Minerals Engineering 23 (2010) 1159–1166

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

Operating strategies to maximise gold recovery at Telfer

X. Zheng *, P. Manton, F. Burns, A. Crawford, P. Griffin
Newcrest Telfer Gold Mine, P.O. Box 6380, WA 6892, Australia

a r t i c l e i n f o a b s t r a c t

Article history: On average, the difference in recovery between copper and gold is 15% among the large scale copper/gold
Received 28 November 2009 operations. At Telfer, a number of operating strategies have been implemented together with a sequential
Accepted 13 March 2010 flotation circuit design to maximise gold recovery. The main operating strategies include targeting a pri-
Available online 2 April 2010
mary grind size optimum for copper recovery, designing and operating the main flotation circuit as cop-
per and pyrite sequential flotation, targeting a minimum saleable concentrate copper grade, allowing a
Keywords: portion of pyrite/gold recovered into the copper concentrate and leaching the pyrite concentrate to
Froth flotation
extract the gold. These operating strategies have lifted the recovery of both copper and gold above
Liberation
Gold
90%. There are opportunities to further improve the metallurgical performance at Telfer, including a sin-
gle stage of cleaning of the copper rougher concentrate, regrinding of the copper scavenger concentrate
prior to cleaning and regrinding of the pyrite rougher concentrate followed by additional copper/gold flo-
tation prior to pyrite leaching.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction In addition to flash flotation and gravity gold concentration in

the grinding circuit, the main flotation circuit consists of two sep-
A review on some of the large porphyry/skarn copper opera- arate sections. Each section has its own rougher/scavenger and
tions reveals that there is a relatively large gap in recovery be- cleaner circuit. Depending on flotation mode, these two sections
tween copper and gold, as shown in Table 1 and Fig. 1. can be used to float copper minerals and pyrite in sequence or to-
On average, the difference in recovery between copper and gold gether (Benson et al., 2007). While the copper concentrate is the fi-
is 15%. Although this difference depends primarily on ore mineral- nal product, pyrite concentrate when produced is further leached
ogy especially gold deportment in the ore, other factors such as in the CIL circuit to extract gold. Fig. 2 is a schematic representa-
treatment method, flowsheet configuration and operating strategy tion of the current Train 1 circuit flowsheet.
may also have a strong influence. At Telfer, these factors are con- A more detailed description of the Telfer Train 1 flotation circuit
sidered in the operating strategy to maximise gold recovery. can be found in a previous publication by Zheng et al. (2009).

2. Telfer ore and circuit configuration 3. General operating philosophy

Telfer mine operates both underground and open pit mines Since the main economic value is derived from gold for the Tel-
(Goulsbra et al., 2003). The underground ore contains predomi- fer operation, the operational focus is on maximising gold recov-
nantly chalcopyrite with a relatively large quantity of pyrite while ery. To achieve this goal, the general philosophy for circuit
chalcocite is the main copper mineral in the open pit ore. Gold is design and operating strategy is:
mostly associated with copper minerals and pyrite. There are
two parallel process trains in the ore treatment plant. More than 1. Targeting a primary grind size optimum for copper recovery.
half of the Train 1 feed comes from the underground mine with 2. Operating the main flotation circuit in a sequential mode.
the remaining ore from the open pit mine. The underground ore 3. Targeting a minimum saleable copper concentrate grade.
has a higher grade and contributes over 80% of the copper and 4. Leaching the pyrite concentrate to extract the gold.
70% of the gold in the Train 1 feed. Table 2 lists the main minerals
in the Train 1 feed based on plant monthly composite samples There are also a number of options considered to further en-
(Brown and Ma, 2007). hance metallurgical performance, including:

1. A single stage of cleaning for the copper rougher concentrate.

* Corresponding author. Tel.: +61 8 9158 6678. 2. Regrinding of the copper scavenger concentrate prior to
E-mail address: [email protected] (X. Zheng). cleaning.

0892-6875/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mineng.2010.03.014
1160 X. Zheng et al. / Minerals Engineering 23 (2010) 1159–1166

Table 1
Comparison of copper and gold recovery in copper/gold operations.

