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

net/publication/303826132

The effect of cell hydrodynamics on flotation kinetics

Thesis · December 2012

CITATIONS READS

6 441

1 author:

Erico Tabosa
Hatch
37 PUBLICATIONS 195 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

The effect of cell hydrodynamics on flotation View project

All content following this page was uploaded by Erico Tabosa on 23 January 2019.

The user has requested enhancement of the downloaded file.

 International Journal of Mineral Processing 156 (2016) 99–107

Contents lists available at ScienceDirect

International Journal of Mineral Processing

journal homepage: www.elsevier.com/locate/ijminpro

The effect of cell hydrodynamics on flotation performance

Erico Tabosa ⁎, Kym Runge, Peter Holtham
Metso — Process Technology and Innovation, 1 Technology Court, Pullenvale, Queensland 4069, Australia

a r t i c l e i n f o a b s t r a c t

Article history: It is clear that along with gas dispersion characteristics, the energy dissipated by the impeller is process determin-
Received 15 February 2016 ing in flotation, and its effect on flotation kinetics has been widely studied. However, turbulent conditions inside a
Received in revised form 27 May 2016 flotation cell have usually only been changed by varying impeller speed or air flowrate or both. Therefore, there is
Accepted 30 May 2016
a need to investigate not only these variables but also how changes in impeller/stator mechanism size and design,
Available online 1 June 2016
cell aspect ratio and cell design affect turbulence. This should lead to a better understanding of the effect of cell
Keywords:
hydrodynamics on flotation performance. The aim of this work was to evaluate the role of flotation cell hydrody-
Flotation cell hydrodynamics namics on flotation performance in a fully instrumented 3 m3 cell. The cell was operated at a copper concentrator
Collection zone flotation rate in Australia with different combinations of airflow rates, impeller speeds and sizes and cell aspect ratio providing
Turbulence a wide range of hydrodynamic conditions. An analysis of the flotation cell performance showed that the overall
Froth recovery copper recoveries were very similar for the conditions tested. However, by decoupling pulp effects from froth ef-
Flotation cell aspect ratio fects it was possible to determine whether the changes made affected the pulp zone and/or froth zone responses.
The analysis showed that the overall recovery had the potential to be up to 10% higher if not limited by froth re-
covery. Comparing the metallurgical performance of the cell with the different hydrodynamic conditions it was
found that the collection zone flotation rate was not directly related to overall energy dissipation, as is commonly
observed at laboratory scale, where energy is usually only changed by varying impeller tip speed. Results suggest
that it is the size of the turbulent zone, rather than just energy input, that affects flotation recovery.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction quiescent zone is believed to be important for flotation as it minimises

Three distinct zones can be identified within a mechanical flotation
cell: the turbulent, quiescent and froth zones. It is known that the highly
turbulent zone is responsible for the key sub-processes (Schubert and
Bischofberger, 1978, 1979; Schubert et al., 1982; Schubert, 1985, 1986,
1999, 2008; Weiss and Schubert, 1988; Weber et al., 1999):
• Suspension of the particles in the pulp;
• Feeding and dispersion of the air into bubbles;
• Mixing of the aerated pulp for better reagent distribution and condi-
tioning; and
• Promotion of particle-bubble collisions (the basic requirement for
particle-bubble attachment).
Too much turbulence can have an adverse effect on flotation because
it results in the detachment of particles from bubbles as they rise in the
pulp, as well as causing destabilisation and loss of recovery within the
froth phase. Thus, a quiescent zone, less energy intensive than the tur-
bulent zone, provides conditions for detaching entrained or entrapped
gangue particles from aggregates created in the turbulent zone. The
⁎ Corresponding author.
the detrimental effects of turbulence on the froth phase.
Turbulence in mechanical flotation cells has been quantified by im-
peller speed or power input by many researchers (Ahmed and Jameson,
1985; Deglon et al., 2000; Schubert and Bischofberger, 1979; Schubert,
1985, 2008; and many others), but few have considered the parameters
necessary for a more comprehensive characterisation of turbulence.
Exceptions are Jordan and Spears (1990); Pyke et al. (2002, 2003)
and Duan et al. (2003) who measured turbulence parameters such as
turbulent fluctuating velocities and energy spectra in batch flotation
cells and incorporated these results into micro-kinetic models for flota-
tion in agitated systems. The authors considered the turbulent microen-
vironment in the flotation cell to have a significant impact on both
bubble breakup and bubble-particle contact.
Schubert and Bischofberger (1978) found that an increase in recov-
ery occurred with increasing impeller speed and hence established op-
timum power inputs for different particle size fractions in a cassiterite
flotation plant. An optimum in the flotation rate as a function of impeller
speed, particle size, density, and bubble size was found by Ahmed and
Jameson (1985).
Other similar studies, showing an increase in flotation rate of fine
particles with an increase in impeller speed, can be found in the litera-
ture for different ore types: galena (Spears and Jordan, 1989), chalcopy-
rite (Jordan and Spears, 1990) and coal (Cheng et al., 1995).

