Improved Cleaner

circuit performance
at the degrussa
copper mine with Abstract
an in situ column Sandfire’s DeGrussa Copper-Gold operation in West Australia is a
new high grade copper sulfide mine, which includes a 1.5 Mtpa
sparging system concentrator which was completed in 2013.
Early after commissioning, it was identified that the flotation capacity
J. Knoblauch1 - Principle Metallurgist in the all-mechanical cleaner circuit was not sufficient. In order to
H. Thanasekaran2 - Senior Process Engineer improve recovery, the tail from the cleaner scavenger was sent
E. Wasmund3 - Global Managing Director back to feed the rougher scavengers. Test-work was done on-site to
evaluate the possibility of using a flotation column as a pre-cleaner. A
CavTube™ sparging system was selected, which is especially effective
for fine ore particle flotation. Test-work was conducted at site using a
150 mm diameter unit, and then confirmed on a 500 mm diameter
unit. A 4250 mm diameter full scale unit was installed in Q4 2014
and commissioned in Q1 2015. This unit gave some significant
advantages including a reduction in re-grinding requirements,
elimination of a circulating load and increase in the recovery of
copper in the cleaner circuit and overall circuit.
In this presentation, the benefits of columns and cavitation sparging
for floating fine particles and rejecting fine entrained gangue will
be explained. Also, the results from a lab unit and a pilot unit will
be compared against the results of the full-scale unit. The paper
will demonstrate the scalability of flotation columns, and which
phenomena and scale-up criteria must be considered to be successful.
Finally, some general observations will be made about the importance
of columns in copper cleaning circuits.

KEYWORDS
Flotation, Copper Cleaner Circuit, CavTube Sparger,
Fine Particle Flotation, Column Flotation Cell

INTRODUCTION
Sandfire Resources’ DeGrussa copper-gold concentrator is located
approximately 900 km north-north-east of Perth in Western Australia. The
project area landscape is scrubby mulga lands with few notable features.
The climate is arid with high temperature and heavy summer downpours.
The ore body was discovered on Tuesday 28 April, 2009 by John
Evans, Sandfire’s chief technical director. The DeGrussa ore body
comprises lateritic gold, oxide copper and supergene copper which
is vertically above hypogene mineralization. On the Hypogene
1. Sandfire Resources NL mineralization more than 90 per cent is massive sulfide, which mainly
Level 1, 31 Ventnor Avenue,
West Perth, WA-6872, Australia consists of pyrite, chalcopyrite, pyrrhotite, sphalerite and subordinate
galena, as well as minnesotaite (talc) and magnetite.
2. Eriez Flotation Division Australia
21 Shirley Way,
Epping, Victoria 3076, Australia
3. Eriez Flotation Division Canada
7168 Venture Street,
Delta, BC, Canada, V4G 1H6

More information is available at www.eriezflotation.com 1

 FIGURE 1 DeGrussa commenced as an open-pit operation but now treats solely
DeGrussa copper mine processing underground ore through its 1.5 Mtpa concentrator. The concentrator
flow sheet, 2014 plant was built in 2011 and commenced commercial production
in September 2012. Now DeGrussa is firmly established as one
of the Asia-Pacific region’s premier, high-grade copper mine. The
concentrator produces up to 300,000 tonnes-a-year of high grade
copper concentrate.
# of Capacity
Cells
Function Model
(m3) PROCESSING PLANT DESCRIPTION
2 Rougher TK-50 50 A simplified flow sheet of the DeGrussa concentrator, from 2014
6 Scavenger TK-50 50 is shown as Figure 1. The concentrator feed is blended from four
6 First Cleaner TK-30 30 different zones of massive sulphide mineralization: DeGrussa,
Cleaner
Conductor 1, Conductor 4 and Conductor 5. This blend is determined
4 TK-30 30 by availability, composition and production targets.
Scavenger
6 Recleaner OK-16-U 16 Ore is processed through a conventional flotation circuit to separate
TABLE 1 the valuable copper sulphide bearing particles. The details of the
DeGrussa processing plant flotation equipment flotation cells are listed in Table 1.
The rougher/scavenger flotation circuit consistently recovers 91%
of the total copper sulphide minerals, with the greatest loss typically
being from the >75µm and <7µm size fraction. The enrichment ratio
on the rougher/scavenger flotation stage is around 4.5 to 5.0. The
tailing from the rougher/scavenger flotation circuit is sent to the final
tailing hopper.
Concentrate from the rougher/scavenger circuit is fed via a cluster
of hydro cyclones to a regrind circuit. Underflow from the cyclones
reports to an IsaMill where it is reduced in size to approximately 80%
passing 18 µm. The product from the IsaMill is combined with the
regrind cyclone overflow before being fed to the cleaner circuit. The
cleaner circuit operates with the following recirculating loads:
• Recleaner tailing to first cleaner
• Cleaner scavenger concentrate to regrind mill
• Cleaner scavenger tailing to scavenger cells

