SAIMM_Oct_11-18:Template Journal 11/5/07 8:02 AM Page 615

Flotation cell technology and circuit J

o
design—an Anglo Platinum perspective u
r
by C.M. Rule* and A.K. Anyimadu* n
a
l

P
two plants designed to treat the Platreef ore. a
Since the early 1990s, an increasing p
proportion of UG2 ore has been processed.
Synopsis
This has increased dramatically in recent e
Froth flotation is the primary mineral processing separation step years, reaching a level of 54% of all ore treated r
employed in the beneficiation of PGM ore bodies. Anglo Platinum in FY 2006 (excluding WLTR, the Rustenburg
operates a significant portion of the total installed flotation capacity tailings retreatment operation).
in the PGM industry, which is installed at 20 operating plants across The capacities of individual concentrators
the Bushveld. range from small plants handling 50 thousand
PGM flotation has undergone a considerable evolution in both tons per month (ktpm) plants to large
the capacity of flotation cells employed as well as in the circuit operations treating in excess of 600ktpm.
design configuration over the past 30 years. Certain of Anglo
A rationalization of concentrator operations
Platinum’s flotation circuits have been in operation for a relatively
where much of the old equipment and plants
long time (in some operations installed equipment dates from the
1960s), and this allows comparative operating performance between are being replaced and modernized is currently
various equipment types and circuits to be evaluated. This paper in progress. This has resulted in a reduction in
summarizes some aspects of this experience. the number and an increase in the unit size of
Research trends in flotation are continuously monitored so that installed flotation cells. Anglo Platinum-
appropriate findings can be incorporate into existing and future managed operations currently treat over
circuit designs. Continuous evaluation and monitoring of flotation 4 million tons of ore per month through
performance is undertaken to identify opportunities for increasing hundreds of flotation cells of varying
flotation efficiencies. Mineralogical and fractional analyses of plant capacities, and produce about 90 thousand
composites and survey samples are analysed to interrogate
tons of flotation concentrate. Examples of the
speciation and association data. This analysis has enabled a better
wide variety of installed capacity of flotation
understanding of priorities and opportunities for improved
metallurgical performance.
units in Anglo platinum are given in Table I.
These opportunities are divided into three improvement areas: The figures illustrate the transition to
stability, liberation, and fines flotation. The paper discusses these, fewer, larger cells, which are shown in red, as
as well as the circuit design improvements required to take full compared with the older generation plants
advantage of the opportunities identified Frank and Klipfontein, built in the 1960s. The
best example is shown in the Rustenburg
operations figures—the new Waterval
complex, UG2 and Retrofit, commissioned in
2002 and 2007.
Introduction
Recent history, evolution and current
Most companies that extract PGMs from base
metal sulphide deposits currently use flotation situation
as a primary extraction process. For the
The evolution of Froth Flotation in Anglo
foreseeable future, comminution followed by
Platinum can be simplistically summarized into
froth flotation will continue to be the major
three different ‘concentrator eras’ as follows:
focus areas in the PGM mineral beneficiation
process. Continuous effort is being made to
improve the technical and operational
efficiency of flotation.
Anglo Platinum has 20 operating concen-
trators sited around the Eastern, Northern and
* Concentrators, Anglo Platinum.
Western limbs of the Bushveld complex, as
© The Southern African Institute of Mining and
illustrated in Figure 1. Metallurgy, 2007. SA ISSN 0038–223X/3.00 +
The concentrators treat mainly Merensky 0.00. This paper was first published at the SAIMM
and UG-2 ores, with Potgietersrust (PPL, Conference, Flotaton Cell Technology in the 21st
Sandsloot) and PPRust North being the only Century, 20 June 2007.
▲

The Journal of The Southern African Institute of Mining and Metallurgy VOLUME 107 NON-REFEREED PAPER OCTOBER 2007 615
SAIMM_Oct_11-18:Template Journal 11/5/07 8:02 AM Page 616

