Optimisation of PB-ZN Flotation
Optimisation of PB-ZN Flotation
Optimisation of PB-ZN Flotation
Circuit
S Schwarz1 and B Kilgariff2
Recleaner
Final Concentrate
EXPERIMENTAL PROGRAM (five samples per cell). The results from these surveys enabled a
‘snapshot’ of the flotation circuits to be obtained.
A methodology for analysing flotation circuit performance has Samples collected from each survey were weighed wet and dry
been developed by the flotation research groups at the Julius to determine per cent solids and sent for assay to determine lead,
Kruttschnitt Mineral Research Centre in Australia, the University zinc, silver, sulfur, iron and copper. Each survey was mass
of Cape Town in South Africa and McGill University in Canada. balanced using all available data to give the best estimates of mass
This methodology is an outcome of the Australian Minerals flows around the circuit. Least squares minimisation techniques
Industry Research Association (AMIRA) P9 Project, titled ‘The were used to estimate the flows and assays in each stream,
Optimisation of Mineral Processes by Modelling and Simulation’, ensuring all data were consistent.
which is a multi-national project currently sponsored by more than
30 companies.
Batch test procedure
The methodology captures actual plant data and applies a
semi-empirical subprocess model to represent these data for useful To map the floatability components, standard laboratory rate
simulations. The model has been described in detail previously tests were conducted on every stream in the circuit. The
(Alexander and Morrison, 1998; Alexander et al, 2000; Harris et following conditions were used in the tests:
al, 2002) and is briefly discussed later in this paper.
• Tests were conducted in the five litre JKTech flotation cell.
Individual flotation circuit models were developed for the lead
and zinc circuits at Perilya Broken Hill, with the capability to • Air rate and impeller speed were fixed for each test.
link the circuits and determine the overall flotation performance. • Six concentrates were taken over various times according to
An extensive on-site test program was conducted by site and their floatability.
JKTech personnel that included:
• Froth depth was maintained throughout each test at 1 cm.
• down the bank surveys (one in each of the lead and zinc The shallow froth depth was used to ensure the froth
circuits), recovery was approximately 100 per cent.
• batch flotation tests, • No collector, activator, pH modifier, or depressant was added
• hydrodynamic measurements, and in any test.
• froth performance measurements. • The froth pull rate from the cell was constant at six pulls per
minute.
This paper is focussed on the model developed for the lead
circuit, and as such, the results and development of the zinc • Other measurements included:
circuit model will not be discussed here. • wet weights of each concentrate,
• water addition rate during each concentrate,
Survey procedure
• cell operating impeller speed, and
Each flotation survey was planned and conducted using the
guidelines outlined by Mosher and Alexander (2002). Prior to the
• cell operating air pressure.
start of each survey, shift operating data was investigated to
determine the stability of the flotation circuit, and no operating Hydrodynamic measurements
parameters were changed during the survey period, including Hydrodynamic measurements including gas hold-up, bubble size
reagent addition rates, grinding throughput rates, cell levels, etc. and superficial gas velocity are widely considered important
The down the bank surveys conducted on each circuit were not parameters when modelling flotation cells (Alexander et al,
typical of metallurgical surveys, in that only one cut was taken of 2000). These measurements are used to calculate the bubble
each stream, and the survey period was between eight and ten surface area flux, which has a strong relationship with cell
hours. This was due to the large quantity of samples required recovery (Gorain et al, 1997).
TABLE 2
Selected mass balanced stream flows.
TABLE 3
Non-floating fraction
Changes in residence time for rougher and scavenger sections. The non-floating fraction for each mineral in the ore was
estimated using the batch test results of the tails from each
Bank Residence time (mins) circuit. The batch test results are given in Figure 2 for the lead
Survey 1 (332.7 tph) Survey 2 (330.3 tph) circuit tail.
Primary rougher 1 11.8 9.6 Although there may have been slight increases in recovery
Primary rougher 2 10.4 8.8 with further flotation time, the maximum recovery of each
mineral was calculated after 11 minutes of flotation, and related
Primary rougher 3 10.3 8.5 to the mass balanced values of these streams to give the
Secondary rougher 8.0 7.0 non-floating fractions. The non-floating fraction of galena in the
Scavenger 4.1 3.7 lead circuit was estimated to be 11 per cent.
increase the gas hold-up in the cell, and cause disturbances to the 70
froth phase (thus reducing the froth recovery). The calculated Sb 60
values were used in developing the models for this circuit.