More information is available at www.eriezflotation.com 2

 The final tailings of the cleaner circuit are returned to the head
of the scavenger circuit. The recycle provides an opportunity to
recover valuable copper sulphide minerals in the cleaner tailings
with additional retention time in the flotation circuit and/or to handle
upsets in the cleaner circuit. The initial design target in the cleaning
stage recovery was 90 per cent.

INITIAL FLOTATION CIRCUIT

PERFORMANCE
In 2013, plant commissioning was progressed with the ore throughput
and production on track to achieve demonstrated nameplate levels
of 125,000 tonnes per month by mid-2013. Under normal operating
conditions, the rougher flotation stage recovers 50 to 60 per cent
of the copper in the feed and the scavenger flotation stage varies
FIGURE 2 from time to time. The total recovery in the rougher/scavenger circuit
Pyrite liberation in float feed (FF), recleaner remains relatively consistent.
concentrate (RCC), cleaner scavenger tail (CST),
scavenger tail (ST), flash cleaner Based on the plant monthly data, the average copper recovery in the
concentrate (FCC) cleaner circuit was generally between 75 to 82 per cent. During April
2014 when the average mill throughput was 125,000 tonnes per
month with an average feed grade of 4.8 per cent copper, the plant
copper recovery was 89.1 per cent. During this peak production
period, a detailed plant survey was conducted. The recoveries in the
rougher, scavenger and cleaner circuits based on individual circuit
feed were 65 per cent, 80 per cent and 81 per cent. The significantly
lower cleaner flotation circuit recovery implies that the cleaner circuit
was heavily overloaded.
When the cleaner circuit is overloaded, recycle of the cleaner tailings
results in floatable valuable minerals building up in the scavenger
circuits. If the condition lasts significantly longer than the circuit
residence time, the rougher and scavenger circuits will eventually
reach their capacity limit and recovery will start to decrease. The
overall plant recovery becomes dependent on recovery in the cleaner
circuit during high mill throughput and feed grade periods.
The plant historical copper concentrate grade data showed that a
large portion of liberated, non-sulfide non- floating gangue material
was being recovered to the copper concentrate via entrainment.
Particle size and mineralogical data for January 2014 were collected
from the plant and reported by ALS, (ALS Quantitative Automated
Mineralogical Analysis March, 2014). This snap-shot data is
FIGURE 3 representative of the character of the operation for that period of
Non sulphide gangue liberation in float feed (FF), time. The flotation plant feed had a grade of approximately 5.04 per
recleaner concentrate (RCC), cleaner scavenger cent copper with P80 of 40 µm. Chalcopyrite was the only significant
tail (CST), scavenger tail (ST), flash cleaner
concentrate (FCC) copper bearing mineral, and the feed contained 31.9 per cent pyrite,
14.2 per cent chalcopyrite, 9.4 per cent pyrrhotite, and 2.2 per cent
sphalerite and lesser quantities of other sulfides. 76.3 per cent of the
chalcopyrite in the float feed was fully liberated.
The plant final concentrate copper had a grade of 25.5 per cent
copper with P80 of 24 µm. The major diluents in the final concentrate
grade were iron sulphide (predominately pyrite, with trace of
pyrrhotite). As shown in Figure 2, 38.5 per cent of the pyrite in the
final concentrate stream was liberated and these were distributed
predominately in the fines and ultra-fines.
More information is available at www.eriezflotation.com 3
 Other diluents included phyllosilicates and other non-sulphide gangue
which accounted for 4.4 per cent of this stream and were distributed
across all liberation classes. Also 37 per cent of the non-sulphide
gangue in the final concentrate stream was fully liberated which is
shown in Figure 3. Considering the high amount of ultra-fines in the
cleaner circuit, it was confirmed that the existing conventional cells
are recovering a significant portion of liberated non-sulphide gangue
to the final concentrate stream through entrainment.

