See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/268300682

OPTIMIZATION OF FLOTATION CIRCUITS BY

MODELLING AND SIMULATIONS

Article December 2006

READS

29

4 authors, including:

A. Lopez-Valdivieso
Universidad Autnoma de San Luis Potos
84 PUBLICATIONS 1,019 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Juan Luis Reyes Bahena
letting you access and read them immediately. Retrieved on: 02 July 2016
 Proceedings of 2006 China-Mexico Workshop
on Minerals Particle Technology
San Luis Potos, Mexico

OPTIMIZATION OF FLOTATION CIRCUITS BY MODELLING AND

SIMULATIONS
J. L. Reyes Bahena1, A. Lpez Valdivieso1, E. V. Manlapig2, J. P. Franzidis2
1
Instituto de Metalurgia, Universidad Autnoma de San Luis Potos, Av. Sierra Leona 550,
Lomas 2a Seccin, San Luis Potos, S.L.P., Mxico
e-mail: [email protected]
2
Julius Kruttschnitt Minerals Research Centre, University of Queensland, Isles Rd, Indooroopilly,
QLD, Australia

Abstract
Optimization of flotation process comprises the development of suitable flotation kinetic models.
Such models account for the recovery of various minerals species based on different floatability
components of the feed stream to a bank or flotation circuit. In order to build a model, metallurgical plant
performance and batch flotation tests needs to be carry out to calibrate an appropriated form of the model
and to assess the influence of changing operating parameters on the ore floatability. The first-order
flotation rate constant includes the cell hydrodynamics as well as the feed characteristics represented by
the ore floatability of the sulfide minerals. A discrete distribution of floatability including the fast, slow
and non floating fractions were investigated for each range of minerals and particle sizes. The main
emphasis of the paper is to show the ability of prediction of the mineral behavior using a modeling and
simulation technique in order to achieve a better grade-recovery relationship in the flotation circuit and to
describe the behavior of the minerals in the circulating loads.

1. Introduction
In the past, flotation circuits were designed using trial and error sequence of laboratory
locked cycle flotation tests followed by pilot plant test. However, most of the designed circuit
based on trial and error approach are forced to be redesigned to achieve the grade and recovery
requirements. In some cases, the cost associated during commissioning could be very high and
that the main reason of the advantage in using modeling and simulation techniques to reduce
costs.
A number of contributions have been focused in developing flotation models to describe the
ore floatability in a flotation circuit. The flotation process is usually represented by a first-order
rate equation in the form:
dC(t)
- = kC (t ) (1)
dt
where C(t) is the concentration of the particles at time t, k is a first-order kinetic rate constant.
156 2006 China-Mexico Workshop on Minerals Particle Technology

It is well known that a feed stream to a flotation cell/circuit has a set of particles of different
size and will therefore exhibit a range of floatabilities. According with the literature, there are a
variety of different techniques used in modeling to incorporate the effect of particle size,
composition, and hydrophobicity on flotation rate by means of a distribution function of rate
constants as described by Chander and Polat (1994):
Continues distributed function of rate constants, such as rectangular function (Klimpel,
1984; Dowling et al., 1985), triangular function (Harris and Chakravarti, 1970), normal
function (Chander and Polat, 1994), sinusoidal function (Diao et al., 1992) and gamma
function (Imaizumi and Inoue, 1963; Loveday, 1966).
Discrete distributed function of rate constants, for example a model of two components
named fast and slow floating components (Kelsall, 1961); two floating components and one
non-floating (Jowet, 1974); and a model to account for all discrete size fractions in the
flotation feed (Huber-Panu et al., 1976).

Dowling et al., (1985) concluded that all previous models fit the data reasonably well and
they suggested that the best model will depend on the operating conditions of the flotation
process. However, the determination of the actual distribution function of the kinetic rate constant
is still missing. For that reason, other researchers are focused their investigations in developing a
more detailed distribution functions:
Mulit-distributed flotation rate constant models where each mineral is not only broken up
into different size classes but these size classes are also broken up into floatability
components either arbitrarily as above or based on liberation or surface reagent coverage
(King, 1976; Niemi et al., 1997).