Recovery (%)
50
40
Froth recovery results 30
20
The experimental measured froth recovery values are given in
10
Table 5. In some of the flotation cells, unstable froths and/or low
0
bubble loads prevented this technique being used. The froth
0 2 4 6 8 10 12
recovery values for these cells were fitted to experimental data in
the model development stage. Also included in this Table is the Time (mins)
froth retention time (FRT), which is calculated from Equation 3 Galena Sphalerite Chalcopyrite Pyrrhotite NSG
(Mathe et al, 2000):
TABLE 4
Hydrodynamic results.
Plant bank Jg (cm/s) ε g (%) d32 (mm) Sb (m2/m2 s)† Residence time (min per cell)
Primary roughers 0.7 - 1.5 1.9 - 6.5 1.4 29 - 61 5.1 - 5.9
Secondary roughers 0.9 - 1.8 1.9 - 4.7 1.0 56 - 111 0.9 - 1.0
Scavengers 1.7 - 2.8 5.9 - 7.2 1.4 73 - 119 0.9 - 1.0
Cleaners 0.8 - 1.3 1.6 - 14 0.7 66 - 106 2.8 - 3.3
Recleaners 1.3 - 1.5 7.8 - 10 1.2 66 - 76 4.0 - 4.6
( Pi ⋅ S b ⋅ R f ⋅ τ )(1 − Rw ) + ENT ⋅ Rw
TABLE 5 Recovery = ∑ mi (4)
Froth performance results. (1 − Pi ⋅ S b ⋅ R f ⋅ τ )(1 − Rw ) + ENT ⋅ Rw
100
The other parameters (Sb and Rf) have been defined previously.
Note that entrainment was not included in this model as sizing
galena recovery
Cumulative %
80
information was not available at the time of this paper.
60
A combination of the down the bank survey data, batch test
40 data, froth recovery measurements, bubble surface area flux
20 measurements and residence time estimates were used as inputs
0 to the model. The water recovery values were obtained from the
0 2 4 6 8 10 12 survey data, and are shown in Table 6, along with the other
Time (mins) parameters required for the development of the model. The
Secondary Rougher Feed Recalculated Feed
floatability rates (Pi) and mass fractions (mi) were estimated
Secondary Rougher Concentrate Secondary Rougher Tail
using a linear least squares fitting routine on this data (Alexander
and Morrison, 1998).
In modelling flotation circuits it is important to model the
FIG 3 - Nodal analysis for the lead secondary rougher bank. mineral flows around the circuit and not the element flows.
Previous studies on the Broken Hill ore (Runge and Manlapig,
1998) included an element-to-mineral conversion table, which
was also used for this analysis (Table 7).
Cumulative % galena recovery
120
100
80
Floatability parameters
60 The model parameter estimates for each mineral are shown in
40
Table 8.
20
Table 8 shows that the ore feeding the flotation circuit can be
characterised by three floatability components – fast, slow and
0
non-floating. It should also be noted that most of the fast floating
0 2 4 6 8 10
minerals had similar flotation rates for this ore type.
Time (mins)
The majority of the galena present in the feed (67 per cent)
Before reagent After reagent was relatively fast floating with 22 per cent slow floating and
11 per cent non-floating. Interestingly, the chalcopyrite appeared
to follow the galena in terms of flotation rates and mass fractions
FIG 4 - Recovery versus time profile for galena using the recleaner
in the feed.
tail stream before and after reagent addition.
Most of the sphalerite (94 per cent) was either slow or
non-floating with a small amount of fast floating material (six
Figure 4 shows that addition of reagent (xanthate) to the per cent), similar to the pyrrhotite. Most of the non-sulfide
recleaner tail increased the recovery of the fast floating galena, gangue was non-floating (97 per cent), although three per cent of
shown by the increased recovery for times less than two minutes. the gangue in the feed was slow floating. Any recovery of
It should be noted that the ultimate recovery of galena was the gangue (pyrrhotite or non-sulfide gangue) will primarily be due
same for both before and after reagent addition, indicating that to entrainment, which will be incorporated into the size by
the reagent addition increased the floatability of the slow floating floatability model (at a later date).
material, but did not affect the non-floating material. This
The model fit results for galena are shown in Figure 5 and
reagent addition point was therefore included in the model
development to account for the change in floatability by show an excellent agreement (correlation coefficient greater than
redistributing the mass fractions across the floatability 99 per cent) between model-fitted recoveries and experimental
components. recoveries. Similar fits were also observed for the other minerals.