Flotation cell technology and circuit design—an Anglo Platinum perspective

A n g l o Pl at i n u m Co n c en t r at o r Op er at i o n s
o n t h e B u s h v el d c o m p l ex
Steady-state operation
PGM projects
Anglo Platinum joint ventures
Ramp-up operation BOIKGANTSHO JV
(ANOORAQ) POLOKWANE
PPRust POTGIETERSRUST
PLATINUMS LIMITED POLOKWANE SMELTER
PROJECT MOKOPANE
LEBOWA PLATINUM
MINES
LIMPOPO GA PHASHA JV
TWICKENHAM
BURGERSFORT
RPM: AMANDELBULT MODIKWA
BELA BELA
Union Smelter NORTHAM DER BROCHEN MOTOTOLO JV

LYDENBURG
RPM: UNION
SHEBAʼS RIDGE JV
NORTH WEST BOOYSENDAL JV
BAFOKENG/RASIMONE
RPM: RUSTENBURG GAUTENG
PTM JV Waterval PANDORA JV
RUSTENBURG Smelter PRETORIA MPUMALANGA
WITBANK 0 10 20 40
PSA1 + 2
Kilometres

Figure 1—Anglo Platinum concentrator operations

Table I

Summary of installed flotation equipment in Anglo Platinum (as at end 2007)

Operation Ore type processed Plants Capacity ktpm No. of cells m3 installed
Roughers Cleaners Roughers Cleaners

BRPM Merensky 1 240 18 24 900 380

Waterval concentrators Merensky/UG2 2 1.120 72 38 4.656 1.325
Klipfontein/Frank Merensky/UG2 2 220 224 68 596 418
WLTR Tailings 1 525 15 12 1.290 240
Union Mortimer UG2 3 425 92 93 715 208
Union Ivan UG2/tailings 1 110 10 20 500 118
Amandebult Merensky Merensky 1 320 50 94 1.068 359
Amandebult UG2 UG2 2 300 78 70 1.864 724
PPL Platreef 1 385 63 84 747 342
PPRust Platreef 1 600 30 32 3.480 1.280
Lebowa Merensky/UG2 2 140 50 26 474 196
Modikwa UG2 1 240 21 39 1.080 455
Mototolo UG2 1 200 13 21 910 459
Waterval slag Converter and furnace slag 1 50 10 7 109 8

➤ Pre 1980s—Mineral extraction circuits were charac- R&D and operational improvements efforts focused on
terized by multiple small flotation machines (typically minimizing flotation losses, advanced process control
1m3) with short residence times, employing a single and circuit stability. Circuits treating UG2 became more
mill-float configuration. Typically a three-stage predominant.
crushing, closed circuit ball mill comminution circuit
was used to prepare feed for flotation. These circuits
Flotation improvement priority areas
treated Merensky ore only.
➤ 1980s–1990s—Individual flotation machine capacities In pursuing opportunities for optimization, Anglo Platinum
increased to up to 30m3. ‘MF2’ circuits, which had constantly evaluates flotation operations within the group.
evolved from rougher tailings regrind work done in mid Technological advances in mineralogical analysis have
1980s, were employed. Typically 60 minutes allowed accurate measurement and quantification of the loss
mainstream residence time was installed. Induced air profiles across all reef types and operating plants within the
rougher cells and forced air cleaning cells were the Anglo Platinum group. A typical PGM distribution profile for
norm. The ‘3kW/m3’ mantra evolved, originally driven selected streams in a UG2 operation is shown in Figure 2.
by grind-psd suspension requirements, principally for Approximately 80% of the PGM minerals in the feed are
primary FAG operation and UG2 ores. either individually liberated, or are attached to or associated
➤ 2000s—Larger tank cells, with capacities up to 130 m3, with base metal sulphides (BMS). The remainder are locked
were installed. Longer residence times were utilized, or middlings particles. The concentrate consists of liberated
especially in concentrate cleaning. The ‘3kW/m3’ and BMS associated/enclosed particles, with a much smaller
doctrine was discarded, particularly on the larger cells. proportion of locks and middlings.
▲

616 OCTOBER 2007 VOLUME 107 NON-REFEREED PAPER The Journal of The Southern African Institute of Mining and Metallurgy
SAIMM_Oct_11-18:Template Journal 11/5/07 8:02 AM Page 617

Flotation cell technology and circuit design—an Anglo Platinum perspective

J
100% o
u
80% r
n
60% a
l
40%

20%
P
a
0%
Feed Concentrate Tailing –10 +10 +53 p
Liberated Enclosed in BMS Attached to BMS
PGM/BMS associated Enclosed in silicate Attached to silicate e
Enclosed in oxide Attached to oxide Grain boundary
r
Figure 2—Typical UG2 PGM association profile—monthly composite