Operations Throughput head grade Recovery Throughput head grade Recovery

k tonnes % Cu g/t Au Cu Au Diff. k tonnes % Cu g/t Au Cu Au Diff.
2009 2008
Aitik – Sweden 9258 0.28 0.13 89.2 55.8 33.4 17,813 0.30 0.14 88.4 48.8 39.5
Alumbrera – Argentina 18,919 0.49 0.56 82.5 73.7 8.8 37,502 0.50 0.55 83.7 76.1 7.6
Batu Hijau – Indonesia 37,818 0.47 0.28 80.6 75.2 5.4
Cadia Hill – Australia 17,163 0.19 0.71 86.0 75.6 10.4 16,792 0.18 0.94 88.8 82.1 6.7
Ernst Henry – Australia 11,406 1.07 0.53 90.9 75.2 15.7
Freeport/Grasberg – Indonesia 43,362 1.11 1.32 90.6 82.9 7.7 70,409 0.83 0.66 90.1 79.9 10.2
Kennecott/Bingham Canyon – US 26,030 0.69 0.48 89.6 70.9 18.6 49,134 0.58 0.35 83.5 66.6 17.0
Northparkes – Australia 2789 0.71 0.24 88.9 69.2 19.6 5244 0.54 0.26 87.6 73.7 13.9
Ok Tedi – Papua New Guinea 10,546 0.83 1.06 85.8 69.0 16.8 21,663 0.85 1.02 86.7 72.5 14.2
Osborne - Australia 2549 1.93 0.65 94.5 79.0 15.5 8020 1.83 0.59 93.8 78.4 15.4
Ridgeway – Australia 5860 0.56 1.52 85.8 80.9 4.9 5775 0.67 1.93 89.6 84.7 4.9
Tintaya – Peru 3354 1.22 0.24 82.0 65.0 17.1 7110 1.38 0.27 85.2 59.5 25.8
2007 2006
Aitik – Sweden 18,178 0.32 0.14 86.8 46.3 40.5 18,481 0.40 0.25 89.5 50.7 38.8
Alumbrera – Argentina 38,607 0.56 0.67 83.4 74.0 9.4 36,350 0.56 0.71 88.5 77.3 11.2
Batu Hijau – Indonesia 46,782 0.60 0.44 86.1 81.9 4.2 47,026 0.55 0.37 87.3 79.5 7.8
Cadia Hill – Australia 17,817 0.18 0.58 89.3 79.7 9.6 15,501 0.14 0.65 84.3 76.7 7.6
Ernst Henry – Australia 11,114 0.94 0.47 91.7 73.4 18.3 10,301 0.89 0.44 91.6 72.4 19.1
Freeport/Grasberg – Indonesia 77,599 0.82 1.24 90.5 86.2 4.3 83,731 0.85 0.85 86.1 80.9 5.2
Kennecott/Bingham Canyon – US 47,525 0.53 0.38 84.2 68.4 15.9 47,857 0.63 0.49 88.1 69.4 18.7
Northparkes – Australia 5,297 0.91 0.62 89.4 74.6 14.8 5,789 1.53 0.64 94.0 79.5 14.5
Ok Tedi – Papua New Guinea 25,771 0.76 0.85 86.4 70.8 15.6 27,561 0.82 0.88 86.0 70.5 15.5
Osborne – Australia 8,270 2.09 0.75 93.7 80.2 13.5 3,147 1.85 0.96 88.2 72.2 16.0
Ridgeway – Australia 5,694 0.72 2.00 90.6 85.7 4.9 5,538 0.71 2.40 91.3 85.8 5.5
Tintaya – Peru 6,767 1.44 0.36 86.0 49.5 36.5 6,556 1.35 0.33 88.5 57.6 30.8

Note: 1. Information sourced from company annual reports, including Xstrata, Rio Tinto, Freeport-McMoRan, Newmont, Barrick, Ok Tedi and Newcrest. 2. Year reported refers
to financial year which may differ from calendar year depending on the countries.

100 Table 2
Copper
Gold Train 1 feed typical mineral composition.

90 Minerals % Mass distribution

Chalcopyrite 0.61
80 Bornite <0.01
Recovery (%)

Chalcocite/covellite 0.034
Cuprite <0.01
70 Native copper <0.01
Pyrite 2.11
Goethite 1.29
60 Other NSG 95.9

