44th Annual Canadian Mineral Processors Operators Conference©

Ottawa, Ontario, January 17-19, 2012

AN OVERVIEW OF OPTIMIZING STRATEGIES FOR FLOTATION BANKS

M. Maldonado, R. Araya and *J.A. Finch.

McGill University
3610 University Street
Montreal, Canada H3A 2B2
(*Corresponding author: [email protected])

ABSTRACT

A flotation bank consists of a serial arrangement of cells. How to optimally operate a bank
remains a challenge. This article reviews three strategies for optimizing bank performance: air profiling,
mass-pull (froth velocity) profiling and Peak Air Recovery (PAR) profiling. These are all ways of
manipulating the recovery profile down a bank, which may be the property being exploited. Mathematical
analysis has shown that a flat cell-by-cell recovery profile maximizes the separation of two floatable
minerals for a given target bank recovery when the relative floatability is constant down the bank.
Available bank survey data is analyzed in respect to recovery profiling. Possible variations on recovery
profile to minimize entrainment are discussed.

KEYWORDS

Air Profiling, Mass-Pull, Peak Air Recovery (PAR), Optimization, Flotation Bank

211
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

INTRODUCTION

Flotation is used to separate valuable minerals from each other and from gangue. To reach a target
metallurgical performance, usually assessed in terms of concentrate grade and recovery, feed is passed
through stages such as roughing, cleaning and scavenging. All these stages comprise serial arrangements of
flotation cells known as banks, lines or rows.

Although a bank is the simplest interconnection of cells in a circuit, i.e. the tails of one cell is the
feed to the next cell down the bank, how to operate each flotation cell in a way that the whole bank
performance is optimized remains a challenge. Significant efforts have been made towards understanding
the effect of operating variables on the flotation performance of an isolated cell (Lynch et al., 1981; Finch
and Dobby, 1990; Gorain et al., 1998). However in the case of banks the variables that can be manipulated
increase with the number of cells in the bank and the problem of how to manipulate them to achieve
optimum performance is difficult even disregarding variation in ore characteristics (Lynch et al., 1981, p.
16). In general, the more manipulated variables available the better for optimization purposes, however
without clear guidance how to effectively use them it becomes an “embarrassment of riches”.

Due to their localized impact, gas rate and/or froth depth are usually used to modify the operating
point of a cell in a bank. The problem then becomes to find the optimal profile (e.g. gas rate profile) that
achieves the target metallurgical objective. This solution is not obvious and a brute force approach based
on a trial-and-error search rapidly becomes intractable even for simulation. To exemplify this point,
consider a bank of 9 cells and assume that only the froth depth in each cell can be manipulated, then for 10
discrete froth depth values in each cell the number of possible froth depth profiles rises to 109!

Attempts to solve this optimization problem have been proposed (Maldonado et al., 2007;
Sbarbaro et al., 2008; Rojas and Cipriano, 2011) but no industrial applications have been reported.

This paper reviews three operational strategies to improve bank performance that have been
successfully implemented in several industrial operations: air rate profiling, mass-pull (froth velocity)
profiling, and Peak Air Recovery (PAR) air profiling. Although different in concept they are all ways of
manipulating the recovery profile down a bank, which may be the property being exploited. Mathematical
analysis has shown that a flat cell-by-cell recovery profile maximizes the separation of two floatable
minerals for a given target bank recovery when the relative floatability is constant down the bank
(Maldonado et al., 2011). It is suggested that understanding the role of recovery profiling may help to link
these strategies.

BANK OPTIMIZING ESTRATEGIES

Air Rate Profiling

The air rate profiling strategy consists of distributing air to each cell to achieve a set pattern down
the bank. Xstrata Brunswick Division pioneered this strategy on the final Zn cleaner bank of seven DR100
Denver cells (Cooper et al., 2004). The bank ‘as found’ had no obvious air distribution (profile) strategy.
Three air profiles were imposed: decreasing, balanced and increasing. It was found that the increasing
profile gave the smallest variation between best and worst performance and the highest Zn concentrate
grade for the target Zn recovery, 75% (Figure 1). The increasing air rate profile was adopted on all four
cleaner stages with total bank air adjusted to achieve target bank recovery; and it remains the practice.