Currently, optimization of flotation process has been successfully achieved through new
modeling techniques of industrial scale circuits based on the development of sub-process model.
The model development was conducted through a collaborative research project (supported by
the Australian International Mineral Research Association, AMIRA) between the Julius
Kruttschnitt Mineral Research Center (JKMRC), the University of Cape Town (UCT) and the
University of McGill relating the hydrodynamics aspects of the cells (Gorain et al., 1997; Deglon
et al., 1999; Gomez et al., 2003), froth recovery (Savassi et al., 1998; Mathe et al., 1998; Vera,
1999), ore floatability (Harris, 1997; Runge et al., 1997; Alexander and Morrison, 1998), and
entrainment (Johnson, 1972; Savassi et al., 1998). A characteristic of the current methodology is
to estimate, by regression techniques, a discrete number of floatability components in the feed
streams of the cell or circuit.
In this paper, industrial flotation circuits were surveyed in order to collect data for the
model building. Several simulations case studies were performed to predict the mineral behavior
across the circuit configuration changes.

2. The Flotation Model

The current methodology used in this paper, assumes true flotation and entrainment are the
two principal mechanisms of recovery for each component in each cell of a flotation circuit, Eq.
P S b R f (1 - R w ) + Ent R w
R=
(1 + P S b R f ) (1 - R w ) + Ent R w (2) (Savassi et al., 1998). Overall recovery is due to
both the true flotation and the recovery by entrainment. True flotation is a function of the
operating conditions in the cell (i.e. bubble surface area flux, froth recovery) and the ore
 2006 China-Mexico Workshop on Minerals Particle Technology 157

floatability of each component. Meanwhile recovery by entrainment is a function of the water

recovery in the cell and the degree of entrainment of each component.
P S b R f (1 - R w ) + Ent R w
R=
(1 + P S b R f ) (1 - R w ) + Ent R w (2)
When recovery by entrainment has no significant contribution in the overall recovery, the
recovery of each component in each flotation cell can be calculated using Eq.
P Sb R f
R=
(1 + P S b R f ) (3).

P Sb R f
R=
(1 + P S b R f ) (3)
All sub-processes models comprising the current JKMRC flotation group methodology is
being incorporated into the JKSimFloat simulator, a computer package for performing flotation
simulations (Runge et al., 2003) which was used to perform the simulation case studies in this
paper.

3. Industrial Application
The current model developed in the JKMRC flotation group is the base platform for the
JKSimFloat simulator which has been applied with success at several flotation plants, comprising
copper, gold, nickel, lead, zinc, and PGM operations. This section includes examples of industrial
applications of the JKSimFloat to confirm the validity of the current modeling methodology.

3.1 Copper Flotation Circuit

Metallurgical survey and laboratory test data were used to develop a flotation model in the
copper flotation circuit (Fig. 1). Copper and other minerals were firstly balanced. Lead, copper,
zinc, iron and sulfur assays were converted to galena, chalcopyrite, sphalerite, pyrite minerals
while the non sulfide gangue was calculated as the remainder.
158 2006 China-Mexico Workshop on Minerals Particle Technology

Preflotation Rougher Scavenger

Final
Tialings

Retreatment
Column

Hydrocyclone

Cleaner
Water
addition

Cleaner-Scavenger
Ball Mill

Final
Concentrate

Fig. 1: Copper flotation circuit

Once balanced the raw data, the adjusted assays were used to build the flotation model and
the regressed parameters to represent the base case of plant metallurgical performance (Table 1).

Table 1: Estimated ore floatability parameters in the copper flotation circuit

Ore Floatability Fraction of Ore Floatability
No-
Mineral Fast Medium Slow Fast Medium Slow
Floating
Chalcopyrite
Primario 0.7236 0.2245 0.0155 0.0364 3.0763 0.9021 0.1045
Agotativo 0.02 0.062 0 0 0.0125 0.6756 0
Pyrite
Primario 0 0.0109 0.5287 0.4604 0 1.172 0.1578
Agotativo 0 0.7393 0 0.0638 0 0.1736 0
Non-Sulfur-Gangue
Primario 0 0.0272 0.5665 0.4063 0 0.1756 0.0265
Agotativo 0 0.1157 0.1856 0.6077 0 0.1015 0.047