Recovery (from tails stream) 2006

40

35

30
% Recovery

25

20

15

10

5
4- 14- 24- 3- 13- 23- 3- 13- 23- 2- 12- 22- 2- 12- 22- 1- 11- 21- 31- 10- 20- 30- 10- 20- 30- 9- 19- 29-
Mar Mar Mar Apr Apr Apr May May May Jun Jun Jun Jul Jul Jul Aug Aug Aug Aug Sep Sep Sep Oct Oct Oct Nov Nov Nov

Before Grind After Grind

Figure 3—Additional ‘hot float’ and ‘regrind hot float’ flotation potential for an operating UG-2 plant

The tailings display appreciable proportions of liberated. Stability

BMS associated and locked/middlings particles, but it is
imperative to note the distribution of these losses between For the purposes of this discussion, all process inefficiencies
the various size classes. In this particular case the total are collected under the definition of plant instability. This
tailings amounted to ~ 0.5 g/t 4E, (Pt, Pd, Rh & Au). Of this, would include the non-recovery of adequately liberated (but
0.35 g/t occurs as locked PGM particles and 0.15 g/t as not ‘fine’ which can be arbitrarily defined as <10 μm) flotable
liberated/BMS associated PGMs. The deportment in size value minerals. Selectivity issues, gangue competition,
classes shows the majority of ‘fines losses’ occur in the -10 surface coating covered by, and inhibited flotation are all
micron fraction.
included in this definition, which includes all the particles
For simplicity, the causes of these losses are loosely
that should have been recovered in the float, but were not.
grouped into three categories:
Operational instability includes, among others, sub-
➤ Inefficient flotation—predominantly fine liberated PGM
optimal process control, plant stops and starts, and sub-
(and BMS) losses
optimal circuit configuration (such as insufficient residence
➤ Incomplete or non-liberated values—composites with
time and inappropriate reagent regimes). The extent of extra
gangue
➤ Plant instability—sub-optimal process control and flotation potential in this category of losses can be gauged by
plant stops/start-ups; poor operation. performing tailings re-floats, or the so-called ‘hot floats’ that
are routinely performed on the various operations. These re-
The second and third categories of loss have almost
obvious solutions—improved grinding (resulting in increased floats generally produce roughly 10% recovery (of the tails
liberation) and improved plant operation and especially stream) for more stable operations, but up to 25% recovery
stability. Three priority areas have been identified for (of the tails stream) for less stable plants. These ‘hot floats’
optimization: stability, liberation, and fine particle flotation. are done on ‘as is’ material and after additional regrinding. A
These three factors are interrelated, but will be discussed typical example of ‘hot float’ results for a UG-2 operation is
separately in this paper. shown in Figure 3.
▲

The Journal of The Southern African Institute of Mining and Metallurgy VOLUME 107 NON-REFEREED PAPER OCTOBER 2007 617
SAIMM_Oct_11-18:Template Journal 11/5/07 8:02 AM Page 618

Flotation cell technology and circuit design—an Anglo Platinum perspective

The biggest opportunity for recovery improvement is However, numerical evaluation of the data certainly
clearly linked to improved liberation. The mineralogical shows the potential of finer grinding—assuming that
analyses and assay work on sized composite samples from flotation can recover fine PGMs.
the plants indicate that the majority of PGM losses in plant In summary the data show:
final tails are due to either un-liberated or partially liberated ➤ Feeds—clearly this is a ‘relatively’ good UG2 ore. The
minerals—composites or locked particles. The proportion majority of PGMs are either liberated or associated with
varies between plants and between ore types. sulphides, say 83% in these classes
Once plant stability has been consistently achieved, the ➤ Concentrates—clearly demonstrates that recovery of
largest opportunities that remain for flotation improvement PGMs is predominantly of liberated or sulphide
are finer grinding of flotation feed, and improvement in the
associated PGMs, say 97% in these classes
flotation of the fines thus generated.
➤ Tailings—as mentioned earlier, the losses are split
The need to grind more finely between liberated and partially liberated particles, fines
losses in the -10 micron fraction predominantly and
The economics of acquiring additional capital equipment and
partially or not liberated losses in the coarser fractions.
the associated operating costs in pursuit of optimum
liberation and hence recovery of metals needs to be justified The following tailings MLA-sourced photomicrographs
by an actual business application. Grinding more finely has indicate the difficulty of PGM liberation—a case of ‘needles in
limitations with respect to flotation performance, and also in haystacks’.
areas such as filtration of concentrates and the thickening In summary, Table II and Figure 4 illustrate the following:
and deposition of tailings. All of these must be considered in ➤ The PGMs in the coarser fraction (+53 micron) of the
the overall proposal. tails are not sufficiently liberated