212
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

Figure 1 – ‘Down-the-bank’ Zn grade-recovery curve showing best and worst performances of the three air
rate profiles: increasing, balance and decreasing.

Analysis concluded the following: that operating with reduced air rate in the first cells improved
selectivity against entrainment by reducing water recovery and that this high starting grade aided
increasing grade at target bank recovery. The need to increase air down the bank, it was argued, was to
compensate for reduced floatability, but more directly it was necessary to provide the total bank air
required to meet the target recovery. The analysis included recovery profiles and it was observed that the
relative floatability of sphalerite to pyrite (S=ksp/kpy) was independent of the air profiling and fairly
constant down the bank (S ~ 2). Other operations have subsequently implemented the increasing air profile
with significant performance benefits (Aslan & Boz, 2010; Hernandez-Aguilar & Reddick, 2007).

Peak Air Recovery (PAR) profiling

Air recovery (α) was first proposed by Woodburn et al. (1994) as a froth stability index. It is
defined as the air entering the cell that overflows as unburst bubbles. It can be calculated using the
following expression:

ξ ⋅ vf ⋅ h ⋅ w
ξ ⋅ vf ⋅ h ⋅ w
α=
Qg

α= (1)
Qg
where Qg is the total air flowrate into the cell, ξ the gas holdup in the froth zone (assumed to be 1), vf the
overflowing froth velocity, h the froth height over the lip, and w the length of the lip where froth is
overflowing. A camera monitors the surface of the froth to calculate the velocity at which froth overflows
(vf) using image processing techniques.

Studies found there was a gas rate that produces the maximum air recovery and that operating at
this gas rate, flotation performance, particularly mineral recovery, can be improved (Hadler et al., 2010a).
Part of the explanation offered is that gas rates below PAR produce a highly loaded froth with low mobility
and at gas rates beyond PAR the loading per bubble decreases producing unstable froth. The PAR air
profiling strategy then consists of operating each cell of a bank at the gas rate that maximizes the air
recovery (Hadler & Cilliers, 2009; Hadler et al., 2010b). Figure 2 illustrates PAR air profiling strategy for
a bank of N cells where Jg is the superficial gas velocity.

213
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

α1 α2 α3 αΝ

Jg1 Jg2 Jg3 JgN

1 2 3 N

Figure 2 – PAR air profiling strategy applied to a bank of N cells

Application of the PAR strategy in a platinum concentrator produced a significant increase in

platinum recovery for a given concentrate grade when operating each cell of a bank of 4 cells at the PAR
air rate (Smith et al. 2010a, 2010b).

Mass-Pull Profiling

This method profiles the solid mass overflow (concentrate) rate. It is usually implemented by
controlling the froth velocity. This strategy has been used in several operations having the attraction that
measurement of froth velocity using image analysis (machine vision) is non-invasive (Aldrich et al., 2010).

Supomo et al. (2008) describe a machine vision based control for a 9-cell rougher bank at the PT-Freeport
copper/gold operation in Indonesia. The froth velocity profile is adjusted by manipulating froth depth
designed to maximize concentrate rate (maximize recovery) subject to a trade-off between incremental
recovery and additional transport and concentrate treatment charges. An exponential decay froth velocity
profile was proposed (Figure 3) based on the exponential characteristic of a typical flotation kinetic curve.
A 1% copper recovery increase at an acceptable concentrate grade was reported after implementation.

Figure 3 – Illustration of the froth velocity setpoint profile targeted by PT Freeport (Supomo et al., 2008)

A similar study was conducted at Los Colorados concentrator at BHP Escondida (Figueroa et al.,
2009). The VisioFrothTM system was installed in the copper rougher circuit of 8 parallel banks of 10
(100m3 OK) cells. Based on tonnage and copper feed grade an expert system control continuously selects
between three predetermined froth velocity profiles labeled as low-velocity, medium-velocity and high-
velocity for each of the rougher banks (Figure 4). To achieve these profiles air rate and froth depth are
adjusted in each cell. An increase of 0.5% Cu recovery was achieved.