As shown in Fig. 2 and Fig. 3, a strong correlation between the base case and the simulation
results was obtained for the solids tonnage.
 2006 China-Mexico Workshop on Minerals Particle Technology 159

800.0

Column 1: Balanced Data

700.0

Column 2: Modeled Data (JKSimFloat)

600.0

500.0
Solid, tph

400.0

300.0

200.0

100.0

0.0
il

lit

te
il

il

il
te

ed

te

te
lit

l
te

ed

te

ils
ai
ai

ai
ta

ta

ta
ta

t
ee
sp
sp
tra

ra

tra

tra

ra
tra

ra

Ta
rt
rt

tt

fe
fe
n

er

er
n

tf
nt

nt
nt

en

ne
ge
en

en

en
en

c
nc
tio

um

er
n
gh

an
on

en
ce

ce
ce
um

ea
en

co

an

nc

nc
nc

nc
ta

ol
ou

le
on

on
C

on

at
flo

l
av

co
co

co

co
ol

le
C

ec
er

at

re
er
R

rc

lc
tc
C

C
Sc
e

re
gh

R
er

et

er

er
n

gh

na
en
ge
Pr

um

et

R
gh

an

an
R

Fi
m
en

R
ol
ou

le

le
+

at
av

ec
il

re
R

ta
Sc

R
et
ol

R
C

Fig. 2: Calibration results of a copper flotation circuit

The simulation case studies were:

o Sim 1b decreasing feed tonnage at 50%.
o Sim 2b increasing retreatment capacity.
o Sim 3b bank residence time assessment.

100.0
Chalcopyrite: R2 = 1
90.0 FeS: R2 = 0.9999
NSG: R2 = 1
80.0
Modeled Mineral Recovery, %

70.0

60.0

50.0

40.0

30.0

20.0

10.0

0.0
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Balanced Mineral Recovery, %
Chalcopyrite FeS NSG

Fig. 3: Comparison between predicted and experimental recoveries of all species en the copper flotation
circuit
160 2006 China-Mexico Workshop on Minerals Particle Technology

3.2. Decreasing Feed Tonnage

Decreasing half of the tonnage (380.5 tph) in the feed of the copper flotation circuit showed
that overall chalcopyrite recovery increased from 96.3% to 98.1% in the final concentrate (Table
2). It means an increase of 1.8% chalcopyrite. However, the final chalcopyrite grade decreases
from 74.4% to 59.4% due to an increase in iron and non-sulfide gangue recovery. This behavior
highlight that cell capacity may be an important issue to increase the grade-recovery curve in this
circuit.

Table 2: Metallurgical summary of the Sim 1b case study

Chalcopyrite Pyrite NSG
Recovery, % Grade, % Recovery, % Recovery, %
Base Case 96.25 74.39 19.31 3.43
Sim 1b: 50% of feed tonnage. 98.11 59.35 38.56 6.99

3.3. Increasing Retreatment Capacity

The additional retreatment capacity (8 extra cells) gave only 0.3% increases in chalcopyrite recovery in the
final concentrate (Fig. 4 and
Table 3). And also, the chalcopyrite grade is decreased due to an increase of the pyrite and
non-sulfide gangue (1.7% with 4 extra cells and 2.3% with 8 extra cells).

Table 3: Metallurgical summary of the Sim 2b case study

Chalcopyrite Pyrite NSG
Recovery, % Grade, % Recovery, % Recovery, %
Base Case 96.25 74.39 19.31 3.43
Sim 2b: increasing 50% capacity (4 cells) 96.51 72.74 21.71 3.69
Sim 2b: increasing 100% capacity (8 cells) 96.59 72.05 22.92 3.80
 2006 China-Mexico Workshop on Minerals Particle Technology 161