http://dx.doi.org/10.1016/j.minpro.2016.05.019
0301-7516/© 2016 Elsevier B.V. All rights reserved.
100 E. Tabosa et al. / International Journal of Mineral Processing 156 (2016) 99–107
The energy dissipation, or specific power input (ε), may be consid-
ered equal to the average energy dissipation in a flotation machine
(Schubert, 1986, 2008), which is related to the power input (P) of the
impeller into the total mass of the fluid system (m) by:
P 2. Methodology
ε¼
m
This equation shows that achieving high local energy dissipation
rates in the impeller stream is dependent on the power input to the im-
peller. The degree of turbulence is at a maximum in the impeller region
and is significantly reduced in the quiescent zone above the impeller
(Wu and Patterson, 1989; Rutherford et al., 1996; Pyke et al., 2002,
2003; Newell and Grano, 2007). Schubert and Bischofberger (1979)
suggest that, in the highly turbulent zones of the cell, values of ε may
be 5 to 30 times higher than the average value.
The impeller-stator system area and its immediate vicinity represent
the high intensity zone, where air is sheared and bubbles are generated,
and significant turbulence and air void fraction are present. This zone
also represents the area where particles and bubbles come into first
contact. These factors suggest this zone presents the highest probability
of collision, and thus the highest flotation kinetics (Koh et al., 2000).
Hence, high collision rates necessitate the use of an efficient impeller-
stator system (Schubert, 2008).
The effect of the high intensity turbulence on flotation recovery has
been widely studied by Schubert and co-workers (Schubert et al., 1982;
Schubert and Bischofberger, 1998; Schubert, 1999; and others). They in-
vestigated recovery in the high intensity zone by means of high-speed
photography, showing that particle-bubble collection occurs almost
solely in this zone of high local energy dissipation.
It is clear that along with gas dispersion characteristics, the energy
dissipated by the impeller is process-determining in flotation. Its effect
on flotation kinetics has been widely studied (Schubert and
Bischofberger, 1978; Jameson and Ahmed, 1983; Deglon et al., 1999;
Pyke et al., 2003; Newell and Grano, 2006).
Conventional mechanical flotation cell size has increased significant-
ly over the last 20 years. However, as cell size has increased, most man-
ufacturers have kept the aspect ratio of the cell constant to maintain
geometric consistency. This has resulted in cells with a highly turbulent
region near the impeller and very large quiescent zones, which poten-
tially result in very little flotation recovery. There may be an opportunity
to increase cell recovery rates by changing the volume ratio between
the high and low turbulent zones in the cell.
Most of the studies performed to investigate the effect of energy on
flotation response have usually only involved changing impeller speed
and/or air rate. Only limited laboratory scale work has been reported
in the literature evaluating the size of the high turbulent zone and the
ratio between high and low turbulent (or quiescent) zones on flotation
performance. Earlier studies (Hall and Dell, 1989; Morris and
Matthesius, 1988) on the effect of the ratio of quiescent to turbulent
zone volumes in batch tests concluded that turbulence appears to play
an important role in promoting particle-bubble attachment, as well as
the generally accepted roles of air diffusion and suspension of solids.
Both studies suggested that the ratio between high and low turbu-
lent zones within a flotation cell affects performance. Hall and Dell
(1989) showed that in a cell with a larger turbulent zone, i.e. a shorter
cell, and at a certain optimum air rate, higher flotation rates were ob-
tained. However, Morris and Matthesius (1988) found that for coarser
particles (N 150 μm) the beneficiation efficiency was enhanced when
flotation was carried out in a cell double in height.
There would therefore seem to be opportunities to improve flotation
recovery rate by optimising the distribution of turbulence in a flotation
cell. The work described in this paper was performed to evaluate the
role of flotation cell hydrodynamics on flotation performance in a fully
instrumented 3 m3 cell. The cell was operated with different combina-
tions of airflow rates, impeller tip speeds and sizes and cell aspect
ratio providing a wide range of hydrodynamic conditions. The objective
was to gain a better understanding of how flotation cell operational var-
iables, which change both the energy inputted to the flotation cell and
how this energy is distributed, impact flotation recovery.
ð1Þ
2.1. The experimental rig and test work program
The experimental work was carried out in Metso's flotation test rig
at the Northparkes copper/gold concentrator, located in the central
west of New South Wales, Australia. The test rig consists of two modules
that can be disassembled to facilitate transportation (Fig. 1). A 3 m3
Metso RCS flotation cell is located on the upper platform, with concen-
trate and tailings being discharged via gravity to a product sump tank
located on the bottom platform. The staircase and the grid mesh plat-
form enable personnel to safely access the top of the cell, where differ-
ent measurements can be carried out in the flotation cell.
The 170 cm diameter RCS flotation cell incorporated a variable side-
wall height, allowing comparison of a standard aspect ratio cell (145 cm
height), with a lower aspect ratio cell (110 cm height, 2.5 m3 volume).
By reducing the cell height, the ratio between high and low shear
zones in the cell was increased. The cell geometries are illustrated in
Fig. 2.
One of the main advantages of the Metso rig is that it incorporates
easily accessible full stream sampling points and is fully instrumented
to allow stable operation, essential requirements for metallurgical test
work (Runge et al., 2007).
The rig can be operated in three modes: open circuit, full recycle and
tailing recycle (Runge et al., 2007). The configuration used for this test
work was full recycle mode, in which the concentrate and tailing
streams are mixed in the product sump and then recycled back as
new feed. In this case the cell operates with nominally unchanging
feed properties. The measured feed flow is kept constant by controlling
the product sump pump speed. Feed flow rate was maintained at
80 m3 h−1 throughout the test work, resulting in a flotation residence
time of 2.3 min which is typical of that observed in industrial flotation
cells. A peripheral launder with froth crowder was used to collect the
froth.
Slurry to fill the test rig was taken from the flotation feed of Module
1 of the Northparkes concentrator. Flotation feed typically assayed 0.3%
Cu at approximately 30% solids by mass, with a P80 of 100 μm. When
processing the flotation feed material, the reagent suite added to the
test rig was identical to that used in the plant (PAX, frother, NaHS and
promoter).
At the start of each day the rig was filled with fresh slurry from the
plant. The rig was then operated at a series of different test conditions
and measurements performed. A factorial design was used in which
each of the following variables was tested at two or three levels:
• Cell aspect ratio (height:diameter): 0.85 and 0.65;
• Impeller design: RCS3 (standard impeller for a 3 m3 Metso RCS flota-
tion cell) and RCS5 (oversized impeller, i.e. impeller for a 5 m3 Metso
RCS flotation cell);
• Impeller tip speed: 3.5 m s−1 and 6.2 m s−1;
• Air flowrate: 100 and 140 m3 h−1
• Froth depth: three different froth depths, (typically 5, 9 and 13 cm).
A total of 103 test conditions were evaluated during the test cam-
paign. This included the 48 test conditions of every combination of the
factor levels in a full factorial experimental design; and also reproduc-
ibility testing.
To change between aspect ratios, the current sidewall was unbolted,
removed with a crane and replaced with a new sidewall. To change
between impellers, the current impeller and stator had to be unbolt-
ed and the new combination bolted into place. These modifications
were time consuming and difficult, logistically and therefore there
 E. Tabosa et al. / International Journal of Mineral Processing 156 (2016) 99–107 101