214
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

15
average low-velocity
Froth velocity (cm/s)

12 average medium-velocity
average high-velocity

0
1 2 3 4 5 6 7 8 9 10
Cell number

Figure 4 –Average froth velocity profiles implemented at Los Colorados concentrator, Escondida Mine
(Figueroa et al., 2009)

A RECOVERY PROFILING APPROACH

Although different in concept and roots the strategies described above are all ways of
manipulating the recovery down the bank. In this section a strategy for optimizing banks based on cell-by-
cell recovery profiling is argued. For the sake of simplicity two floatable minerals A and B are considered.
Figure 5 depicts a mass-balance on each cell of a bank composed of N cells where Ri is the recovery of the
cell i.

1 (1-R1) 2 (1-R1)(1-R2) N (1-R1)(1-R2)…(1-RN)

1

R1 R1(1-R2) RN(1-R1)(1-R2)…(1-RN-1)

Figure 5 – flotation bank composed of N cells.

Making the common assumption of first-order flotation kinetics and fully mixed transport, recovery of
mineral A and B in an isolated cell is given by:

k A ⋅τ k B ⋅τ
RA = , RB = (2)
1 + k A ⋅τ 1 + k B ⋅τ

where kA and kB are the flotation rate constants for mineral A and B and τ is the average residence time.
The relative floatability, S, is defined as the ratio of the flotation rate constants (Gaudin, 1957) and can be
expressed as a function of the recovery of each mineral.

kA R A 1- R B
S= = ⋅ (3)
kB 1− R A R B

215
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

The relative floatability provides an indication about how difficult is the separation: when S=1
(RA=RB) no separation is possible. Notice that for a given relative floatability the recovery of mineral B is
completely determined by the recovery of mineral A. The operational objective of a bank can be expressed
as finding the recovery of A in each cell (recovery profile) such that for a target bank recovery of mineral
A the bank recovery of mineral B is minimized. Taking as a measure of separation efficiency for two
floatable minerals as E =RA-RB (Agar et al., 1980) the operational objective is equivalent to maximizing the
separation efficiency for a given target bank recovery of mineral A. Relative floatability is assumed to be
constant down the bank and independent of the operation as observed at Brunswick Mine (Copper et al.,
2004). The optimization problem can then be expressed mathematically as follows:

Max E = R CA - R CB
R A1 , R A2 ,...R AN

subject to:
R CA = R target
(4)
R CA = R A1 + R A2 ⋅ (1 − R A1 ) + ... + R AN ⋅ (1 − R A1 )(1 − R A2 )K (1 − R AN-1 )
R CB = R B1 + R B2 ⋅ (1 − R B1 ) + ... + R BN ⋅ (1 − R B1 )(1 − R B2 )K (1 − R BN-1 )
R Aj 1 - R Bj
S = ⋅ , j = 1,..., N
1 - R Aj R Aj

where superscript c stands for cumulative.

The optimal strategy that solves this problem was found to be a flat cell-by-cell recovery profile,
i.e. each cell having the same recovery based on the feed to that cell (Maldonado et al., 2011). Figure 6
gives the general result that includes the solution for the bank recovery. This bank recovery expression is
often quoted as ‘the simplified solution’ (all Rs equal) but it appears that there may be good reason to
operate that way. To illustrate, Figure 7 shows the result for a bank comprising two cells: for two target
bank recoveries, 75% and 90%, the maximum separation efficiency occurs when the two cell recoveries
are equal, respectively 50% and 68.4%. Figure 8 shows the result for a bank of 9 cells, the flat cell-by-cell
recovery (i.e., recovery based on cell feed) and the corresponding recovery based on feed to the bank
which shows a monotonic decrease reminiscent of the froth velocity profiles set to control mass pull
(Figures 3 and 4).