100.0
Column 1: Base Case (JKSimFloat)
90.0
Column 2: 12 Cells in Retreatment
80.0
Chalcopyrite % Recovery

Column 3: 16 Cells in Retreatment

70.0

60.0

50.0

40.0

30.0

20.0

10.0

0.0

ils
l

il

il

te
d
te

te
lit

te
te
i

ai

ta
ta

ta
ee
ee
sp
ra

tra

ra
tra
tra

Ta
tt
er

er

er
rf
tf

nt
nt

en

en

en
n
c
ng

an

an
en
on

ce
e
e

ce

an
nc

nc

nc
e

le

le

on
m

on
rC

at
av
co

co

co
le

ec
at

re

lc
tc
e

C
Sc

re

R
er

et

er

er
gh

na
en
et

R
ng

an

an
R

Fi
m
R
e

le

le
+

at
av

ec
il

re
ta
Sc

R
et
ol

R
C

Fig. 4: Chalcopyrite recovery at increased capacity of the retreatment bank

By adding four and eight flotation cells in the retreatment bank the solid flow rate in the

retreatment concentrate is increased. However, the incremental solid flow rate is rejected in the

cleaner tailings, increasing the circulating load, based on rougher feed, of slurry in the circuit

(32.3% base case to 51.9% with eight additional flotation cells) as shown in

Table 4. As observed in this table, the main components in the circulating load are those
mineral particle characterized as medium and slow floatability components which requires more
residence time to be floated (Fig. 5 and Fig. 6).

Table 4: Circulating load of chalcopyrite floatability in the copper flotation circuit

162
Conc Agotativo 2006 China-Mexico Workshop on Minerals Particle Technology
0.0

47.5

85.4

Retreatment

Colas Re-flotacin
0 50 100

92.9
Fast Medium Slow

1.8 10.9
Colas Limpia 1

100.0

0 50 100
52.5

Fast Medium Slow

14.6

0 50 100 150
Fast Medium Slow Conc Re-flotacin
7.1

89.1
98.2

0 50 100 150

Fast Medium Slow

Fig. 5: Chalcopyrite proportion in the floatability components around the retreatment bank
28.7
Conc Agot

92.4
100.0

Retreatment
Colas Re-flotacin

0 50 100 150
68.2

Fast Medium Slow

3.4
0.0
Colas Limpia 1

71.3

0 20 40 60 80
7.6

Fast Medium Slow

0.0

0 20 40 60 80

Fast Medium Slow

Conc Re-flotacin

31.8

96.6 100.0

0 50 100 150

Fast Medium Slow

Fig. 6. Non-sulfide gangue proportion in the floatability components around the retreatment bank

As observed in Fig 5 and Fig 6, the proportion of non-sulfide gangue is the main source of
the dilution of the chalcopyrite concentrate. It may be related to poorly liberated particles and
therefore further mineralogical analysis need to be carried out to validate this conclusion.
 2006 China-Mexico Workshop on Minerals Particle Technology 163

3.4. Bank Residence Time Assessment

Several simulation studies were carried out in which the number of cells in the scavenger,
cleaner, recleaner, and retreatment flotation bank was increased. The aim of these simulations
was to determine which bank in the copper circuit was residence time deficient. The
characteristics of the simulations performed are outlined in Table 5.

Table 5: Simulations performed in copper flotation circuit for bank residence time assessment (Sim 3b)
Simulation
Base Case
1 20 cells in scavenger
2 24 cells in scavenger
3 16 cells in cleaner
4 12 cells in cleaner/scavenger
5 24 cells in scavenger & 16 cells in retreatment
6 24 cells in scavenger & 16 cells in retreatment & 16 cells in cleaner

Fig. 7 and Table 6 show the assessment of the residence time in the scavenger, retreatment
and cleaner flotation banks of the copper circuit. Either increasing scavenger or retreatment
capacity will result in increasing the circulating load by rejecting more material in the cleaner
tailings.
It was observed that increasing the cell capacity in the scavenger bank is more important
than increasing retreatment capacity due to a mayor chalcopyrite recovery. It is important to
address at this point that most of the chalcopyrite recovered (as slow floatability component) in
the scavenger bank may be poorly liberated and the extra recovery of this mineral may result in a
dilution of the final chalcopyrite grade.