FIGURE 4
CLEANER CIRCUIT OPTMISATION
Copper recovery at various size fractions (%) Based on the eighteen months of operational and plant survey data,
Sandfire decided to install additional flotation cells in the cleaner
circuit to improve recovery and concentrate grade. The choice of
flotation column over mechanical cells was based on four main
considerations. Column flotation offers improved recovery over
conventional cells due to their ability to simultaneously provide
high superficial bubble area and collision rates, relatively plug flow
conditions, and a low-turbulence environment. Most importantly,
columns allow for higher grades given deep froths that can be
effectively washed. Column cells provide more flexibility with air-rate,
froth depth and wash water addition adjustments. Additional benefits
include a smaller foot print and savings in steel work and concrete.

THEORY OF IN SITU SPARGING

(CAVTUBE SPARGING SYSTEM)
It is well known that recovery by flotation is strongly dependent on
particle size, as well as other considerations such as liberation,
bubble size, density and hydrodynamics. Figure 4 (Bulatovic, 2007)
shows typical copper flotation performance by particle size for major
operations that are using mainly conventional mechanical cells.
This graph shows a passband between 50 and 150-200 microns
where conventional flotation gives good performance. As the size is
decreased below 50 microns or above 200 microns, the recovery
attenuates rapidly.
An explanation for lower recoveries of fine liberated particles has
been proposed by Sutherland (1948), who described the flotation
recovery rate process mathematically as the product of the probability
of collision, the probability of attachment and the inverse of the
probability of detachment. The probability of collision is often written
as being proportional to the second power of the ratio of the particle
size to the bubble size. In other words fine particles are not effectively
collected by large bubbles. The physical explanation for this is that
smaller particles have lower inertia and are more likely to remain in
the fluid streamlines that travel around the bubble’s boundary layer,
than to diverge from the streamline and intercept the bubble. Larger
particles, relative to bubble size, are less likely to track the fluid
streamlines and will successfully collide with the bubble when they
come in close proximity.

More information is available at www.eriezflotation.com 4

 FIGURE 5 In-situ sparging systems such as Eriez’ CavTube™ sparger, introduce
Schematic illustration of a full scale CavTube fine bubbles and bubbles that will form and grow on hydrophobic
column flotation cell surfaces. In this device, ore slurry is typically mixed with pressurized
air pumped through a sudden contraction and expansion. As the
compressed feed is expanded, bubbles on the order of 100 microns
in size are formed by shearing apart large gas slugs. Additionally,
as the local pressure exerted on the slurry drops, the concentration of
dissolved gases like nitrogen and oxygen become supersaturated and
can nucleate as fine bubbles of the order of 1 micron in size. If the local
pressure of the liquid drops below the vapor pressure, the liquid will
also become supersaturated with respect to water vapor, resulting in
cavitation. As with other super-saturation phenomenon, nucleation onto
existing surfaces (heterogeneous nucleation), especially hydrophobic
surfaces, will take place preferentially over homogeneous nucleation,
due to the additional energy required to create a new interface. So
when ore slurry is pumped through an array of CavTubes, it is possible
to create a bimodal bubble size distribution with two main modes at
about 1-2 microns and 100-200 microns respectively. Many of the fine
bubbles, which only make up on the order of 2% of the total bubble
volume, will be mostly tethered to the ore particles because of the
preponderance of heterogeneous nucleation as shown by Fan (Fan et
al, 2010a, Fan et al, 2010b). A recent study (Krasowska and Malysa,
2007) using a model system showed that large bubbles had a higher
sticking efficiency when impinging against solid surfaces that had
pre-existing adsorbed fine bubbles. Therefore, there is good evidence
to support the idea that a combination of large and small bubbles
produced by the CavTube is especially effective for the attachment of
bubbles onto fine ore particles.
In the CavTube column, a stream from the pulp phase is extracted,
and pumped through a ring manifold containing CavTubes in parallel,
which is recirculated back into the column. The ore slurry is pumped
through the array of CavTubes, which also receive pressurized air
from a second manifold. This creates a multiphase mixture of micron-
sized bubbles attached to the ore surfaces and larger free bubbles.
This configuration is shown in Figure 5.

