circuit and limiting performance are shown in Fig. 5. The combination of
oxidation, attritioning with silica sand in lead condition- the use of high intensity conditioning in the copper and
ing and using high power intensities in cleaner circuit lead circuits and increased lime addition to the grinding
feed conditioning. circuit to inhibit oxidation greatly improved selectivity

Table 3 Hellyer flotation plant operating data

Item Feed Rougher Scavengers Cleaner Recleaner 3rd cleaner

Copper float
Feedrate, t h21 120–156
Concentrate, % solid 30–40 10–30 35–45
Residence time, min 32 55
Grade Cu, % 0.4 3 12
Lead float
Feedrate, t h21 120–156
Concentrate, % solid 30–40 30–40 30–40 40–50 40–50 40–50
Residence time, min 74 74 52 29 28
Grade Pb, % 6.5 25 12 35 50 60
Zinc circuit
Feedrate, t h21 75–115
Concentrate, % solid 30–40 30–40 30–40 30–40 30–40
Residence time, min 84 57 40 34
Grade Zn, % 14 30–35 45 48 50
Bulk float
Feed rate, t h21 40–60
Concentrate, % solid 10–20 40–50 40–50 40–50
Residence time min 69 75 76
Grade Pb, % 10 14 18 30
Grade Zn, % 16 29 25 76

Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4 209
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy Dunne et al. Flotation data for the design of process plants: Part 2

4 Hellyer concentrator: modelling Zn recovery from optical mineralogy parameters

5 Hellyer concentrator metal recoveries (1989–2000)

and flotation kinetics due to modification of the surface Drill core was cut in half and one metre portions
chemistry of the sulphide minerals. submitted for gold and copper assays. Composited assay
pulp samples, contained within the mineralised zone,
Case study two: a simple copper–gold were analysed for sulphur, cyanide soluble copper and
flotation circuit (the Cadia Hill Project) certain smelter penalty elements (mainly mercury).
A mineral assemblage classification was developed for
The Cadia Mine is located 25 km southwest of Orange the Cadia ore. The classification is based on Cu/S ratios
in the central west of New South Wales. Cadia has a and the amount of cyanide solution copper and provides
history of mining and exploration dating back to 1851 a classification of the sulphides as shown in Table 4.
and shortly after the first Australian gold rush at Ophir, Applying this classification to the geological block
north of Orange.10 models and cumulating the tonnage of each ore type
Exploration commenced at Cadia Hill in 1992 and provides the deportment of the different mineral types
this identified a substantial zone of copper and gold within the deposit.
mineralisation. Subsequent resource definition drilling
defined an orebody with a mineable reserve of Benchscale flotation test work
2006106 t ore with an average gold grade of
The flotation programme involved reagent scoping
0?74 g t21 and 0?17% copper. A full feasibility study
studies, ore variability test work, a pilot plant campaign
commenced in 1994, approvals to proceed with devel-
and orebody mineral (metal) recovery modelling.
opment were granted in September 1996 and the
The flotation reagent scoping study was undertaken in
processing plant was commissioned in July 1998.
three phases. Initial ‘in-house’ flotation test work by
Mineralogy
The mineralisation of economic interest consists of Table 4 Mineral assemblage classification system
native gold and chalcopyrite with lesser bornite located adapted for Cadia
within or disseminated near quartz veins mostly within Cu/S ratio S, % CNSolCu, % Interpreted sulphides
the quartz porphyry. Most veins are 1 to 20 mm thick
with a vein density of 2 to 5 m21. These rarely occupy ,0.5 .0.01 ,20 Pyritezchalcopyrite
more than 5% of the rock mass. ,0.5 .0.01 .20 Pyritezbornite
Initial diamond core drill spacing was on 1006150 m 0.5–1.5 .0.01 ,20 Chalcopyritezpyrite
grid. This was followed by both 1006100 m and 0.5–1.5 .0.01 .20 Chalcopyritezbornite
,1.5 .0.01 All Bornite
50650 m patterns to obtain a resource estimate
All ,0.01 All Oxide
classifiable as Inferred and Indicated.