50
Edwards and Vien (1999). By plotting the cumulative percent pass-
ing size distribution against the cumulative percent passing metal
40 distribution and fitting a cubic spline function to the data (Ahlberg
et al., 1967), the cumulative metal distribution can be predicted at
Fig. 1. Comparison of copper and gold recovery in copper/gold operations.
any given grind size. Fundamentally, this relationship accounts for
both preferential breakage and classification segregation effects
3. Regrinding of the pyrite rougher concentrate followed by addi- observed in the plant. The overall flotation recovery can then be
tional flotation of copper and gold prior to leaching. predicted by combining the newly calculated metal distribution
with the size-by-size recovery obtained from the plant survey.
The model of Bazin et al. (1994) has been independently validated
4. Optimum grind size by Runge et al. (2007) and proved to be robust. Fig. 3 shows a typ-
ical size-by-size recovery profile in the Train 1 copper flotation cir-
For a given ore, mineral liberation is generally directly related to cuit. Based on this size-by-size recovery profile, recovery at various
particle size. Flotation could only proceed if the particle is suffi- grind sizes is simulated and plotted in Fig. 4. Note that the cyclo-
ciently liberated. Meanwhile, particles of different sizes react dif- sizer cut size (C5) shown in Fig. 3 is approximately 6.4 lm for cop-
ferently to the hydrodynamic conditions in a flotation cell, per sulphides and 2.8 lm for gold.
affecting the probabilities of bubble-particle collision, attachment The optimum grind size for copper recovery is estimated to be
and detachment. Hence, particle size has a strong influence on close to 80 lm in Fig. 4. Meanwhile, within the particle size range
recovery. simulated, gold recovery continues to increase as the grind size de-
The relationship between grind size and recovery is studied creases. Simulation based on a wide range of survey or monthly
using a modelling technique developed by Bazin et al. (1994). composite data produced consistent results with the optimum
The procedure is to predict particle size distribution at different grind size for copper recovery likely in a range of 75–85 lm and
grind sizes according to the Rosin–Rammler model proposed by gold recovery improving with a finer grind.
 X. Zheng et al. / Minerals Engineering 23 (2010) 1159–1166 1161

Flash Flotation and Gravity in Grinding Cu Flotation Circuit Pyrite Flotation

Cu Rougher Pyrite Rougher

8 x OK-150TC 6 x OK-150TC to
COS Grinding Circuit COF C RT PRT Final
Tailings
Flash Rougher
1 x SK-1200 Pyrite Cleaner
Cu Re-Cleaner Cu Cleaner Cu Cleaner Scavenger Pyrite Cleaner Scavenger
3 x OK-8 3 x OK-30TC 3 x OK-30TC 5 x OK-30TC 3 x OK-30TC
PCST
CCST
Falcon
4 x SB-5200

Knelson Flash Cleaner

1 x XD-30 2 x OK-6.5TC

to Gold room to Gold room to Final Concentrate to CIL

to Final Concentrate

Fig. 2. Telfer Train 1 circuit flowsheet.

Au Cu feed size P80 of 172 lm and 180 lm at Laguna Seca and Los Color-
100 ados concentrators respectively. It can be estimated that the losses
90 in the +106 lm size fraction were equivalent to approximately 7%
80 of the total copper in the feed. Huls and Hill (2005) proposed a
Recovery (%)

70 coarse particle recovery process (CPR) to re-float the coarse frac-

60
tion of the rougher tailings at Escondida. It was acknowledged that
50
CPR flotation would not be achieved in conventional flotation cells
40
and would require extreme operating conditions. To date, the CPR
30
20 process has not been implemented.
10 A similar phenomenon can be observed at Freeport (Zheng et
0 al., 2007a). At a feed size P80 of approximately 140 lm at the Free-
+150 +75 +38 +C5 -C5 port C4 concentrator, more than half of the copper lost in the
Particle Size Fraction rougher tailings was in the +106 lm size fraction. With increase
in feed size among the four concentrators at Freeport, the propor-
Fig. 3. Size-by-size recovery of copper and gold in Train 1 copper flotation circuit. tion of copper loss also increased. This can be explained by the
size-by-size recovery results: copper recovery in the 20 lm size
fraction was higher than that in the +100 lm fraction at Freeport.
94.0 75 At Telfer, the plant flotation feed had an average P80 of 95 lm,
93.5 74 significantly finer than the other major copper/gold operations.
93.0 73 This gives Telfer an advantage to achieve a higher recovery despite
Cu Recovery (%)

Au Recovery (%)

92.5 72 no regrinding of rougher/scavenger concentrate.