1 PGM
2 Ni—Sulphides
3 Fe—Sulphides
4 Cu—Sulphides
5 Chromite
6 Pyroxene
7 Feldspar
8 Altered—Silicates
9 Quartz

150 μm

Figure 4—MLA output for PGM losses in the flotation tails—typical sample

Table II

UG2 Plant—Typical PGM association table for feeds, concentrates and tailings

Feed Concentrate Tailings Tailings-10 Tailings+10 Tailings+53

Liberated 61.0 54.8 25.1 89.0 4.2 0.0

Enclosed in BMS 5.2 30.3 3.2 0.8 7.0 0.2
Attached to BMS 12.2 7.4 1.7 1.2 3.3 0.1
PGM/BMS associated 5.4 4.0 10.5 0.4 17.5 9.7
Enclosed in silicate 3.9 1.1 28.6 2.0 26.4 53.3
Attached to silicate 4.3 1.3 17.2 6.4 28.5 11.3
Enclosed in oxide 0.1 0.1 4.8 0.0 9.6 2.4
Attached to oxide 1.4 0.1 0.7 0.2 0.0 1.9
Grain boundary 6.5 0.9 8.2 0.0 3.5 21.1
Total 100.0 100.0 100.0 100.0 100.0 100.0
No. of particles 132 958 255 100 80 75
▲

618 OCTOBER 2007 VOLUME 107 NON-REFEREED PAPER The Journal of The Southern African Institute of Mining and Metallurgy
SAIMM_Oct_11-18:Template Journal 11/5/07 8:02 AM Page 619

Flotation cell technology and circuit design—an Anglo Platinum perspective

J
Liberated Fines: Intermediate: fast floating
High surface area Lower collector need, o
Need high collector Composites need depressant
and low depressant Coarse Particles u
Low liberation
r
n
Recovery
a
l

5 10 30 50 80 100 150 200

P
Size (microns) a
Figure 5—Typical flotation size/recovery curve p
e
r

Figure 6—Flotation recovery versus size for different minerals

Pt Recovery per size, and % of total lost Pt from size class

100

90

80

70
Pt Recovery (%)

60

50

40

30

20

10

0
5 25 53 75 106 150 212 300
Screen size (micron)
% lost Pt Recovery per size

Figure 7—Particle size versus flotation recovery; and % of lost PGM from different particle size classes in an Anglo Platinum UG2 plant2

➤ The PGMs contained in the ultra-fines fraction (-10 have a flattened ‘n’ shape as indicated in Figure 5, which also
micron) of the tails are well liberated. provides some explanations for the drop-off in ultra-fines
and very coarse particles.
This is not surprising information—given that in typical
Even though the general shape remains for most
UG2 ore, PGMs are typically less than 10 microns in size. In
minerals, the absolute numbers of the threshold (for drop-
fact, it can be said that the typical industry-wide definition of
offs etc) vary for different minerals and flotation systems, as
‘poor’ or ‘good’ metallurgically characterized ore types are indicated in Figure 6, an excerpt from a paper by Professor
simply shows relativity with respect to the degree of sulphide Jameson that illustrates typical relationships.
mineral association with total PGMs: ‘poor’ ore has higher For typical UG2 ores, for example, the finer natural PGM
PGM particle association with silicates. grain size in the in situ ore and the relative performance of
PGM flotation plants indicate that the ‘watershed’ size where
Having ground more finely, can we float ‘fines’?
flotation drops off significantly must be relatively fine. This is
Classically, flotation recovery per size classes is known to indicated in Figure 7, where data collected from a UG2
▲