800 Column 1: 20 Cells in Scavenger

700 Column 2: 24 Cells in Scavenger

Column 3: 16 Cells in Cleaner

600
Solid Flowrate, tph

Column 4: 12 Cells in Recleaner

500
Column 5: 24 Cells in Scav; 16 Cells in Retreat

400 Column 6: 24 Cells in Scav; 16 Cells in Retreat; 16 Cells in Cleaner

300

200

100

0
ils
il

il

il

il
te

lit

ce il
te
a n ail

nc ed

te
te

nc ed
il

te

o n e ed

e
te
i
ta

ta

ta
ta

a
ta

t
sp

sp

tra
ra

ra
tra

tra

ra

tra

Ta
rt
t

fe
fe
er

er

er
t
n

tf

nt
nt

nt

en

ne
en
en

en

en
gh onc

c
um
tio

er
n
gh

ng

n
en
ce

on

ce
m

an lea

ea
nc

nc
ta

ol
ou

e
on

on
u

m
rc

at
flo

av

l
co
ol
co

co
co

le
C

ec
C
at

re
R

er
rc

lc
he

tc
C

C
Sc

re
e

R
et

er
er

er

na
en
ge

g
Pr

um

et

an
gh

Fi
m
en

R
ol

le
ou

le
+

at
av

ec
il

re
R

ta
Sc

R
et
ol

R
C

Fig. 7: Solid flow rate as a function of the bank capacity in the copper flotation circuit
164 2006 China-Mexico Workshop on Minerals Particle Technology

Table 6: Metallurgical summary of the Sim 3b: assessment of the bank residence time
Chalcopyrite Pyrite NSG
Recovery, % Grade, % Recovery, % Recovery, %
Base Case 96.25 74.39 19.31 3.43
20 Cells in Scavenger 96.54 74.05 19.62 3.50
24 Cells in Scavenger 96.67 73.78 19.88 3.56
16 Cells in Cleaner 96.31 72.79 21.18 3.71
12 Cells in Recleaner 96.34 71.33 22.07 4.04
24 Cells in Scavenger/16 Cells in
97.17 71.73 23.13 3.90
Retreatment
24 Cells in Scavenger/16 Cells in
97.21 69.79 25.58 4.27
Retreatment/16 Cells in Cleaner

The best recovery-grade relationship for chalcopyrite was observed in the simulation case
study 5, where the additional scavenger and retreatment capacity allowed increasing about 1.0%
recovery but decreasing the final chalcopyrite grade in 2.7%.

4. Conclusions
A discrete floatability component model accounting for the contribution in mineral
recoveries by true flotation and entrainment has been developed. The main factors affecting the
mechanism recovery inside the flotation cell have been measured and estimated as: the flotation
rate constant, the mean residence time, the bubble surface area flux, the froth recovery, the degree
of entrainment and the water recovery from the feed to the concentrate. All these sub-processes
of the flotation allow the mineral behavior in conventional mechanical cells to be predicted over a
broad range of operating conditions.

Acknowledgements
The authors wish to thank the AMIRA P9 sponsor and mine site for the collection of the
data. Juan Luis Reyes Bahena also thanks to K. Runge, X. F. Zheng for their invaluable
supporting time and discussion of the results and to CONACyT for the scholarship granted (No.
66764/122882) to pursue a PhD study.

References
1. Alexander, D.J., Morrison, R.D., (1998), Rapid estimation of floatability components in industrial flotation
plants, Minerals Engineering, 3: 4-5.
2. Chander, S., Polat, M., (1994), In quest of a more realistic flotation kinetics model, In: Proceedings of the IV
Meeting on the Southern Hemisphere on Mineral Technology and III Latin American Congress on Froth
Flotation, S. Castro and J. Alvares (Eds), Chile, 481-500.
3. Deglon, D.A., Sawyerr, F., OConnor, C.T., (1999), A model to relate the flotation rate constant and the bubble
surface area flux in mechanical flotation cells, Minerals Engineering, 12(6): 599-608.
4. Diao, J., Fuerstenau, D.W., Hanson, H.J., (1992), Kinetics of coal flotation, In: SME-AIME Annual Meeting,
Phoenix, AZ, Preprint # 92-200.
5. Dowling, E.C., Klimpel, R.R., Aplan, F.F., (1985), Discrimination in the flotation of a porphyry copper ore,
Minerals and Metallurgical Processing, 2: 87-90.
6. Gomez, C.O., Cortes-Lopez, F., Finch, J.A., (2003), Industrial testing of a gas hold-up sensor for flotation
systems, Minerals Engineering, 16: 493-501.
7. Gorain, B.K., Burgess, F., Franzidis, J.P., Manlapig, E.V., (1997), Bubble surface area flux: a new criterion for
 2006 China-Mexico Workshop on Minerals Particle Technology 165