92.0 71
91.5 70
5. Sequential flotation versus copper only flotation
91.0 69
90.5 68 Sequential flotation produces two separate concentrates of cop-
90.0 Cu 67 per and pyrite. The pyrite concentrate is then further treated in a
89.5 Au 66 CIL circuit to extract gold. In addition to losses of valuables in
89.0
the final pyrite flotation tailings, the other main losses include cop-
65
0 20 40 60 80 100 120 140 160
per in the pyrite concentrate and gold in the leached tailings. Nev-
ertheless, by producing an additional pyrite concentrate,
P80 (µm)
sequential flotation has a number of advantages over the copper
Fig. 4. Simulated grind size versus recovery of copper and gold in Train 1 copper
only flotation even though the pyrite flotation circuit can be uti-
flotation circuit. lised to extend copper flotation during copper only flotation.
Pyrite is generally abundant in the Train 1 feed ore and better
liberated than copper minerals. This can be seen in the production
It is common to see a plant design with a relatively coarse pri- data at Telfer. Train 1 produced more pyrite concentrate than the
mary grind followed by a regrind on the rougher/scavenger con- combined copper concentrate despite the Train 1 pyrite rougher
centrate. One of the metallurgical arguments for such an bank having two cells less than the copper rougher bank. Mean-
arrangement is to reduce the potential risk of valuable minerals while, on average, the pyrite concentrate contained more than
being over-ground. However, coarse particles are generally diffi- 90% sulphides compared with more than 20% non-sulphide gangue
cult to float due to poor liberation and/or unfavourable cell hydro- in the copper concentrate. As a result of abundant, well liberated
dynamic and froth transportation conditions. This is evident in pyrite in the ore, pyrite flotation generally produced more stable
operations such as Escondida and Freeport. froth and improved composite recovery.
Coleman et al. (2006) reported that in the two surveys con- Based on the production data of the first 6 months of year 2009
ducted at the Laguna Seca and Los Colorados concentrators at when Train 1 was operated in a sequential mode, approximately
Escondida, copper losses in the +106 lm size fraction were greater 2.1% of the total copper reported to the pyrite concentrate. The va-
than those in the 53 lm size fraction in the rougher tailings. The lue of copper recovered in the pyrite concentrate currently cannot
rougher copper recovery was measured as 83.1% and 85.9% at a be counted as revenue. However, once the pyrite concentrate is
1162 X. Zheng et al. / Minerals Engineering 23 (2010) 1159–1166

available, it provides opportunities to recover not only gold but oratory flotation feed had a P80 size of 99 lm, close to that of the
also more copper. plant cyclone overflow. The pyrite flotation stage used 8 g/t of PAX.
Because of the reasons mentioned above, Telfer Train 1 flotation Nearly 90% of the copper with 71% of the gold in the sample was
circuit is designed and always operated in a sequential flotation recovered by collectorless flotation alone. The flotation kinetics of
mode unless the CIL circuit is not available. In addition pyrite, chalcopyrite appeared to be fast. The recovery reached a plateau
especially the proportion as composite with copper minerals, is of- during the collectorless flotation. At the same time, the total pyrite
ten not actively depressed in the copper flotation circuit. This is recovery from the collectorless flotation was 3.6%, the same as the
aimed to reduce copper reporting to the pyrite circuit and more non-sulphide gangue. This implies that pyrite was non-floatable in
importantly to improve gold recovery. Higher copper content in the absence of a collector and recovered into the copper concen-
the pyrite concentrate increases the chance of higher cyanide con- trate by entrainment.
sumption and/or lower gold recovery in the CIL circuit. Further- Pyrite floated in the presence of PAX. After 6 min, an additional
more, with unground feed, gold recovery in the CIL circuit is 5.3% of the copper and 22% of the gold were recovered together
relatively limited at approximately 75%. Operating cost per unit with 74.7% of the pyrite. Although not all the copper and gold
of gold produced in the pyrite flotation and CIL circuit is also higher recovered in this stage with collector were associated with pyrite,
than copper flotation. the results from this sequential flotation test can still be used as
When operating the circuit in a copper only flotation mode, a indicators for how much copper and gold are associated with pyr-
minimum acceptable concentrate grade is usually targeted to max- ite and what recoveries are to be expected when the plant circuit is
imise gold recovery. This is similar to a bulk flotation strategy operated in a sequential flotation mode or a copper only flotation
which produces only one combined copper and pyrite/gold con- mode.
centrate. Given sufficient circuit capacity, bulk flotation and Based on the plant production data, the Train 1 pyrite flotation
sequential flotation should achieve the same total recovery of cop- circuit recovered 2.1% of the copper and 9.2% of the gold in the mill
per and pyrite/gold. However, it is difficult to achieve a saleable feed. It should be noted that a portion of the pyrite is allowed to be
concentrate copper grade by bulk flotation on a copper sulphide recovered into the copper concentrate at Telfer. Typically, the Train
ore or a mixture of sulphide/oxide ore due to pyrite dilution. Bulk 1 final copper concentrate contains more than 15% pyrite.
flotation generally can only be used for oxide copper ores with low
pyrite content. Depending on the oxide copper and pyrite content,
the reagent scheme can be modified to suit not only for copper
7. Current operation performance
only flotation or bulk flotation but also for copper only flotation
with additional pyrite recovery or bulk flotation with incomplete
Under the current operating strategies i.e. relatively fine pri-
pyrite recovery.
mary grind size, copper/pyrite sequential flotation and allowance
of partial pyrite recovery into the copper concentrate, Telfer
6. Gold associated with pyrite achieved a copper recovery of 91.5% and a gold recovery of 90.7%
at a head grade of 0.28% copper and 1.4 g/t gold in Train 1 during
The proportion of the gold associated with pyrite is estimated the first 6 months of year 2009. During the same time period, the
by conducting sequential flotation tests in the laboratory. Train 1 copper concentrate contained an average of 18.7% copper and
ore contains predominantly chalcopyrite. Chalcopyrite is known 60.7 g/t gold. Figs. 6 and 7 compare the gold and copper recovery
to be floatable without collector in a moderate oxidising environ- at Telfer with the other operations listed in Table 1.
ment while pyrite does not float naturally under an alkaline condi- It should be noted that recovery alone does not represent the
tion (Heyes and Trahar, 1977; Gardner and Woods, 1979; Janetski optimum metallurgical objective. The optimum metallurgical
et al., 1977; Ahlberg et al., 1990). Therefore, a collectorless flotation objective should take into account recovery, concentrate grade
followed by another stage of flotation in the presence of a collector, and mill throughput depending on ore properties and other site
relatively pure copper concentrate can be obtained separately from specific conditions, and can only be determined ultimately by the
pyrite. Gold distribution in chalcopyrite and pyrite can then be economic return. This is beyond the scope of the current paper.
estimated. Fig. 5 shows the recovery of chalcopyrite, pyrite and This paper simply provides a methodology and a working example
gold obtained from a laboratory flotation test. The sample was ta- for an option of maximising gold recovery in a specific copper/gold
ken directly from the plant SAG mill trommel undersize stream ore at Telfer. One of the important characteristics of the Telfer ore
and ground in a laboratory ball mill. The final grinding product/lab- is that it has a higher gold to copper ratio as shown in Fig. 8.