1 (1-R) 2 (1-R)2 3 (1-R)3 (1-R)N-1 N (1-R)N

1

R R(1-R) R(1-R)2 R(1-R)N-1

Figure 6 – Optimal operation for two floatable minerals (flat recovery profile, i.e., each cell has same
recovery R)

216
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

0.6 0.6
S = 10 0.55 S = 10
0.55

Separation efficiency, E
Separation efficiency, E

0.5
0.5
0.45
0.45
S=5 0.4
0.4 S=5
0.35
0.35 0.3
0.3 0.25
RA target = 0.75 0.2
0.25
RA target = 0.9
S=2 0.15 S=2
0.2
0.1
0.15
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Recovery of mineral A first cell, RA1 Recovery of mineral A first cell, RA1
(a) (b)
Figure 7 – Separation efficiency vs. recovery of mineral A in the first cell for a bank composed of 2 cells
and three relative floatabilities. (a) Target cumulative recovery of mineral A equal 0.75. (b) Target
cumulative recovery of mineral A equal 0.9.

This is the result for separating two floatable minerals, which is our focus. In this case the
optimum flat cell-by-cell recovery profile result is independent of changes in rate constant along the bank
provided the relative rate constant is unchanged. If we allow for the moment that the rate constant is
unchanged then an additional interesting property emerges relevant when recovery of a single floatable
mineral is the concern (e.g. bitumen). The flat cell-by-cell recovery profile produces the maximum
cumulative bank recovery for a given installed volume. This is illustrated in Figure 9 for a 2-cell bank
where the total cell volume is fixed but the volume of each cell is altered, which is equivalent to changing
cell recovery. Setting total bank recovery when cells are equal volume (V1/V2 = 1) at 75%, we note that all
other volume combinations give less than 75%. It is logical to construct banks with cells of the same size
but Figure 9 demonstrates a fundamental case for doing so. Aris (1964) made an equivalent observation for
any chemical reactors in series, that they should all have the same residence time, or volume.

25
22.57
20
flat cell-by-cell recovery profile
Recovery (%)

15 Recovery
referred to the feed of the bank

10

5
Rc = 90%

0
1 2 3 4 5 6 7 8 9
Cell number
Figure 8 – Optimal flat cell-by-cell recovery profile for a target bank recovery of 90% giving R = 22.57%;
and recovery profile based on bank feed

217
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

0.74
Bank Recovery (%)

k = 1min-1
Rtarget = 0.75
0.72

0.7

0.68
0 1 4 8
V1/V2
Figure 9 – Two-cell bank recovery relative to bank recovery when cells are equal volume (V1/V2 = 1)

CASE STUDIES

There are few studies in which grade-recovery down a bank has been reported under different
operating conditions. This is not surprising given the effort that these surveys entail. We have found three
that give sufficient data to compare with theory.

Brunswick Mine, Canada (Cooper et al., 2004)

Figure 10(a) shows the average (over all trials) cumulative zinc recovery for the decreasing,
balanced and increasing air profiles and Figure 10(b) gives the corresponding cell-by-cell recovery (which
corrects these data given in the original paper). The increasing profile represents slowing down flotation in
the first cells and re-distributing recovery (mass) down the bank. The corresponding cell-by-cell recovery
is not the flat profile from theory but rather trends upwards along the bank. (The profiles probably reveal
that cell 5 is over-pulling.) The high initial recoveries seen with the decreasing and balanced air profiles
proved detrimental while the low recovery in the first cells of the bank with the increasing profile produced
the increase in grade that was maintained down the bank. Recovery profiling was not the initial ambition
but it does provide a new way to consider what was achieved. There is no way of knowing if the increasing
air profile used gives the true optimum. An argument for reducing recovery in the first cells below the
balanced value may be that it benefits entrainment rejection. The original analysis focused on rejection of
entrained particles in the first cells as the mechanism of grade enhancement.