Fig. 1. Metso 3 m3 test rig installed at Northparkes copper/gold concentrator.

was no choice but to perform all testwork associated with a particu-
lar impeller/aspect ratio in a block before moving onto a new combi-
nation. Randomisation of the test sequence was therefore not
possible.
Each day of test work, six conditions were evaluated for a particular
impeller/aspect ratio combination. A particular air rate, impeller tip
speed combination would be randomly chosen and then the pulp level
(froth depth) would be set to a shallow, intermediate and deep froth
depth. Another combination of impeller tip speed and air flowrate was
then set and tests were again run for different froth depths. The aim
was to perform tests at different froth depths for a particular combina-
tion of variables on the same day over a relatively short time interval
to maximise the accuracy of the technique used to estimate collection
and froth zone recoveries (outlined below).
Besides the six conditions tested per day, a standard condition was
set at the beginning of each day in order to assess the day-to-day repro-
ducibility of the tests. The standard condition was 3.5 m s−1 impeller tip
speed, 100 m3 h−1 air flowrate and 9 cm froth depth. This condition was
also repeated six times during one testing day to assess reproducibility
of the testing procedure over a day of operation.
The main response variables measured at each condition were as
follows:
• Metallurgical performance (feed, concentrate and tail copper grades);
• Volumetric flowrates (feed and concentrate);
• Overall copper recovery (Ro);
• Overall flotation rate (k);
• Collection zone flotation rate (kc);
• Froth recovery (Rf);
• Collection zone recovery (Rc); and
• Power input to the impeller and overall energy dissipation
characterisation.
To enable the concentrate grade and recovery to be determined,
samples of the feed to the flotation cell, concentrate and tailings were
collected. All samples were weighed wet, filtered and dried, weighed
dry and analysed for copper in the Northparkes laboratory.
A mass balance was performed for each test condition to establish
unmeasurable flows and to make all measurements consistent. This
was done by using the measured feed, concentrate and tails assays
and solids concentration data as well as the measured concentrate and
feed flowrates. This balance was set up in a Microsoft Excel spreadsheet
with Solver used to find the flows/assays that resulted in a minimum
sum of squares error between the experimental data and the mass bal-
anced data.