2009 2008 2007 2006 2005 Telfer Train 1 2009

100 93.9
89.6
94.9 100
88.6
90 86.9 93.0
Cumulative Recovery (%)

88.3
90
78.7 78.3
80 80
Gold Recovery %

70 71.0 70
68.7 70.3
59.5
60 60
60.6
50 50
40 40
30 30
20 20
Chalcopyrite

10
Gold 10
3.2 3.6 Pyrite
1.9 2.7
0 0
0 2 4 6 8 10 12 14 16 18 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Flotation Time (min) Head Grade Au g/t

Fig. 5. Laboratory sequential flotation of chalcopyrite and pyrite. Fig. 6. Gold recovery versus feed grade from various operations.
 X. Zheng et al. / Minerals Engineering 23 (2010) 1159–1166 1163

2009 2008 2007 2006 2005 Telfer Train 1 2009 9. Copper scavenger concentrate regrinding
100
95 After the liberated, medium size copper mineral particles have
90 been removed in the copper rougher, the copper scavenger cells
Cu Recovery %

85 pick up proportionally more composite particles as well as slow

80 floating fines. Regrinding of the copper scavenger concentrate to-
75 gether with classification would provide an effective means to lib-
70 erate the copper minerals and gold from the composites and likely
65 improve concentrate grade in the cleaner circuit. The preliminary
60 results from the laboratory regrinding/flotation on the Train 1 cop-
per scavenger concentrate indicate that the final concentrate cop-
55
per grade increased from 15% Cu without regrinding to 21% Cu
50
with regrinding and gold from 27 g/t to 54 g/t with a head grade
0.0 0.5 1.0 1.5 2.0 2.5
of 4.7% copper and 8.8 g/t gold in the initial Train 1 copper scaven-
Head Grade %Cu
ger concentrate sample (Zheng et al., 2009).
Fig. 7. Copper recovery versus feed grade from various operations.