218
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

80
24
70
22
Cumulative Zn Recovery

Cell by cell Recovery (%)

60
20
50
18
40
16

30
14

20 Increasing 12 Increasing
Balanced
Balanced
10 Decreasing 10 Decreasing

0 8
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Cell number Cell number
(a) (b)
Figure 10 – (a) Cumulative Zn recovery for the three air profiles. (b) Corresponding cell-by-cell Recovery
profile

Los Pelambres, Chile (Maldonado et al., 2007)

Figure 11 shows the results of two campaigns on a 9-cell rougher flotation line at Los Pelambres
Mine, Chile. The line comprises 5 banks in a 1-2-2-2-2 configuration (first bank is 1 cell followed by
subsequent banks composed of 2 cells). Figures 11(a) shows, respectively, Cu cumulative recovery and
cumulative grade down the line. The final (i.e., line) Cu cumulative recovery for both campaigns is
essentially the same (93.58 vs. 93.48%) but kinetics was slower for campaign 2. Figure 11(a) shows that
the grade in Campaign 2 starts significantly higher, and this is maintained down the line, the total line
grade being higher. Figure 11(b) shows the corresponding bank-by-bank recovery profiles. Of the two,
campaign 2 tends towards a flat recovery profile, at least up to bank 4. This more equitable distribution of
recovery in campaign 2 lends support to the theoretical analysis.

100 40 70
Cumulative Cu Recovery (%)

Campaign 2
Cumulative Cu grade (%)

35 60 Campaign 1
80
Campaign 2
Cu cell Recovery

Campaign 1
30 50
60
25 40

40
20 30

20
15 20

0 10 10
0 1 2 3 4 5 1 2 3 4 5
Bank number Bank number
(a) (b)
Figure 11 – (a) Flotation cumulative recovery and cumulative Cu concentrate grade profiles for two
sampling campaigns. (b) Bank-by-bank recovery profiles.

Northparkes, Australia (Hadler et. al., 2006)

Hadler et al.(2006) reported grade-recovery curves and air recoveries for a 4-cell bank at three
total air rates all with an increasing air rate profile. Figure 12(a) shows that the lowest total air gave the
best result and also gave highest cell-by-cell air recoveries down the bank, supporting the PAR strategy.
Figure 12(b) opens another possibility: the result at highest total air rate also gave the flattest cell-by-cell
recovery profile.

219
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

60

3.4 cm/s
50
3.7cm/s
4.2 cm/s

Recovery (%Cu)
40

30

20

10

0
1 2 3 4
Cell number
(a) (b)

Figure 12 – (a) Grade-recovery curves for three increasing air rate profiles at three total air rates (Hadler et
al., 2006); (b) Corresponding cell-by-cell Cu recovery profiles

DISCUSSION

There appears to be no agreed method of operating a bank of cells to achieve optimal

performance, however defined. Three strategies to search for the optimum have been described in recent
literature: air rate profiling, air recovery profiling, and mass-pull profiling. We introduce the notion that the
feature common to these strategies is the way recovery is distributed down the bank, the cell-by-cell
recovery profile. What has emerged is the potential benefit of a flat cell-by-cell recovery profile. While
plant data are limited, the three cases examined lend some support to this conclusion.

The mathematical analysis assumed constant relative rate floatability, S. The Brunswick Mine
data supported this assumption and it is a more defendable that assuming a constant floatability (rate
constant) (Lynch et al., 1981). What a constant S means is that changes in physical (as opposed to
chemical) variables tend to change the rate constant of different floatable minerals proportionally. Lynch et
al (1981 p. 24) noted this in the case of air rate: “Change in the flotation rate of the Mount Isa chalcopyrite
by change in aeration will change the flotation rates of all other components proportionally.” Although
not as directly stated they made similar remarks regarding froth depth (p. 41); and we might now add
bubble size to the list of physical variables. While the validity of constant S may be debated the lesson of a
flat cell-by-cell recovery profile drawn from the commonly applied kinetic model seems worth taking on
board in designing strategies for bank operation. It offers to reduce the search effort by using the flat
profile as a starting point.

Air profiling is a low cost approach to improve performance of flotation banks which is
encouraging its growing application. For cells provided with gas flowrate sensors it can be implemented
without capital investment. Note the case at Brunswick Mine, use was made of a gas velocity sensor
(Gomez & Finch, 2007). For self-aspirated cells air rate can still be controlled this time by manipulating
froth depth. Although the evidence is that an increasing air rate profile will be better than other profiles it
does not specify the best shape of the increasing profile nor the total bank air to be distributed. Brunswick
Mine set an essentially arbitrary increasing profile shape and used total bank air as another degree of
freedom to achieve a target recovery.