Fig. 2. (a) High and (b) low cell aspect ratio configurations.
102 E. Tabosa et al. / International Journal of Mineral Processing 156 (2016) 99–107
2.2. Analysis
Using the mass balance data, the overall copper recovery (Ro) was
determined by the ratio between the mass balanced copper flow rate
in the concentrate and the copper flow rate in the feed:
mCu concentrate
Ro ¼  100
mCu feed
The overall flotation rate (k), was calculated from the overall recov-
ery by assuming the system is a perfect mixer (Finch and Dobby, 1990),
in which k is a function of the cell residence time (τ) and the recovery
(R) measured in the process:
R represents the loss in recovery within the froth itself and does not in-
k¼
τ  ð1−RÞ
Both cell aspect ratio tests were operated with the same feed flow
rate (80 m3 h−1). Due to the lower cell volume, the low cell aspect
ratio resulted in a shorter residence time (1.9 min, versus 2.3 min for
the high aspect ratio). Therefore, to allow a consistent comparison, the
overall flotation rates were first calculated using the actual residence
time of each test and the measured overall copper recovery (Eq. (3)).
The overall flotation rate was then used to recalculate the overall copper
recovery in the low cell aspect ratio for a residence time of 2.3 min (Rτ=
2.3 min), as follows:
kτ
Rτ¼2:3 min ¼ ð4Þ
1þkτ
The determination of the collection zone flotation rate (kc) was
achieved by operating the cell at three different froth depths for each
set of combination of cell aspect ratio, impeller size, impeller tip speed
and air flowrate.
To determine the froth depth associated with each cell level, the
total froth depth was measured at each cell condition using a ball
float. All the measured froth depths were correlated with the cell
level, and a linear model was built for each cell aspect ratio. The froth
depth was then predicted based on the instrumented cell level.
For each froth depth, the overall flotation rate (k) was determined as
described in Eq. (3). Thus, by applying the linear relationship between
overall flotation rate and froth depth (Fig. 3) observed by Feteris et al.
(1987) and Vera et al. (1999), the collection zone rate (kc) could be pre-
dicted by extrapolating this relationship to zero froth depth. At this
point, there should be no loss in recovery due to the froth phase and
therefore the flotation rate at zero froth depth should be the collection
zone flotation rate.
Fig. 3. Overall flotation rate versus froth depth relationship used to estimate the collection
zone flotation rate constant (after Vera et al., 1999).
Vera et al. (1999) showed that froth recovery can be defined as the
total rate of transfer from the pulp to the concentrate (k) divided by
the rate of transfer from the pulp to the froth phase (kc):
k
Rf ¼ ð5Þ
kc
ð2Þ
Froth recovery at each point on Fig. 3 can therefore be calculated by
dividing the measured overall flotation rate (k) at each condition by the
interpolated collection zone flotation rate (kc). This method enables
pulp effects to be decoupled from froth effects, and hence it is possible
to evaluate whether the tested changes either affect the pulp zone or
froth zone responses. It should be noted that this technique, as demon-
strated by Seaman et al. (2006), estimates a froth recovery that only
ð3Þ clude any losses that might occur at the pulp-froth interface as the par-
ticle-laden bubbles decelerate and form the froth.
To be able to compare the overall cell performance with both froth
and collection zone performance simultaneously, a new method of rep-
resentation was adopted. This involved plotting overall flotation rate (k)
against froth recovery (Rf), calculated using Eq. (5). The magnitude of
the collection zone flotation rate (kc), by re-arranging Eq. (5), is the
slope of the fitted line that passes through the experimental data points
and intercepts zero (Fig. 4). Hence, the steeper the slope the higher the
collection zone flotation rate (kc).
3. Results
3.1. Overall cell performance and reproducibility assessment
The experimental data collected generally proved to be very consis-
tent and could be mass balanced relatively easily in terms of solid flow
rate, percent solids and copper grades.
In order to assess the reproducibility of the tests performed using the
test rig, the rig was operated for a period of 6 h using the same operating
conditions, first when the cell was in the high aspect configuration and
then again when the cell was modified to a low aspect configuration.
Samples in each case were taken at one-hour intervals. There was
some evidence of changing operation over the six hour period of testing,
with variation in the copper feed grade, a decrease in the concentrate
solids flow rate and an increase in the copper concentrate grade. Inter-
estingly, however, the copper overall recovery remained relatively con-
stant. It was concluded that the change in results were a consequence of
a reduction in water recovery which reduced entrainment flow.
This paper involves an analysis of the copper overall recovery, which
was reproducible: copper overall recovery and overall flotation rate re-