flotation scale-up. In: Proceedings of the VI Annual Mill Operators Conference, Madang, Papua New Guinea, 6-
8 October, 1997, AusIMM, 1997 Publication series 3/97, 141-148.
8. Harris, C.C., Chakravarti, A., (1970), Semi-batch flotation kinetics : species distribution analysis, Trans. AIME,
247: 162-172.
9. Harris, M.C., (1997), A practical framework for flotation circuit modelling and simulation, In: Presentation at
SAIChE 97, the 8th National Meeting of the South African Institute of Chemical Engineers, Cape Town, 16-18
April, 1997.
10. Huner-Panu, I., Ene-Danalache, E., Cojocariu, D.G., (1976), Mathematical models of batch and continuous
flotation, In: Flotation A. M. Gaudin Memorial vol. 2, M.C. Fuerstenau (Ed.), AIME, New York, 675-724.
11. Imaizumi, T., Inoue, T., (1963), Kinetic considerations of froth flotation, In: Proceedings of the VI International
Mineral Processing Congress, Cannes, (Ed: A. Roberts), 581-605.
12. Johnson, N.W., (1972), The flotation behaviour of some chalcopyrite ores, PhD Thesis, The University of
Queensland.
13. King, R.P., (1976), The use of simulation in the design and modification of flotation plants, In: Flotation A.M.
Gaudin Memorial vol. 2, M.C. Fuerstenau (Ed.), AIME, New York, 937-961.
14. Kelsall, D.F., (1961), Application of probability assessment of flotation systems, Trans. Instn. Min. & Met.,
70:191-204.
15. Klimpel, R.R., (1984), Froth flotation: the kinetic approach, In: Proceeding of MINTEK 50, International
Conference on Mineral Science and Technology, 385-392.
16. Loveday, B.K., (1966), Analysis of froth flotation kinetics, Trans. IMM, 75: C219-C225.
17. Mathe, Z.T., Harris, M.C., OConnors, C.T., Franzidis, J.P., (1998), Review of froth modeling in steady state
flotation systems, Minerals Engineering, 12(3): 397-421.
18. Niemi, A.J., Ylinen, R., Hyotyniemi, H., (1997), On characterization of pulp and froth in cells of flotation plants,
International Journal of Mineral Processing, 51: 51-65.
19. Runge, K.C., Harris, M.C., Frew, J.A., Manlapig, E.V., (1997), Floatability of streams around the Cominco Red
Dog lead cleaning circuit, In: Proceedings of the VI Annual Mill Operators Conference, Madang, Papua New
Guinea, 6-8 October, 1997, AusIMM, 1997 Publication series 3/97, p157-163.
20. Runge, K.C., Franzidis, J.P., Manlapig, E.V., Martin, M.C., (2003), In: Proceedings of the XXII International
Mineral Processing Congress, I. Lonrezen, D.J. Bradshaw (Eds.), 29 Sept 3 Oct, Cape Town, South Africa,
973-984.
21. Savassi, O.N., Alexander, D.J., Franzidis, J.P., Manlapig, E.V., (1997), Measuring froth recovery of attached
particles in industrial flotation cells, In: Proceeding of the VI Mill Operators Conference, AusIMM, 149-155.
22. Savassi, O.N., Alexander, D.J., Franzidis, J.P., Manlapig, E.V., (1998), An empirical model for entrainment in
industrial flotation plants, Minerals Engineering, 11(3): 243-256.
23. Vera, M.A., Franzidis, J.P., Manlapig, E.V., (1999), Simultaneous determination of collection zone rate constant
and froth recovery in a mechanical flotation environment, Minerals Engineering, 12(10): 1163-1176.