It should be noted that the plant recovery data had a closer
It is also known that regrinding will affect the floatability of a
correlation than the batch test data. This was due to the higher
particular stream. Unfortunately no batch tests were performed
standard deviations placed on the batch test data during the
on the combined primary rougher tail to enable comparison of
model fitting procedure.
before and after the secondary grinding stage in this circuit.
However, this change in floatability was incorporated in the It should be noted that this model was developed on an
model development process, using the mass balanced results to un-sized basis, and no entrainment has been included.
fit the floatability parameters around this node. Size-by-size analysis is currently being conducted to develop a
more detailed model, including entrainment; however, Table 9
shows a very good comparison between the overall recovery
MODEL DEVELOPMENT results for each of the minerals.
Harris et al (2002) proposed that the overall recovery can be The reagent and regrinding nodes were incorporated in the
represented by the following equation: model development, with a redistribution of the mass fractions of
TABLE 6
Parameters and data required for development of model.
Note that scavenger cell 2 was not operating at the time of the surveys
80
in:
60
• feed throughput (assuming that floatability remains constant
40 and residence time varies);
20 • bank residence time;
0 • cell operating parameters, eg air flow rate, froth depth, etc;
0 20 40 60 80 100 120 and
Mass Balanced (%)
• circuit stream destination.
Plant data Batch test data
In this study, the simulations conducted involved changes to
the stream destinations and operating parameters. The flotation
FIG 5 - Lead flotation circuit model fit results for galena. performance was estimated from these changes. There was also a
TABLE 8
Lead circuit model floatability parameters.
120 TABLE 9
100 Lead circuit model recovery results.
Mass fraction (%)
80
Mineral Overall recovery (%) Overall grade (%)
60
MBal Calc MBal Calc
40 Solids 3.84 3.95 12.8 13.1
20 Galena 79.3 79.4 82.7 80.4
0 Sphalerite 3.79 3.76 11.6 11.2
Chalcopyrite 68.4 67.9 6.5 6.3
na
ite
G
ite
r it
NS
er
e
ot
py
al
r rh
ha
G
co
al
Ch
40
ite
G
ite
NS
er
e
ot
py
al
r rh
ha
co
Py
Sp
al
Ch
being investigated). Note also that the base case results were
slightly different to those reported in Table 9 as the water
Fast Slow Non-floating
recovery was calculated based on the concentrate solids-water
flowrate according to Equation 5:
FIG 7 - Change in floatability component distribution across the
reagent addition node – solid columns indicate mass fractions
Qwater = a × Qsolids
b
(5)
prior to reagent addition, dotted columns indicate mass fractions
after reagent addition.
Both the simulated and experimental results indicated that
operating at the reduced feed rate should increase the lead
large scope to continue the study at Perilya Broken Hill in
recovery with a slight decrease in final concentrate lead grade.
investigating the effect of optimising cell operating parameters,
This was due to the increased residence times throughout the
as well as reducing the recirculating loads. However, this cell
operating optimisation was beyond the current project scope and circuit. A series of simulations can be conducted to estimate a
has not been included in this paper. range of feed throughput rates and generate a grade-recovery
curve on an unsized basis.
Reduced feed tonnage There was good agreement between the experimental trends
and simulated trends. However, although the trends are similar,
The model was developed based on a feed flow rate of there was a difference between the concentrate recoveries and
approximately 110 tph per grinding line. Plant personnel were grades. The main reason for the different results was because the
interested in investigating the effect of operating the circuit with feed grade (and hence floatability) changed between the two
a feed flow rate of 100 tph per grinding line. Simulations were surveys, with 3.5 per cent Pb in Survey 1 (330 tph) and 2.6 per
performed using JKSimFloat V6 for both feed tonnages, and the cent Pb in Survey 3 (300 tph). There was also a change in circuit
REFERENCES Laplante, A R, Toguri, J M and Smith, H W, 1983. The effect of air flow
rate on the kinetics of flotation Part 2: the transfer of material from
Alexander, D J, Franzidis, J P and Manlapig, E V, 2003. Froth recovery the froth over the cell lip, International Journal of Mineral
measurement in plant scale flotation cells, Minerals Engineering, Processing, 11:221-234.