More information is available at www.eriezflotation.com 5

 PILOT TEST-WORK, SCALE-UP
AND INSTALLATION
To identify whether the addition of columns could make an
improvement to the cleaning circuit, a number of candidate locations
were evaluated on-site using a “lab-scale” 150 mm (3 inch) diameter
test column. On-site test-work, if available, is the best way to evaluate
the metallurgical performance of a flotation column. The main
scale-up parameters are residence time and carrying capacity. In
other words, the amount of time the pulp spends in the column and
the rate of froth being recovered over a fixed cross-sectional area.
These results can be checked against Eriez’ in-house database
FIGURE 6
Photograph showing the installation of the developed from hundreds of applications, to confirm that the test-
industrial column flotation cell and 500mm pilot work is representative. The Eriez Flotation Division technical team
column unit worked with the experts at Sandfire for more than four months to
identify the best location for a column in the existing flow sheet to
produce optimal results for DeGrussa. This initial lab-scale survey
indicated that good results would be obtained by installing a column
at the head of the existing cleaner circuit, as a cleaner scalper. These
results were subsequently validated on a 500 mm (20 inch) diameter
pilot plant unit which reduced the amount of risk for Sandfire to go
ahead with a full-scale 4.27 meter diameter by 10.0 meter high unit.
Photographs of the 500 mm pilot unit and the 4.27 meter full-scale
unit are shown as Figure 6.
In 2014, Sandfire allocated approximately $14 million towards
processing plant enhancements at DeGrussa and the copper cleaner
column was part of that package. The column was subsequently
installed in December 2014 and commissioned in February 2015.
The modified flow sheet is shown as Figure 7.

FIGURE 7
Flotation circuit configuration after cleaner circuit upgrade

More information is available at www.eriezflotation.com 6

 FIGURE 8
Test results of full-scale industrial column IMPROVEMENT IN PLANT
and pilot test unit.
PERFORMANCE
A comparison of the grade-recovery curve for each of the three sizes
of column cells in this study is shown in Figure 8. The metallurgical
performance for a scale-up of diameter approximately 30 times is
roughly on the same grade-recovery curve, which provides confidence
that the scale-up criteria for columns is well known and the results can
be predictable.
As a result of this single new column, the DeGrussa operations were
able to achieve some considerable benefits. The main benefit of the
CavTube column was derived from rejection of non-sulfide gangue
allowing the recovery of valuable composite particles as well as
liberated fines. Secondly, the rougher concentrate stream could be
sent directly to the column cell (compare Figure 1 and Figure 7),
thereby minimizing the load on the regrind mill and allowing higher
kinetics on the cleaner circuit. Finally, the circulating load between
the cleaner scavenger tail and the scavenger feed, was no longer
required so that the cleaner scavenger tail could go to final tails.
Prior to installation of the column, froth in the existing conventional
cleaner cells was tenacious and stable, with poor mobility and
heavily loaded. As a result, entrainment of non sulphide gangue
was very high due to the poor froth drainage and lack of wash
water. In effect the cleaner circuit had to be run less aggressively
to increase the copper grade which also resulted in lower copper
recoveries. The operators must choose between recovery and grade.
More often operators had to split the cleaner and recleaner feed
into multiple cells to minimize the loading on the cells. The cleaner
feed stream was split into three streams and fed into cell 1, 3 &
5. There was also a similar configuration on the recleaner cells.
Since the column commissioning the issue was resolved as there is
now much less floatable material reporting to the mechanical cells.
Unlike conventional cells, column flotation technology allows for froth
washing which improves froth mobility, decreases entrainment and
maximizes grade.