The Journal of The Southern African Institute of Mining and Metallurgy VOLUME 107 NON-REFEREED PAPER OCTOBER 2007 619
SAIMM_Oct_11-18:Template Journal 11/5/07 8:02 AM Page 620

Flotation cell technology and circuit design—an Anglo Platinum perspective

concentrator illustrate the relative recovery by size fraction in IsaMilling for production of fine particle size distributions
the mainstream roughers and scavengers. The surprising presents useful lessons applicable to in the PGM industry.
feature is the very good fines flotation and steep drop-off in
coarse fraction PGM recovery. Fines flotation at Anglo Platinum
This shows that finer particles have a better chance of
Typical size-by-size analysis of flotation feed products and
recovery than coarser (> 53 micron) particles. More evidence
tails across a full scale flotation plant produces the curves
for the ability to float fines is given for Zn-Cu systems (Mt.
Isa) as well as PGMs (Anglo Platinum) in the sections that shown in Figure 10.
follow. The data in Figure 10 show that fines have a much better
chance of flotation. This trend is in agreement with the
Lessons from Pb/Zn, McArthur River and Mt Isa mines analysis of UG2 monthly composites of feed tails and concen-
The ore mineralogy required that ultra-fine flotation was trates (an example of which is shown in Figure 7). Even with
needed in the operating regime to produce a saleable the present flotation equipment, sub 10 micron particles are
concentrate product. The only option was therefore to develop floating reasonably well. Work is currently ongoing to
an ultra-fine grinding technology, and design the appropriate
identify better ways of improving ultra-fine (less than 5
circuit to float the fines generated. McArthur River and
micron) flotation. It is the authors’ opinion that a different
George Fisher, Mt Isa plants therefore implemented a flotation
circuit performance and optimization through ultra-fine flotation technology will solve this problem eventually. It is
grinding (UFG), as illustrated in Figure 8. unlikely to be achieved by the introduction of more and more
IsaMill technology was developed by scaling up from power into conventionally available mechanical cells. In this
existing grinding technology specifically to enable these increasingly energy-aware world, the use of energy forinef-
particle size distributions to be achieved. The technology is ficient flotation systems in inherently wrong!
recognized as the reason that both plants are in operation
today. The enhanced fines flotation from the Mt. Isa Mine,
Circuit considerations
George Fisher plant, can be seen in Figure 9, which shows
recovery per size class results. The successful use of A typical Anglo Platinum UG-2 circuit is as shown in Figure 11.

Figure 8—Mount Isa lead zinc concentrator flow sheet

▲

620 OCTOBER 2007 VOLUME 107 NON-REFEREED PAPER The Journal of The Southern African Institute of Mining and Metallurgy
SAIMM_Oct_11-18:Template Journal 11/5/07 8:02 AM Page 621

Flotation cell technology and circuit design—an Anglo Platinum perspective

J
Recovery by size in the zinc circuit
wrt rougher con o
Zinc distribution in feed Zn recovery by size u
100.0% 50.0% r
Zinc recovery in size fraction %
90.0% 45.0%
n
80.0% 40.0%
a

Zinc distribution in feed %

70.0% 35.0%

60.0%
l
30.0%

50.0% 25.0%

40.0% 20.0%

30.0% 15.0% P
20.0% 10.0% a
10.0% 5.0% p
0.0%
C7 C6 C5/C4 C3-C1 38/53 75
0.0% e
size fraction
r
Figure 9—Recovery by size class in the zinc circuit

200.00

180.00

160.00

140.00
2E + Au grade (g/t)

120.00 53 μm
+10 -53
100.00
-10 μm
80.00 unsized

60.00

40.00

20.00

0.00
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00
2E + Au recovery (%)

Figure 10—PGM grade-recovery profile down a rougher bank for different particle size fractions2

Figure 11—Flowsheet overview of a UG-2 operation at Anglo Platinum

▲

The Journal of The Southern African Institute of Mining and Metallurgy VOLUME 107 NON-REFEREED PAPER OCTOBER 2007 621
SAIMM_Oct_11-18:Template Journal 11/5/07 8:02 AM Page 622