10. Copper separation from pyrite concentrate

Difference in Cu-Au Recovery (%)

2009 2008 2007 2006 2005 Telfer Train 1 2009

45 There is an incentive to separate copper from the pyrite concen-
40
trate as much as possible prior to leaching as copper in the pyrite
concentrate could increase cyanide consumption and reduce gold
35
recovery in the CIL circuit. Currently, pyrite concentrate is not
30 being reground prior to CIL processing. An experimental investiga-
25 tion was conducted in the laboratory, involving regrinding the pyr-
20 ite concentrate, separating copper by flotation and leaching the
15 flotation tailings.
10
5
Cumulative Concentrate Grade (%Cu)

P80 = 36 um P80 = 43 um P80 = 73 um

0 20%
0 1 2 3 4 5 6
Head Grade Au:Cu Ratio
15%
Fig. 8. Difference in copper and gold recovery as a function of head grade gold to
copper ratio.
10%

Due to the relatively high gold to copper ratio in the ore and
gold providing the main revenue for the operation at Telfer, 5%
emphasis has been on maximising recovery while allowing a rela-
tively low copper concentrate grade to be produced. However, the
0%
current studies indicate that there are opportunities to increase the 0% 10% 20% 30% 40% 50% 60% 70%
recovery further and improve concentrate grade without sacrific- Cumulative Cu Recovery (%)
ing recovery. The main opportunities include an additional single
stage of cleaning of the copper rougher concentrate, regrinding Fig. 9. Comparison of copper recovery and grade in flotation of pyrite concentrate
the copper scavenger concentrate prior to cleaning and regrinding samples without additional collector at various grind sizes.
the pyrite rougher concentrate for additional copper/gold flotation
with flotation tailings for CIL gold leaching.
Cumulative Concentrate Grade (g/t Au)

8. Copper rougher concentrate cleaning P80 = 36 um P80 = 43 um P80 = 73 um

250
The copper rougher consisting of two cells preferentially re-
cover liberated fast floating copper mineral particles. The amount 200
of composite material in the rougher concentrate is relatively min-
or. The main dilutant is entrained non-sulphide gangue, which typ- 150
ically constitutes more than 35% of the total mass of the Train 1
copper rougher concentrate. A single stage of cleaning using a lab- 100
oratory flotation cell reduced the non-sulphide gangue content
from 38% in the initial copper rougher concentrate sample to 19%
50
and increased the copper concentrate grade from 20% to 26.4% at
a copper recovery of 94.5% (Zheng et al., 2009). It is anticipated that
0
a higher concentrate grade can be achieved in an industrial scale 0% 5% 10% 15% 20% 25% 30% 35% 40% 45%
cleaner cell, especially if a column type of flotation cell is used,
Cumulative Au Recovery (%)
at a higher recovery (Zheng et al., 2007b). The tailings from the
copper rougher cleaner can then be combined with the copper Fig. 10. Comparison of gold recovery and grade in flotation of pyrite concentrate
scavenger concentrate for further regrinding and cleaning. samples without additional collector at various grind sizes.
1164 X. Zheng et al. / Minerals Engineering 23 (2010) 1159–1166

Existing pyrite cleaners

Existing pyrite roughers & cleaner scavengers
Feed from copper
rougher tailings after To final tailings To pyrite thickener
conditioning
To copper concentrate

Regrind To copper concentrate

Existing de-slime
cyclones cyclones

Regrind
Vertimill

Fig. 11. Pyrite flotation circuit incorporating regrinding and additional copper/gold flotation on pyrite concentrate.