16:1197-1203. Mathe, Z T, Harris, M C and O’Connor, C T, 2000. A review of methods
Alexander, D J and Morrison, R D, 1998. Rapid estimation of floatability to model the froth phase in non-steady state flotation systems,
components in industrial flotation plants, Minerals Engineering, Minerals Engineering, 13(2):127-140.
11:133-143. Mosher, J and Alexander, D J, 2002. Sampling high throughput grinding
Alexander, D J, Runge, K C, Franzidis, J P and Manlapig, E V, 2000. The and flotation circuits, in Proceedings SME Mineral Processing Plant
application of multi-component floatability models to full scale Design, Practice and Control Conference, pp 63-76 (SME:
flotation circuits, in Proceedings Seventh Mill Operators’ Vancouver).
Conference, pp 167-178 (The Australasian Institute of Mining and Power, A, Franzidis, J P and Manlapig, E V, 2000. The characterisation of
Metallurgy: Melbourne). hydrodynamic conditions in industrial flotation cells, in Proceedings
Engelbrecht, J and Woodburn, E T, 1975. The effects of froth height, Seventh Mill Operators’ Conference, pp 243-255 (The Australasian
aeration rate, and gas precipitation on flotation, Journal of the South Institute of Mining and Metallurgy: Melbourne).
African Institute of Mining and Metallurgy, pp 125-132. Runge, K and Manlapig, E V, 1998. An investigation of stream
Feteris, S M, Frew, J A and Jowett, A, 1987. Modelling the effect of froth floatability in Pasminco’s Broken Hill lead circuit, AMIRA P9L
depth in flotation, International Journal of Mineral Processing, report.
20:121-135. Savassi, O N, Alexander, D J, Franzidis, J P and Manlapig, E V, 1998. An
Gorain, B K, Franzidis, J P and Manlapig, E V, 1995. Studies on impeller empirical model for entrainment in industrial flotation plants,
type, impeller speed and air flow rate in an industrial flotation cell – Minerals Engineering, 11(3):243-256.
part 2: effect on gas hold-up, Minerals Engineering, 8:1557-1570. Thorne, C G, Chilman, G F and Lean, P J, 1983. The design of an
Gorain, B K, Franzidis, J P and Manlapig, E V, 1996. Studies on impeller incremental grinding circuit expansion at New Broken Hill
type, impeller speed and air flow rate in an industrial flotation cell – Consolidated Limited, in Proceedings The AusIMM Annual
part 3: effect on superficial gas velocity, Minerals Engineering, Conference (The Australasian Institute of Mining and Metallurgy:
9:639-654. Melbourne).
Gorain, B K, Franzidis, J P and Manlapig, E V, 1997. Studies on impeller Tilyard, P A, 1991. Developments at Pasminco Mining’s South
type, impeller speed and air flow rate in an industrial flotation cell – Concentrator, The AusIMM Bulletin.
part 4: effect of bubble surface area flux on flotation performance, Tucker, J P, Deglon, D A, Franzidis, J P, Harris, M G and O’Connor, C T,
Minerals Engineering, 10:367-379. 1994. An evolution of a direct method of bubble size distribution
Harris, M C, Runge, K C, Whiten, W J and Morrison, R D, 2002. measurement in a laboratory batch flotation cell, Minerals
JKSimFloat as a practical tool for flotation process design and Engineering, 7(5/6):667-680.
optimisation, in Proceedings SME Mineral Processing Plant Design, Vera, M A, Franzidis, J P and Manlapig, E V, 1999. Simultaneous
Practice and Control Conference, pp 461-478 (SME: Vancouver). determination of collection zone rate constant and froth zone
Hart, E K, 1983. The Barrier Mines – a historical outline, in Proceedings recovery in a mechanical flotation environment, Minerals
The AusIMM Annual Conference, pp 1-10 (The Australasian Institute Engineering, 12(10):1163-1176.
of Mining and Metallurgy: Melbourne).