210 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4
 Dunne et al. Flotation data for the design of process plants: Part 2

Newcrest indicated that a grind size of y150 mm was Locked cycle test work
near optimal. The ‘in-house’ procedure, similar to that The aim of the locked cycle test work was to evaluate
used at Newcrest’s Telfer gold mine, incorporated a and reinforce the metallurgical recovery models recovery
selective flotation regime to produce a copper–gold for developed for Cadia Hill. The models were based on
concentrate and a pyrite–gold concentrate. It was found the correlation of assay head grade and other physical
also that regrinding of the rougher concentrate had factors with laboratory batch and pilot plant concen-
benefit in improving copper concentrate grade by y2%. trate grades and recoveries. The head samples used for
Drill core samples crushed to 22 mm and charac- the flotation test work were composited from the core
terised mineralogically were sent to four vendors of samples used to conduct the laboratory batch tests for
flotation reagents for collector evaluation. Data from model construction. A batch test was also conducted on
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

the ‘standard’ flotation test were provided in the data the composite sample for comparison.
package sent to the vendors. The vendors were requested The locked cycle test results compared favourably
to evaluate collectors for the copper–gold float with a with the batch rougher test result. Most locked cycle
preference for the float to be conducted at a pH below tests showed some instability, with gold building up in
10 because of pulp viscosity problems and the potential the copper circuit circulating loads. The bulk flotation
to reduce lime costs. For the pyrite float a combination test conducted on the bornite ore was very stable in
of the collectors PAX and A208 were selected. terms of copper flotation, but still retained gold in the
The vendors were allowed two months to carry out in- recycle streams. A typical result is shown in Fig. 6.
house reagent testing before nominating their preferred The locked cycle test programme for Cadia was
flotation collector for evaluation. The appropriate flotation simpler than that for Hellyer. The total time involved in
procedure and collector were then sent to the Lakefield completing a single cycle was 50 min. Only 20 rougher
Laboratory, in Canada, who on the behalf of Newcrest floats and 20 recleaner floats were involved in a complete
carried out four flotation tests for each vendor with the test, while there were a total of 60 reagent addition steps.
purpose of optimising flotation performance (grade, The predicted copper, gold and sulphur recoveries
recovery, reagent quantities). At the completion of each recorded for the test were 84, 75 and 31% respectively.
test, the vendor was supplied with metallurgical results and The poor equilibrium conditions were predominantly
after reviewing these nominated the next changes to the associated with the occurrence of bornite in the ore. The
flotation procedure. Interestingly and to some degree not bornite floated slowly, particularly in cleaning. The gold
unexpectedly, all vendors were able to produce similar in these ore also responds particularly poorly to
copper–gold flotation results. However, the operating flotation. The magnitude of the difference between the
conditions, especially pH, varied considerably. two calculated recoveries for a given test is another
After reviewing the Lakefield test results, it was indication of the quality of the data.
decided, on the basis of reagent costs and flowsheet The bornite ore responded well to bulk flotation of the
simplicity, to carry out further evaluations of the copper sulphides and the stronger flotation collector regime.
float on only two of the collectors tested, i.e. S701, an The other ores contained high levels of pyrite, negating
bulk flotation options. Copper recoveries in the locked
ethylthiooctane base collector, and RTD11A, a thiono-
cycle tests were close to those indicated by the rougher
carbamate base collector. The next phase of the flotation
batch tests conducted on the heads. Gold recoveries
programme focused on the ‘robustness’ of these two
varied from those indicated in the batch test work, with
collectors. For this work, 34 different drill core samples
two ores achieving better results and two ores worse
were selected having both varying mineralogy and head
results in the locked cycle tests. The two poor results
grades of copper and gold. During the variability test
were obtained with bornite and chalcopyrite–bornite
programme, the collector addition was kept constant
ores in differential flotation.
while some latitude, on the part of the flotation
operator, was allowed to vary the frother addition. Pilot plant operation
The results of this programme clearly showed S701 to be The flotation pilot plant test work commenced soon
more ‘robust’ in providing acceptable copper grades and after the completion of the flotation variability test
recoveries of gold and copper for a wide range of work. Four bulk samples with varying mineralogy,
mineralogy and head grades. Consequently, S701 was selected from ore mined from a decline, were treated in
selected as the collector for all further laboratory and the pilot plant. Survey data from the pilot plant tests
pilot plant test work. were used to provide design numbers for the final