A fresh pyrite concentrate slurry sample was taken from the Based on the above metallurgical justifications, it is proposed
plant. The sample had a head assay of 5.3 g/t gold, 0.13% copper that the existing pyrite flotation circuit can be modified to improve
and 42.4% sulphur. P80 was measured as 73 lm. Three flotation copper concentrate production. Fig. 11 shows the proposed pyrite
tests were conducted on one original sample and two after being flotation circuit configuration.
reground in a laboratory rod mill to a P80 of 43 lm and 36 lm No additional flotation cells are required in the proposed circuit
respectively. No collector was added in the flotation test. The slurry modification. Regrinding is to be carried out on the pyrite rougher
pH was adjusted to 10.5. Flotation was conducted for 10 min with concentrate. Based on the plant operational data, pyrite rougher
four concentrates collected. The results of the flotation tests at concentrate normally contains over 80% sulphides. The reduction
three different grind sizes are compared in Figs. 9 and 10 for copper in concentrate mass after the cleaning stage is insignificant. The
and gold, respectively. existing pyrite cleaner circuit can hence be used for the additional
As can be seen in Figs. 9 and 10, relatively limited recovery and copper/gold flotation with the tailings going to the existing pyrite
poor concentrate grade were obtained when flotation was per- thickener and onwards to the CIL circuit.
formed on the original pyrite concentrate sample without regrind- With this supplemental copper/gold flotation circuit available
ing. Regrinding significantly improved both recovery and in the pyrite flotation circuit, it will also reduce the risk of losing
concentrate grade. Gold was more sensitive to grind size than copper when fluctuations and disturbances occur in the main cop-
copper. per flotation circuit.
At the concentrate grade achieved at the finest grind, the new
concentrate produced from the pyrite concentrate can be blended
into the existing plant copper concentrate. This would increase the 11. Summary
overall plant Train 1 copper recovery by more than 1% while low-
ering the final combined concentrate copper grade by less than A number of operating strategies have been implemented to-
0.5%. The grade of gold in the final combined concentrate would gether with a sequential flotation circuit design to maximise recov-
not decrease, as gold grade in the re-float concentrate of the pyrite ery of the gold in the copper/gold ore at Telfer. The main operating
concentrate sample was more than two times higher than the aver- strategies include targeting a primary grind size optimum for cop-
age gold grade of the existing plant copper concentrate i.e. 215 g/t per recovery, designing and operating the main flotation circuit as
versus 65 g/t. copper and pyrite sequential flotation, targeting a minimum sale-
Regrinding of the pyrite concentrate would also improve the able concentrate copper grade, allowing a portion of pyrite/gold
gold recovery in the CIL circuit. The preliminary test results indi- recovered into the copper concentrate and leaching the pyrite con-
cate that by reducing the CIL feed size from a P80 of 75 lm to centrate to extract the gold. These operating strategies have lifted
35 lm, the leaching recovery of gold increased by more than the recovery of both copper and gold above 90% in Train 1 in the
10%. The potential increase in cyanide consumption in the CIL cir- first 6 months of year 2009.
cuit due to finer feed would be offset by a decrease in the total Opportunities to improve final copper concentrate grade exist,
amount of gold and copper in the feed to the CIL circuit after the including a single stage of cleaning of the Train 1 two cell copper
additional flotation. In addition, maintenance costs associated with rougher concentrate and regrinding of the copper scavenger con-
the CIL circuit would likely decrease due to finer CIL feed and less centrate prior to cleaning. There is also a potential to increase both
wear. copper and gold recovery further by regrinding the pyrite rougher
The laboratory experimental test results and the anticipated concentrate and utilising the existing pyrite cleaning circuit for
improvement in the plant as a result of pyrite concentrate regrind additional copper/gold flotation. A schematic representation of
are supported by the recent mineralogy study (Wightman, 2009). It the future process route is shown in Fig. 12. Potential improvement
was found that: in recovery and concentrate grade is estimated in Table 3.
Other benefits to be gained from the proposed sequential flota-
 Gold in the pyrite concentrate is present as native gold and gold tion circuit configuration and operating strategies are a potentially
alloy (electrum) which is flotation recoverable and cyanide reduced risk of losing copper due to fluctuations and disturbances
leachable. in the main copper flotation circuit, lower maintenance and oper-
 Approximately 80% of the gold is locked with pyrite. ating costs for the CIL circuit and lower reagent (collector, lime and
 More than half of the gold grains present are less than 10 lm. cyanide) consumption.
 X. Zheng et al. / Minerals Engineering 23 (2010) 1159–1166 1165

Flash Flotation and Gravity in Grinding Circuit Cu Flotation Circuit

Cu Rougher Cu Scavenger

Feed COF
SAG/BM Grinding Circuit
Cu Scavenger Cu Cleaner
Flash Rougher Cu Rougher
Cleaner Scavenger
Column Cleaner
Cu Regrind
Cyclone

Cu Re-Cleaner
Flash
Cleaner Falcon
Concentrator

Knelson Cu Regrind Mill

Concentrator

to Gold room to Gold room to Final Cu Concentrate to Final Cu Concentrate

to Final Cu Concentrate

Pyrite Flotation Circuit

Pyrite Conditioning
Pyrite Rougher tanks

to Final Tailings
Pyrite Copper Pyrite Copper
Scavenger Rougher Pyrite Deslime
Cyclone
to CIL circuit
Pyrite Regrind
Cyclone

Pyrite Regrind
Mill

to Final Cu Concentrate to Final Cu Concentrate

Fig. 12. Sequential flotation circuit configuration.

Table 3
Estimated potential metallurgical improvement with the proposed circuit.

Present Potential
Cu recovery % Au recovery % Concentrate % Cu Cu recovery % Au recovery % Concentrate % Cu
Gravity 21.7 21.7
Flash flotation 15.7 22.2 19.5 15.7 22.2 19.5
Copper flotation 76.7 41.0 18.6 76.7 41.0 21.0
Pyrite flotation 9.2 1.3 9.2 14.0
Pyrite flotation/CIL 6.9 8.3
Total 92.4 91.8 18.8 93.6 93.2 20.7

Notes:
1. The present data in Table are based on the first 6 months of production results in 2009 when the circuit was operating in a sequential mode.
2. Increase in concentrate grade from the copper flotation circuit is estimated based on reduction of non-sulphide gangue content in the first two rougher cell concentrate
from 38% to 15%.
3. Additional copper recovery from the pyrite flotation circuit is estimated based on 2.1% of the copper in the ore currently being recovered into the pyrite rougher concentrate
and 60% of this copper to be recovered into a separate copper concentrate after regrinding the pyrite rougher concentrate.
4. Additional gold recovery from the pyrite flotation and CIL circuit is estimated based on an improvement in recovery from 75% to 90%.