sults had a standard deviation of around 1.3% and 0.08 min−1,
respectively.
Fig. 4. Overall flotation rate and froth recovery relationship, where “a” has higher
collection zone flotation rate than “b”.
 E. Tabosa et al. / International Journal of Mineral Processing 156 (2016) 99–107
Reproducibility was also assessed by running repeat tests for a num-
ber of different operating conditions (with three froth depths per condi-
tion) on different days of operation. Results showed a standard
deviation of 1.7% for overall copper recovery and 0.12 min−1 for overall
flotation rate values.
Finally, reproducibility was also assessed by collecting samples from
a standard condition sampled at the start of each day of testing. The
standard deviation of these results was 2% for overall copper recovery
and 0.31 min−1 for the overall flotation rate. This bigger variation for
the day-to-day reproducibility tests is probably due to variation created
by changes in operation of the Northparkes plant (i.e. grind size, reagent
use). In summary, it is concluded that the test rig was capable of produc-
ing repeatable copper recovery results.
Some variability in head grade between tests was expected due to
day-to-day changes in the feed material to the rig. Variability in head
grade on the same day was however not expected to be high because
the rig was operated in recycle mode. It was therefore anticipated that
the intra-day head grade would remain nearly constant.
However, the head grade was found to vary between test conditions
within each test day. In agreement with previous studies (Runge et al.,
2007), an analysis of the data showed that the head grade variability be-
tween tests on the same day was not random. The head grade was
found to be strongly influenced by the froth depth, especially for the
low cell aspect ratio. Deeper froths were associated with lower head
grades. Copper is distributed between three regions of the test rig –
the sump, the pulp phase of the flotation cell and the froth phase of
the flotation cell. It is believed that deeper froths result in more copper
being contained within the froth phase of the flotation cell thereby caus-
ing a drop in the average grade of the feed in the sump (since the total
amount of copper in the system must remain constant).
Variability in head grade would make analysis of the metallurgical re-
sults problematic as it may have a significant impact on flotation perfor-
mance. Therefore, in order to compare the metallurgical performance
with no variation in copper feed grade, a model was developed for overall
flotation rate (k) which incorporated a copper feed grade parameter (see
below). Feed grade could then be kept constant in subsequent analysis.
3.2. Overall flotation rate and collection zone flotation rate modelling
In order to compare the metallurgical performance with no variation
in copper feed grade, a model was developed for overall flotation rate
(k) and collection zone flotation rate (kc). Overall and froth recoveries
can then be calculated using Eqs. (4) and (5) that are directly compara-
ble, allowing evaluation of whether the variables changed in the test
program affect the pulp or froth zone responses or both.
Multiple regression was used to build the models. Stepwise linear
regression was chosen, as it generates an optimal equation for
predicting a dependent variable from several independent variables.
The dependent variables were the overall flotation rate (k) and collec-
tion zone flotation rate (kc). The independent variables were the cell as-
pect ratio (AR), impeller tip speed (IS), impeller mechanism size (ISz),
air flow rate (Air), froth depth (FD) and copper feed grade (Grade).
The systematic iterative construction of the regression model in-
volved the selection of the independent variables, and their interactions,
by including all variables in the model and eliminating those that are
not statistically significant (95% confidence interval). Minitab statistical
software was used to optimise the model. By successively adding or re-
moving variables based on the t-test statistics of their estimated coeffi-
cients the linear model for overall flotation rate (k) and collection zone
flotation rate (kc) were found (Eqs. (6) and (7)).
k ¼ −5:66 þ 13:75AR þ 7:64Grade−0:18ðAR  ISÞ−0:58ðAR  ISzÞ
−0:033ðAR  AirÞ−0:16ðAR  FDÞ−11:58ðAR  GradeÞ
ð6Þ
þ 0:018ðIS  ISzÞ−0:24ðIS  GradeÞ þ 0:0019ðISz  AirÞ
þ 0:00093ðAir  FDÞ þ 0:037ðAir  GradeÞ
103
kc ¼ 8:86−6:92AR−0:30IS þ 1:25ISz−0:036Air−1:37ðAR  ISzÞ
þ 0:043ðAR  AirÞ ð7Þ
Fig. 5 shows the predicted and experimental results for overall flota-
tion rate (k) and collection zone flotation rate (kc). The dashed lines
show the 95% prediction interval of the model. It can be seen that the
model fits the experimental results very well.
An analysis of the error of the regression was also done. The standard
error of the regression (S) is used as a measure of model fit in regression.
S is measured in the units of the response variable and represents the
standard distance data values fall from the regression line, or the stan-