6 Typical Cadia batch locked cycle flotation test showing recovery data

Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4 211
 Dunne et al. Flotation data for the design of process plants: Part 2

flotation plant design (flotation parameters and liquid– Metallurgical recovery model
solid separation). Products from the different flotation After the completion of the 297 batch flotation tests, the
streams (i.e. copper and pyrite concentrates) were results were compiled for each mineral assemblage and a
collected to either send to smelters, or for further linear regression model was fitted to the data. Input data
laboratory test work. For the pyrite concentrates, test comprised head assay data and sample location while
work was undertaken to establish if gold contained in the outputs were copper and gold recoveries. Copper
the pyrite could be recovered by gravity, after fine concentrate grades were fixed as there was little observed
grinding, or by cyanide leaching. The gravity test work variation in concentrate grade from batch tests on
was not successful and although cyanide leaching samples of similar mineralogy. The regressed mineral
provided acceptable gold recoveries the economics were assemblage recovery models were given to the mining
not favourable.
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

engineers to incorporate then with the geological ‘block’

model to allow for pit optimisation and ore scheduling.
Variability test work In the final analysis, an overall copper and gold
An extensive laboratory flotation programme on 297 recovery–grade profile was produced for the deposit
composite drill core samples, equally spaced throughout and the data is summarised in Table 6.
the deposit, was undertaken soon after the completion
of the pilot plant campaign. The purpose of this test Flotation plant design and description
work was to provide data to determine both the copper Flotation feed from two-ball mill circuits are sent to two
and gold recovery–grade profiles for the deposit. Early parallel banks of rougher–scavenger flotation cells, each
in the batch flotation programme it because evident that bank having seven Outokumpu OK150 Tank cells. The
‘gold drop-out’ occurred during the batch cleaning tests. design residence time per bank is 20 min which is double
A typical example of this is shown in Table 5. the residence time for the batchscale test work (scale-up
During the flotation pilot plant trials at Amdel, it was factor of 2). The scavenger concentrate is rougher in a
found that gold ‘losses’ in the cleaner tailings formed Svedala VTM400 VertiMill. The reground concentrate is
only a small portion of the total gold in the circuit. From cleaned in six Outokumpu OK30 Tank cells followed by
this observation, it appeared that the gold drop-out was four cleaner scavenger cells of similar dimensions. The
only a problem in the batch flotation test and would cleaner scavenger tailings are either recycled to the
mostly be recovered in continuous flotation operation. rougher circuit or open circuited to the tailings
However, there was still the problem of evaluating the thickener. The cleaner scavenger concentrate is recycled
results from batch rougher flotation test work. This was back to the front of the cleaner (regrinding with
overcome by a procedure, incorporating both a batch scavenger concentrate), whilst the cleaner concentrate
flotation test and a locked cycle flotation test, on is pumped to a recleaner circuit consisting of four
individually composited samples representing specific Outokumpu OK8 trough flotation cells. All flotation
mineral assemblages. From the batch and locked cycle reagents are supplied by bulk tankers and reagents are
test, a factor was obtained to take account of losses of stored in large holding tanks on site.
copper and gold during the cleaning cycle. These factors The final flotation concentrate is screened to remove
compared favourably to results obtained during the coarse particles and reduce abrasion in the concentrate
Amdel pilot flotation surveys. The factors varied pipeline. The screened concentrate is dewatered in a
between 0?93 to 0?99 for copper and 0?86 to 0?95 for 12 m diameter Outokumpu high rate thickener. The
gold (i.e. copper rougher recovery6factor5final copper thickened concentrate is pumped to a storage tank
cleaner recovery). before being pumped 35 km in an underground pipeline