PAR air profiling strategy translates the problem of optimizing a bank of cells to a local problem
of optimizing each cell. The total bank air flowrate in this case is completely determined and corresponds
to the sum of the gas rate that produces PAR in each cell (JgT=Jg1+Jg2+…+JgN, see Fig. 2). Although
analysis of the data from Hadler et al. (2006) suggested the best result also corresponded to a flat cell-by-
cell recovery profile the link between PAR air profiling and recovery profiling is not clear.

220
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

A limitation in implementing the PAR strategy is the extensive measurement effort and
instrumentation required on each cell to calculate air recovery. It may be possible to find a surrogate for air
recovery such as equilibrium froth height which is another measure of froth stability.

A component in the calculation of air recovery is froth velocity and that may be used
independently as in the third strategy, mass-pull control. This now requires a search for the froth velocity
profile to achieve the target bank performance. An exponentially decreasing froth velocity profile is
usually targeted (Supomo et al., 2008; Figueroa, 2009). The form of the decay function appears to be based
on mimicking how the banks are currently operated, which is not necessarily optimal, as the air profiling
and PAR studies have shown. Taking the flat cell-by-cell recovery solution the function could be derived
and would, we contend, be a practical starting point. A problem that remains is that mass-pull rate and
froth velocity may not be uniquely related.

The analysis considers only separation between floatable minerals, not entrainment. The evidence
at Brunswick Mine is that probably both are affected by air profiling and by extension recovery profiling.
The increasing air rate profile benefits reduces air and mass-pull rate in the first cells both of which benefit
entrainment rejection (Neethling and Cillers, 2003; Zheng et al., 2006). The evidence points to either a flat
or increasing cell-by-cell recovery profile for both minimizing entrainment and maximizing selectivity
between floatable minerals.

CONCLUSIONS

Three strategies for optimizing flotation bank performance have been reviewed: air profiling, peak
air recovery profiling, and mass-pull (froth velocity) profiling. The possibility that the property underlining
these strategies is the way recovery is distributed down the bank (recovery profiling) is discussed. It is
shown that a flat cell-by-cell recovery profile maximizes separation efficiency for a target bank recovery
when relative floatability is constant down the bank. Three case studies lend some support to this as the
optimal strategy. It suggests that the flat recovery profile offers a starting point in bank optimization.

ACKNOWLEDGEMENTS

Funding of this work is under the Chair in Mineral Processing co-sponsored by Vale Inco, Teck
Cominco, Xstrata Process Support, Agnico-Eagle, Shell Canada, Barrick Gold, SGS Lakefield Research,
COREM and Flottec under the CRD (Collaborative Research and Development Program) of NSERC
(Natural Sciences and Engineering Research Council of Canada) and through the AMIRA P9O project also
under an NSERC-CRD.

REFERENCES

Agar, G.E., Stratton-Crawley, R., & Bruce, J. (1980). Optimizing the design of flotation circuits. CIM
Bulletin, pp. 173-181.

Aldrich, C., Marais, C., Shean, B.J., Cilliers, J.J. (2010). On-line monitoring and control of froth flotation
system systems with machine vision: A review. International Journal of Mineral Processing, 96
(1-4), pp. 1-13.

Aris, R. (1964). Discrete dynamic programming: An introduction to the optimization of staged processes.
Blaisdell Publishing Company.

Aslan, A., & Boz, H. (2010). Effect of air distribution profile on selectivity at zinc cleaner circuit. Minerals
Engineering, 23 (11-13), pp. 885-887.

Cooper, M., Scott, D., Dahlke, R., Finch, J.A., & Gomez, C.O. (2004). Impact of air distribution profile on
banks in a Zn cleaning circuit. CIM Bulletin, Vol. 97, No. 1083, pp. 1-6.

221
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

Figueroa, L., Peragallo, E., Gomez, A., & Orrante, F. (2009) Determination of rougher froth velocity
profiles and their implementation through expert systems. Proceedings of the VI International
Mineral Processing Seminar (Procemin), P. Amelunxen, W. Kracht and R. Kuyvenhoven (Eds),
December 2-4, Santiago, Chile, pp. 395-401.