dard deviation of the residuals. In this case, S was equal to 0.064 and
0.2 min−1 for k and kc, respectively. This means a calculated confidence
interval (2S) of 0.13 and 0.4 min−1 on the predicted k and kc results, re-
spectively. The standard deviation in overall flotation rate (k) is in
agreement with the error calculated using the reproducibility tests.
Therefore, this model can be used to eliminate the influence of the var-
iation in copper feed grade observed during testing and to investigate
the results of the test work.
This was done by fixing the copper feed grade (‘Grade’) in Eq. (6) at
0.3% Cu. This value corresponds to the average copper feed grade during
the Northparkes campaign. Once the variation in copper feed grade was
eliminated, the overall flotation rate (k) and collection zone flotation
rate (kc) were predicted for different operational conditions. By doing
so, the overall copper recovery and froth recovery could also be predict-
ed using Eqs. (4) and (5) with cell residence time (τ) equal to 2.3 min.
The standard deviations of the calculated overall copper recovery
and froth recovery were estimated using a Monte Carlo technique.
This procedure is a practical and efficient mean of estimating error prop-
agated from input parameters of known standard deviations (Ogilvie,
1984). It randomly creates a normally distributed dataset based on the
input's mean and standard deviation.
The technique was used to create 1000 random values of k and kc.
Overall copper recovery and froth recovery were then predicted and
their standard deviations calculated. It is estimated that overall copper
recovery would have a standard deviation of 2.1%, which is in agree-
ment with the standard deviation found from the analysis of the repro-
ducibility tests. Froth recovery was calculated to have a standard
deviation of 10.6%. The standard deviation of froth recovery is expected
to be higher than that of overall copper recovery. This is because froth
recovery has been calculated based on an extrapolation method that
will be sensitive to any error in the froth depth and overall flotation
rate measurements.
3.3. Effect of cell hydrodynamic on flotation performance
Plotting overall flotation rate against froth depth (as described for
Fig. 4) allows the interactions between overall cell performance and
both froth and collection zone performance to be evaluated
simultaneously.
3.3.1. Effect of impeller size
Fig. 6 shows that for every cell aspect ratio, impeller tip speed com-
bination, the collection zone flotation rate of the oversized RCS5 impel-
ler mechanisms was higher than the standard RCS3. This difference was
more significant for the low cell aspect ratio (around 30% and 8% higher
for low and high cell aspect ratio respectively). It is anticipated that, for
the same impeller tip speed, the oversized impeller mechanism will dis-
sipate more energy than the standard impeller mechanism. Therefore, it
will result in higher local energy dissipation rates in the impeller stream
and thus higher particle-bubble collision rates (Schubert, 2008).
However, froth recovery tends to be lower for the oversized impeller
(RCS5) than the standard RCS3 mechanism and this is presumably due
to instability of the froth base caused by greater turbulence near the
104 E. Tabosa et al. / International Journal of Mineral Processing 156 (2016) 99–107
Fig. 5. Predicted versus experimental overall flotation rate (k) and collection zone flotation rate (kc).
froth pulp interface. This resulted in lower overall recoveries for the
oversized RCS5 impeller.
3.3.2. Effect of impeller tip speed
Froth recovery was not overly affected by impeller tip speed. For low
and high aspect ratio, the collection zone flotation rate at high impeller
tip speed was lower than at low impeller tip speed (Fig. 7).
The reduction in collection zone flotation rate with an increase in
impeller speed is not consistent with the increase in collection zone flo-
tation rate constant observed with an increase in impeller size. Both an
increase in impeller tip speed and an increase in impeller size increase
the amount of energy dissipated in the flotation cell and it would there-
fore be expected that the direction of change in performance would be
the same (not the opposite). These results are an indication that it is not
just total energy input that affects flotation response.
3.3.3. Effect of cell aspect ratio
Fig. 8 shows that for low and high impeller tip speeds, the collection
zone flotation rate of the low aspect ratio cell was greater than the high
aspect ratio cell, presumably due to the greater proportion of turbulence
within the cell.
The higher turbulence closer to the froth pulp interface adversely af-
fects froth recovery, which is about 40 to 50% lower in the low cell as-
pect ratio than in the high cell aspect ratio. This is presumably due to
instability of the froth base.
3.3.4. Effect of air flowrate
For a particular cell aspect ratio and impeller tip speed, the collection
zone flotation rate (kc) and froth recoveries were similar for both low
and high air flowrates (Fig. 9).