More information is available at www.eriezflotation.com 7

 Survey 1 (2014)

Survey 2 (2014)

Post Column
Pre-Column

Pre-Column
Installation

Installation

Installation
Circuit
As the column cell was installed at the head of the cleaning circuit,

(2015)
it can recover most of the copper minerals prior to the existing
cleaner cells, minimizing the froth overloading issues and providing
operational flexibility of the conventional cells. It should be noted that
Total Circuit Cu
89.8 87.4 91.7 wash water cannot be employed on conventional flotation machines
Recovery (%)
Cleaner Block Cu
as quiescent bubble rise conditions are necessary to gain any benefit
80 67 98.5 from this approach. In addition, wash water added to conventional
Recovery (%)
Column Cu
cells will result in a significant decrease in residence time and hence,
64.2 recovery.
Recovery (%)
Recleaner Cu
85 82 53.3 Plant metallurgical surveys were conducted on 26th April 2015 (post
Recovery (%) column commissioning). The data collected was mass balanced.
Before and after performance data for the surveys conducted are
TABLE 2
Cleaner circuit survey results shown in Table 2.
Prior to the column cell installation, the cleaner block performance
is variable and operates below 90 per cent copper recovery.
Comparing the before and after column cell cleaner block
performance, it is clear that the overall cleaner block performance has
significantly improved in excess of 95 per cent.

SUMMARY AND CONCLUSIONS

A cleaner scalping stage column was installed at the DeGrussa
copper mine in February 2015. Plant results shows improvement
in copper recovery and final concentrate grade since column
commissioning. In addition to the metallurgical improvements,
operability and stability of the cleaner circuit have greatly improved.
Scale-up of a CavTube sparged column was successfully demonstrated
to an industrial scale in the copper cleaner circuit application.
These results show that scale-up of a column from lab scale to a full
production unit is straightforward.

ACKNOWLEDGEMENTS
The authors acknowledge the permission of Sandfire Resources to publish
this work. Many people at DeGrussa copper mine were involved in the
laboratory, piloting and concept development, as well as in the actual
implementation of the work at DeGrussa. The authors are indebted to all
those at DeGrussa who have made this project a success.

More information is available at www.eriezflotation.com 8

 REFERENCES
1. ALS Quantitative Automated Mineralogical Analysis on Monthly
Plant Composites from DeGrussa Mar 2014 (2014).
2. Bulatovic, S. M., (2007). Handbook of flotation reagents: chemistry,
theory and practice: volume 1: flotation of sulfide ores. Elsevier.
3. Sutherland K.L., (1948). “Physical chemistry of flotation. XI. Kinetics
of the flotation process”, J. Phys. Chem., Vol 52, pp 394-425.
4. Fan, M.M. et al, (2010). “Nanobubble generation and its
application in froth flotation (part I): nanobubble generation
and its effects on properties of microbubble and millimeter scale
bubble solutions”, Mining Sci. & Tech, 20, pp 1-19.
5. Fan, M.M. et al, (2010). “Nanobubble generation and its
applications in froth flotation (part II): fundamental study and
theoretical analysis”, Mining Sci. & Tech, 20, pp 159-177.
6. Krasowska, M., Malysa, K., (2007). “Kinetics of bubble collision
and attachment to hydrophobic solids: 1. Effect of surface
roughness, Int. Journal of Mineral Processing, 81, pp 205-216

This paper was originally

published at the Canadian
Mineral Processors National
Conference, Ottawa, Canada,
Jan 20, 2016

Eriez Flotation Division | Canada Inc

7168 Venture St
Delta, BC, V4G 1H6
Canada
Office: +1 604-952-2300
[email protected]

316-WEB-AHA More information is available at www.eriezflotation.com 9