Gaudin, A. M. (1957). Flotation. McGraw Hill, New York.

Gomez, C.O., & Finch, J.A. (2007). Gas dispersion measurements in flotation cells. International Journal
of Mineral Processing, 84, 2007, pp. 51-58.

Gorain, B. (1998). The effect of bubble surface area flux on the kinetics of flotation and its relevance to
scale-up. Ph.D Thesis, University of Queensland, Australia.

Hadler, K., Barbian, N., & Cilliers, J.J. (2006). The relationship between froth stability and flotation
performance down a bank of cells. Proceedings of XXIII International Mineral Processing
Congress, Istanbul, Turkey.

Hadler, K., & Cilliers, J.J. (2009). The relationship between the peak in air recovery and flotation bank
performance. Minerals Engineering, 22, pp. 451-455.

Hadler, K., Smith, C.D., & Cilliers, J.J. (2010a) Recovery vs. mass pull: The link to air recovery, Minerals
Engineering, Vol. 23 (11–13), pp. 994–1002.

Hadler, K., Smith, C., & Cilliers, J. (2010b), Flotation performance improvement by air recovery
optimization on roughers and scavengers. XXV International Mineral Processing Congress
(IMPC) 2010 Proceedings, Queensland, Australia.

Hernandez-Aguilar, J.R., & Reddick, S. (2007). Gas dispersion management in a copper/molybdenum

separation circuit. In: del Villar, R., Nesset, J.E., Gomez , C.O., Stradling, A.W. (Eds.), The Sixth
International Copper-Cobre conference, August 25-30, vol. 2. Toronto, Canada, pp. 173–184.

Lynch, A.J., Johnson, N.W., Manlapig, E.V., & Thorne, C.G. (1981). Mineral and coal flotation circuits:
Their simulation and control. Developments in Mineral Processing, Vol. 3, DW. Fuerstenau
(advisory editor), Elsevier Scientific Publishing Company.

Maldonado, M., Sbarbaro, D., & Lizama, E. (2007). Optimal control of a rougher flotation process based
on dynamic programming. Minerals engineering, 20, pp. 221-232.

Maldonado, M., Araya, R., & Finch, J. (2011). Optimizing flotation bank performance by recovery
profiling. Minerals Engineering, 24, pp. 939-943.

Neethling, S.J., Cilliers, J.J. (2003). Modelling flotation froths. International Journal of Mineral
Processing, Vol. 72, 267–287.

Rojas, D., & Cipriano, A. (2011). Model based predictive control of a rougher flotation circuit considering
grade estimation in intermediate cells. DYNA, Vol. 78, 166, pp. 37-47.

Sbarbaro, D., Maldonado, M., & Cipriano, A. (2008). A two level hierarchical control structure for
optimizing a rougher flotation circuit, Proceedings of the 17th World Congress, The International
Federation of Automatic Control (IFAC), Seoul, Korea, July 6-11, pp. 11018-11022.

Smith, C.D., Hadler, K., & Cilliers, J.J. (2010a). The total air addition and air profile for a flotation bank.
Canadian Metallurgical Quarterly, Vol. 49, No. 4, pp. 331-336.

222
 Proceedings of the 44th Annual Canadian Mineral Processors Conference – 2012

Smith, C.D., Hadler, K., & Cilliers, J.J. (2010b). Flotation bank air addition and distribution for optimal
performance. Minerals Engineering, 23, pp. 1023-1029.

Supomo, A., Yap, E., Zheng, X., Banini, G., Mosher, J., & Partanen, A. (2008) PT Freeport Indonesia’s
mass-pull control strategy for rougher flotation. Minerals Engineering, 21, pp. 808-816.

Woodburn, E.T, Austin, L.G., & Stockton, J.B. (1994). A froth based flotation kinetic model. Chemical
Engineering Research and Design, Vol. 72, A2:211-226.

Zheng, X., Franzidis, J.P., Johnson, N.W. (2006) An evaluation of different models of water recovery in
flotation. Minerals Engineering, Vol. 19, pp. 871-882.

223