Fig. 6. Effect of impeller size on the flotation performance.
There is an interactive effect between air rate and aspect ratio that is
significant. At the low cell aspect ratio, high air flowrate resulted in a
slightly higher overall flotation rate and froth recovery.
3.4. Correlating cell hydrodynamics with collection zone flotation
performance
Many authors have reported that flotation performance is propor-
tional to the bubble surface area flux (Sb), notably in the k-Sb relation-
ship proposed by Gorain et al. (1998). In the present study, however,
it was demonstrated that the different cell hydrodynamic conditions
do affect flotation performance in the collection zone.
It seems therefore that for the k-Sb relationship to be valid in this
particular study, some term related to the hydrodynamic conditions in
the cell should be included as kc = P ∙ f(hydrodynamics) ∙ Sb, where P is
the floatability of the particles which is a function of the ore and chem-
ical conditions in the cell (Gorain, 1997 and Gorain et al., 1998). A sim-
ilar approach has been proposed by Amini et al. (2015).
By comparing the collection zone flotation rate (kc) with cell overall
energy dissipation (Fig. 10a), it can be seen that kc was not related to
overall energy input alone as is commonly observed at laboratory
scale, where energy is usually only changed by varying impeller speed.
Correlations, however, were observed between collection zone flotation
rate and overall energy dissipation, for each cell aspect ratio and impel-
ler mechanism size combination (Fig. 10b).
This may suggest that flotation rate in the collection zone (kc) is pro-
portional not only to the amount of energy being dissipated by the im-
peller but also to the way energy is dissipated. For example, an increase
in energy dissipation by increasing the ratio between high and low tur-
bulent zones (i.e., lower cell aspect ratio or oversized impeller mecha-
nism) significantly increases the collection zone flotation rate. Whilst
 E. Tabosa et al. / International Journal of Mineral Processing 156 (2016) 99–107
Fig. 7. Effect of impeller tip speed on the flotation performance.
an increase in energy dissipation by increasing impeller tip speed tends
to reduce the collection zone flotation rate.
Therefore, better flotation performance in the collection zone might
be achieved when, for a given total energy dissipation rate, more of the
energy is dissipated in the impeller-stator region (e.g. by using a low cell
aspect ratio or increasing impeller mechanism size). Improvements to
the froth phase performance are also necessary to avoid losses in this
zone.
To make a quantitative assessment of how an improved collection
zone flotation rate would translate into an overall recovery improve-
ment without the froth recovery effect, Eq. (9) was used to calculate
the collection zone recovery (Rc). The residence time used in Eq. (9)
was 2.5 and 1.9 min for the high and low cell aspect ratio, respectively.
The curves that resulted from this analysis are shown (dotted lines) in
Fig. 11. Overall recovery (Ro) in a flotation cell is a function of recovery
in the collection (Rc) and froth (Rf) phase, (Dobby, 1984):
R f Rc
Ro ¼
R f Rc þ 1−Rc
kc τ
Rc ¼
1 þ kc τ
The collection zone copper recovery (Rc) may therefore be obtained
when there is no froth recovery effect, i.e. when froth recovery equals
100%. By comparing the collection zone copper recoveries for a particu-
lar cell aspect ratio, it can be seen that the low impeller tip speed in-
creased the collection zone recovery by 3% for the low cell aspect ratio
and 6.4% for the high cell aspect ratio.
Best overall performance would have been achieved with a low cell
aspect ratio running at a low impeller speed. Recovery under these
Fig. 8. Effect of cell aspect ratio on the flotation performance.
105
conditions is 10% higher than the standard conditions under which
the flotation cell is usually operated (i.e. high aspect ratio and high im-
peller speed).
4. Discussion
It has been demonstrated that cell hydrodynamics play an important
role in flotation performance. The results presented in this paper have
the following implications for flotation cell design, scale-up and
operation:
• An increase in the size of the high turbulent zone results in an increase
in the rate of flotation in the collection zone of a flotation cell. The pro-
portion of the cell that is highly turbulent has decreased significantly
as the size of flotation cells has dramatically increased in size over
the last 20 years. It is likely that this has resulted in a quiescent zone
that is too large in size for what is required to protect the froth zone
from the turbulence of the impeller. The height of the cell could be re-
ð8Þ duced without an adverse effect on the overall recovery achieved in
the flotation cell – thus reducing the amount of steel used and there-
fore the cost of manufacture of the cell. Alternatively, in the larger flo-
tation machines multiple impellers (possibly with different designs)
ð9Þ
at different heights would potentially result in a greater proportion