Acknowledgements Brown, P., Ma, W., 2007. Modal Assessments of Flotation Circuit Samples (KM2038),
Telfer External Report by G&T Metallurgical Services Ltd., Canada.
Coleman, R.G, Urtubia, H.E., Alexander, D.J., 2006. A comparison of BHP-Billition’s
The authors wish to thank Newcrest Mining Limited for allow- Minera Escondida flotation concentrators. In: Proceedings of the 38th Annual
ing this paper to be published. Many metallurgists and operation Meeting of the Canadian Mineral Processors, Ottawa, Ontario, Canada, 17–19
January, 2006.
personnel have been involved in the current project at various
Edwards, R., Vien, A., 1999. Application of a model based size recovery
stages. Their contribution is gratefully acknowledged. methodology. In: Control and Optimisation in Minerals, Metals and Materials
Processing, 38th Annual Conference of Metallurgists of CIM, Quebec City,
Canada.
References Gardner, J.R., Woods, R., 1979. An electrochemical investigation of the natural
flotability of chalcopyrite. International Journal of Mineral Processing 6, 1–
Ahlberg, E., Forssberg, K.S.E., Wang, X., 1990. The surface oxidation of pyrite in 16.
alkaline solution. Journal of Applied Electrochemistry 20, 1033–1039. Goulsbra, A., Dunne, R., Lane, G., Dreisinger, D., Hart, S., 2003. Telfer project process
Ahlberg, J.H., Nilson, E.N., Walsh, J.L., 1967. The Theory of Spline Functions and Their plant design. In: Proceedings of Eight Mill Operators’ Conference, Townsville,
Applications. Academic Press, New York. Queensland, 21–23 July, 2003.
Bazin, C., Grant, R., Copper, M., Tessire, R., 1994. A method to predict metallurgical Heyes, G.W., Trahar, W.J., 1977. The natural flotability of chalcopyrite. International
performances as a function of fineness of grind. Minerals Engineering 7, 1243– Journal of Mineral Processing 4, 317–344.
1251. Huls, B.J., Hill. G.S., 2005. The coarser particle recovery process. In: Proceedings of
Benson, M.A., Headley, C.R., Hille, S.O., Jacob, N.P., 2007. Differential flotation the 37th Annual Meeting of the Canadian Mineral Processors, Ottawa, Ontario,
commissioning at Telfer. In: Proceedings of Ninth Mill Operators’ Conference, Canada, 18–20 January, 2005.
Fremantle, WA, 19–21 March, 2007.
1166 X. Zheng et al. / Minerals Engineering 23 (2010) 1159–1166

Janetski, N.D., Woodburn, S.I., Woods, R., 1977. An electrochemical investigation of Zheng, X., Zarantonello, A., Mosher, J., Banini, G., 2007a. Optimum distribution of
pyrite flotation and depression. International Journal of Mineral Processing 4, flotation circuit capacity based on ore and machine characteristics – a plant
227–239. case study. In: Proceedings of the Flotation ’07 Conference, Cape Town, South
Runge, K., McMaster, J., Ijsselstijn, M., Vien, A., 2007. Establishing the relationship Africa, 6–21 November, 2007.
between grind size and flotation recovery using modelling techniques. In: Zheng, X., McCaffery, K., Halomoan, R.P., 2007b. Optimisation of flotation circuit
Proceedings of the Flotation ’07 Conference, Cape Town, South Africa, 6–21 configuration at PT Newmont Nusa Tenggara. In: Proceedings of the Ninth Mill
November, 2007. Operators’ Conference, Fremantle, WA, 19–21 March, 2007.
Wightman, E., 2009. Preliminary Evaluation of Visible Gold Deportment in the Zheng, X., Crawford, A., Manton, P., 2009. Telfer Train 1 copper flotation circuit
Pyrite Thickener Feed, Telfer External Report by Julius Kruttschnitt Mineral reconfiguration. In: Proceedings of the Tenth Mill Operators’ Conference,
Research Centre (JKMRC). The University of Queensland, Brisbane, Australia. Adelaide, South Australia, 12–14 October, 2009.