Flotation cell technology and circuit design—an Anglo Platinum perspective

Research into important factors considered for flotation other factors that influence overall residence time and
circuit design are monitored and incorporated into Anglo cell sizes are froth launder lip length to cell volume
Platinum’s designs where appropriate. Much of this work is ratio and duty
done with external initiatives and partners, e.g. AMIRA. ➤ A combination of mechanical and column cells should
Factors include parameters such as: be used for UG2—overall grade-Cr2O3 optimization
drivers also taking into account, flexibility from
➤ Mineralogical analysis
‘flotability’ matching, cell size, lip length ratios and
➤ Residence time and circuit configuration—size, number
open circuit arrangementsd
and arrangement of cells
➤ Froth recovery and launder lip length are important
➤ Fundamental flotation cell design
issues for optimizing mainstream recovery. In
➤ ‘Flotability’ matching or kinetic matching philosophy
particular the recovery of ‘difficult’ and ‘slow’ floating
➤ Machine type, cell size, launder and froth transport,
species can be affected by reagents and more signifi-
and grade control.
cantly by intermediate concentrate regrinding
The need for ultra-fine flotation (especially for UG2 ores
➤ Process control and circuit assessment will drive open
and altered ores generally) presents challenges that must be
circuit operation in conjunction with UFG (ultra-fine
met in current and future flotation circuit designs. In order to
grinding) of intermediate concentrate streams. Open
achieve metals recoveries in excess of 90% for all UG2 ores,
circuit operation has numerous advantages, not the
for example, these problems and issues must be successfully
least of which is being of which is able to monitor
addressed. Issues that are current areas of focus include the
performance in real time
following:
➤ New flotation technology will be required to improve
➤ There is a physical constraint on fines flotation due to ultra-fine flotation i.e. sub 5 micron particles. This may
the energy requirement for a very small particle to well be an external bubble generator, possibly in hybrid
attach to a bubble. How is this to be overcome? design Flotation in size classes is desirable
Recovery decreases because viscous and electrostatic ➤ The flotation circuit of the future could be a multi-
forces have a strong influence on ultra-fine particle stream plant targeting specific size fractions with
attachment. With ultra-fines, increased surface area
specific equipment employed to maximize the recovery
flux is required to carry a given mass of solids. Will a
characteristics of the particular stream
new flotation cell technology in a design different from
the currently available cells evolve for widespread
Acknowledgments
commercial use?
➤ Scaling up flotation performance designs from the Executive Management of Anglo Platinum for permission to
laboratory to the pilot stage and then full commercial publish the content of this paper.
application raises questions about, Is putting more and Xstrata Technology (XT), for permission to use excerpts
more power into conventional cell designs a solution? from published XT information.
Does the increased turbulence resulting have a negative
impact? Drive designs and associated cost for large
capacity cells are a real constraint on cost effectiveness. References
Is this an energy-efficient route?
Internal report: Mortimer UG2 Tailings Re-floats. By Vusumuzi Mahlangu.

Future circuit considerations and conclusions December 2006.

Certain factors will have to be considered in the flotation Internal report: The Effect of Circuit Design on UG2 flotation—a comparison of
circuits of the future, in order to take advantage of research the two Amandelbult UG2 Concentrators, (210 ktpm and 75 ktpm). By
findings for enhanced PGM flotation. These include, but are A.K. Anyimadu, M. Smith and S. Naik. November, 2006.
not limited to:
Internal Anglo Platinum mineralogical reports—Anglo Research, Mineralogy
➤ The marriage of available and new technology flotation Section.
machines in ‘hybrid circuits’ for optimal flotation circuit
design could easily involve conventional cells, ‘contact’ Excerpts from various papers and private communications from Xstrata
cells and column cells, arranged optimally in one Technology, published information related to the development of the

flotation circuit IsaMill technology and the impact on the McArthur River and Mt Isa-

➤ The use of mechanical cells of sizes to match required George Fisher Concentrators.

residence time and configuration, with high power cells

Various reports conducted in the Amira programme.
restricted to portions of the mainstream is likely.
Classic cell configurations could be developed with Analyses of numerous studies at plant, pilot and bench scale conducted over
respect to by passing driving installed cells per bank; the years within Anglo Platinum. ◆
▲

622 OCTOBER 2007 VOLUME 107 NON-REFEREED PAPER The Journal of The Southern African Institute of Mining and Metallurgy