Table 5 Batch flotation results on Cadia ore showing gold drop-out during cleaning

Grade Recovery

Flotation products Mass, % Cu, % Au, g t21 S, % Cu, % Au, % S, %

Copper final concentrate 0.53 25.9 49.4 30.6 68.0 39.4 26.6
Copper recleaner tailing 0.30 8.6 51.2 10.8 12.6 22.9 5.2
Copper cleaner tailing 1.58 0.7 3.0 1.9 5.7 7.1 4.9
Pyrite concentrate tailing 1.98 0.5 1.6 19.4 4.8 4.9 62.6
Tailings 95.6 0.19 0.18 0.01 8.9 25.7 0.8
Calculated head 100 0.20 0.67 0.62 100 100 100

Table 6 Metallurgical performance for Cadia ores

Overall float recovery Overall process recovery*

Ore type Copper concentrate grade, % Au, % Cu, % Au, % Cu, %

Bn 34 64.1 79.4 62.5 77.1

Py/Cpy 25 77.3 84.0 75.4 80.6
Cpy/Py 25 75.1 89.1 73.2 85.6
Cpy/Bn 32 57.4 77.0 56.0 74.6
Py/Bn 25 61.5 78.4 60.0 75.3
*Includes gravity recovery.

212 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4
 Dunne et al. Flotation data for the design of process plants: Part 2
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

7 Schematic Cadia flowsheet

to the filtration plant at the railhead in Blayney. A single concentrator. However, the magnitude of the problem
Svedala VPA 1540/24 plate and frame filter, treats up to was greater than anticipated. At start-up, it became
24 t h21 of flotation concentrate and produces a filter apparent that the installed flotation concentrate pipe
cake of about 9 to 10% moisture. launders, were inadequate, leading to concentrate
The rougher scavenger flotation tailings gravitate to a spillage, reduced pulling rates and consequently lower
53 m diameter Weir Envirotech high rate thickener. The than predicted flotation recoveries. Froth factors of up
thickened tailings at 55% solids are pumped 4 km to the to 10 were measured on the plant compared to a design
tailings dam. The schematic Cadia flowsheet is shown in of 3. All the rougher, cleaner and cleaner–scavenger
Fig. 7. launders have since been replaced with open trough
Pertinent flotation design criteria are shown in launders.
Table 7. Another and more difficult problem relates to the
clarity of the concentrate thickener overflow solution.
Plant operation and metallurgical development Air entrainment, the main reason for the voluminous
Although general performance statistics for Cadia were froth, is exacerbated before pumping by the requirement
good, the start-up at Cadia was not without its own to screen the final concentrate. Numerous remedies were
share of operating and mechanical issues.11 attempted to overcome this problem including installa-
The presence of sericite/kaolinite in the Cadia ore was tion of water sprays around the circumference of the
expected, from pilot plant and batch laboratory flota- thickener, use of de-foamers, coagulants, flocculants and
tion test work, to create froth stability problems in the deaeration of the concentrate to name but a few. The

Table 7 Cadia flotation design criteria

Items Rougher Scavenger Cleaner Cleaner–scavenger Recleaner

21
Feed rate per line, t h 1024
Residence time, min 12 8 7 10 10
Pulp density, % 34 … … … …
Concentrate Cu, % 5 2 15.5 7 25
Concentrate, % solids 24 16 25 15 30
Froth factor 2 2 2.5 2 3

Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4 213
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy Dunne et al. Flotation data for the design of process plants: Part 2

8 Comparative data for copper recovery

best solution thus far is to deaerate the concentrate using copper and 8% for gold. The differences between actual
a novel deaeration cyclone developed by Outokumpu. plant recoveries and those predicted from laboratory
During commissioning it soon became apparent that and batch flotation test work were due to:
the OK150 flotation cell discharge dart valves were N under estimation of recovery by the linear regression
undersized. Larger darts were installed and modifica- models at the high end of gold and copper in the feed
tions were made to the shaft of the dart to make it more N preferential liberation of gold from pyrite in the
robust. grinding circuit because of the recycle of higher
Three flotation agitator rotors have failed because of density solids in the cyclones
damage to the tips and blades. This was caused by the N lower grades of flotation concentrate
presence of oversize pebbles pumped from the SAG mill
hopper and occurred when a number of SAG mill
N additional gold recovery due to the use of the gold
selective collector S8761
trommel screen panels failed.
The predicted concentrate regrind feed rate, has been
N additional gold recovery due to the inclusion of the
flash flotation–gravity circuit.
considerably lower than predicted. As a consequence New metallurgical recovery models for the different ore
smaller cyclones have been installed and open vane types have been developed and adopted. These predict
impellers placed in pumps. The design particle size for the gold and copper content in the final flotation tailings
the regrind circuit was a P80 of 38 mm. However, size by rather than the overall recovery. Comparative data for
size analysis of the final copper concentrate still shows copper recovery is shown in Fig. 8.
the presence of coarse unliberated particles that dilute
the overall copper content. Work has since been
conducted to modify the regrind circuit and direct
Lessons and useful hints from the case
rougher concentrate to the regrind circuit to increase studies
copper mineral liberation. The following dot points highlight a few of the more
Flotation reagent consumptions and in particular relevant lessons and outcomes from the two case studies.
frother consumption are lower than those predicted.
Frother consumption (MIBC) is currently around Hellyer Mine: complex copper, lead, zinc
10 g t21 compared to the design value of 40 g t21. The flotation circuit
predicted consumption was based on de-rated laboratory
and pilot plant addition rates. Collector addition is N The significant impact of mineral grain size and
slightly lower at 6 g t21, compared to the design of mineral association on metallurgical performance.
11 g t21, and is the total for the two collectors added. N Close correlation between mineral associations and
Originally it was intended to add only S701. However, a zinc recovery
few months before start-up it was found that addition of a N Metal ratios as an indicator of mineralogical content
Cytec collector S8761 (a monothiophosphate) had both and ore variation. Also a useful measure to select
recovery and cost saving benefits. In addition, it was felt samples for variability test work.
that the S8761 would contribute to a reduction in frother N Preferential overgrinding of the heavier sulphide
addition and possibly improved froth properties. minerals in the plant and pilot plant situation
For the first 4 months of operation, when lower grade compared to batch grinding leading to poorer
bornite type ores dominated in the feed, actual flotation performance (slower rates) between plant
concentrator copper and gold recoveries matched those and laboratory.
from the predicted recovery models. As the proportion N Vastly different chemical environment in the plant
of chalcopyrite–pyrite, pyrite–chalcopyrite and the gold compared to laboratory resulting in different flota-
content in the feed increased a progressive positive tion response. Oxidation, mineral galvanic interac-
divergence between actual and predicted recoveries tion, presence of large amounts of very fine sulphides
became apparent, the difference being around 3% for and large circulating loads all contributing to solution

214 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4
 Dunne et al. Flotation data for the design of process plants: Part 2

chemical changes and poorer flotation performance Lawrence Ruwoko of Ausenco Minerals for editing the
in the plant situation. original paper. Permission from Western Metals Ltd and
N Energy intensity in conditioning and flotation Newcrest Mining Ltd to publish results from the Hellyer
impacts on metallurgical performance when surface Mine and Cadia Hill Gold Mine is gratefully acknowl-
active species are present. edged. This paper has been reproduced with the kind
permission of the Australasian Institute for Mining and
Cadia Hill Gold Mine: simple copper–gold
Metallurgy from Metallurgical Plant Design and
flotation circuit Operations Strategies (MetPlant) 2002, 15–16 April
N Inadequate flotation launder design that was partly 2002, Sydney, Australia.
due to the application of lower ‘froth factor’ values
for design. Open trough launders are preferred to References
Published by Maney Publishing (c) IOM Communications Ltd and the Australasian Institute of Mining and Metallurgy

pipe launders for froth removal.

1. M. Eager: ‘The Hellyer Mine – feasibility to production with
N Estimates of the amount of scavenger concentrate confidence’, Proc. Conf. Mincost ’90, 133–140; 1990, Sydney,
were lower that predicted leading to an over design AusIMM.
and poor grinding performance in the regrind circuit. 2. Aberfoyle Limited Hellyer Operation, 1989, Information bulletin.
3. G. D. Richmond: ‘Zinc–lead ore concentration by Aberfoyle
As a general observation it has been found that Limited, Hellyer, Tasmania’, in ‘Mining and Metallurgical
regrind circuits are difficult to design for low grade Practices in Australasia’, The Sir Maurice Mawby Memorial
ores. Therefore, a more appropriate approach would Volumes, 2nd edn, 530–534; 1992, Sydney, AusIMM.
be to install the correctly sized equipment after the 4. G. D. Richmond and K. F. Lai: ‘Metallurgical development of the
flotation circuit has been in operation for some time Hellyer ore’, Proc. 3rd Mill Operators’ Conf., 9–14; 1988, Cobar,
AusIMM.
and design parameters are well establish. An alter- 5. G. Lane: ‘Column flotation at Hellyer’, Proc. Column Flotation
native is to install a smaller unit and increase regrind Workshop, Adelaide, SA, Australia, July 1989, AMIRA Ltd.
capacity as required. 6. G. Lane, G. Richmond and B. Quilliam: ‘Flotation cells at Hellyer
N Preferential grinding of the sulphides leading to more – horses for courses’, Proc. 4th Mill Operators’ Conf., 37–44; 1991,
Burnie, AusIMM.
gold liberation from pyrite in the plant situation 7. J. R. Glen and R. J. Anderson: ‘A review of tower milling at
compared to the laboratory. Hellyer’, Proc. 4th Mill Operators’ Conf., 13–18; 1991, Burnie,
N Flotation reagent consumption had been much lower AusIMM.
8. G. S. Lane and G. D. Richmond: ‘Improving fine particle flotation
than design. This is partly due to better reagent
control when treating large tonnages in large flotation selectivity at Hellyer’, Proc. XVIII Int. Mineral Processing Cong.,
897–904; 1993, Sydney.
cells. 9. J. Rumball and G. Richmond: ‘Measurement of oxidation in a base
metal flotation circuit by selective leaching’, Int. J. Miner. Process.,
Acknowledgements 1996, 48, 1–20.
10. R. Dunne, R. Chittenden, G. Lane and S. Morrel: ‘The Cadia Gold
The authors would like to thank the following people Copper Project – exploration to start up’, Proc. SME Ann. General
who provided data for the original paper published by Meet., Orlando, FL, USA, March 1998, SME.
11. S. Hart, J. Dioses, M. Reed, P. Gelti, B. Clements, W. Valery and
the AusIMM: Peter Wassel, Peter Bourke, Ian R. Dunne: ‘Cadia mines – reflections after one year’s operations’,
Arbuthnot, Steve Hughes, Asa Weber, Andrew Platts, Proc. SME Ann. General Meet., Denver, CO, USA, February
Steve Hart, and Paul Gelfi. Thanks are also extended to 1999, SME.

Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 4 215