of the cell being highly turbulent, resulting in higher flotation recover-
ies.
• To be able to maximise the turbulence used in a flotation cell, design
features which minimise or protect the flotation froth from the turbu-
lence effects should be implemented. For example internal baffle rings
around the cell circumference located above the impeller region can
be used to divert flow so it does not travel upwards and disrupt the
froth zone. Alternatively, some sort of turbulence diffuser could be
placed below the froth to minimise turbulence effects.
106 E. Tabosa et al. / International Journal of Mineral Processing 156 (2016) 99–107
Fig. 9. Effect of air flowrate on the flotation performance (RCS3 impeller).
• Oversized mechanisms can potentially result in increased collection
zone rates and are therefore an option worth evaluating.
• In terms of cell operation, it is likely that the optimal impeller speed
will be different, depending on the particle size and hydrophobicity
of the ore being processed. Mechanical flotation cells should be pro-
vided with a variable speed drive motor in order to optimise the im-
peller speed.
• Flotation recovery rates would be expected to be higher in a smaller
volume cell than a larger volume cell (same aspect ratio) because a
larger proportion of the cell is highly turbulent.
• Cell hydrodynamic conditions could be optimised in order to improve
coarse (less liberated) particle recoveries. Evidence suggests that
coarse particle recovery is extremely sensitive to froth phase effects
with recovery optimal at shallow froth depth and when turbulence
at the pulp-froth interface is minimised (Tabosa et al., 2013). Further
study on the effect of cell hydrodynamics on flotation performance of
coarse and less liberated particles is recommended.
Understanding how cell hydrodynamic conditions affect perfor-
mance potentially allows them to be better controlled, thereby provid-
ing an additional means by which flotation can be enhanced.
Furthermore, because turbulent conditions inside a flotation cell have
usually only been changed by varying impeller tip speed and/or air
flow rate, there remains a need to better understand the relationship
between changes in impeller tip speed combined with changes in im-
peller/stator mechanism size and cell design. The results in this paper
suggest that it is the turbulence distribution within the flotation cell,
rather than just the energy input, that needs to be optimised.
New techniques have been developed to measure the distribution of
turbulence in an abrasive opaque slurry inside a flotation cell (Tabosa,
2012; Tabosa et al., 2012; Meng et al., 2015). These will allow the
Fig. 10. Effect of energy dissipation on collection zone flotation rate for (a) ungrouped data and (b) grouped by aspect ratio and impeller size.
turbulence distribution to be measured as a function of cell design and
operating conditions, and for the results to be correlated with flotation
performance. Further work will be required to elucidate the mecha-
nisms by which impeller speed and size, and cell aspect ratio affect col-
lection and froth phase recovery rates.
5. Conclusions
The aim of this study was to investigate the effects of cell hydrody-
namics on flotation kinetics in a pilot-scale flotation cell. The metallur-
gical experimental data collected proved to be very consistent and,
besides the variability with regards to the copper feed grade observed
throughout the campaign, tests showed that the test rig was capable
of producing very reproducible results.
Although the overall copper recoveries were very similar for the
conditions tested, by decoupling the pulp effect from the froth effect it
was possible to evaluate whether the changes tested affected the collec-
tion zone response, the froth zone response or both. The analysis
showed that the overall recovery had the potential to be up to 10%
higher if not limited by froth recovery.
The cell aspect ratio was the variable that had the most notable effect
on copper collection zone recovery. A change from the standard high
cell aspect to the low cell aspect ratio, however, had little effect on the
overall cell performance. A detailed analysis showed that the high cell
aspect ratio had a lower collection zone recovery, but was operating at
higher froth recoveries. Whereas the low cell aspect ratio presented
higher collection zone recovery, but its overall recovery was limited
by the low froth recovery.
The effect of other operational variable tested can be summarised as
follows:
 E. Tabosa et al. / International Journal of Mineral Processing 156 (2016) 99–107
Fig. 11. Overall copper recovery versus froth recovery relationship.
• Air flow rate (for the range evaluated here) had little effect on the flo-
tation performance;
• Low impeller tip speed had better flotation performance than the high
impeller tip speed (within the range used in this study);
• The oversized impeller mechanisms (RCS5) resulted in higher collec-
tion zone flotation rate compared to the standard RCS3 mechanism,
when operating at the same impeller tip speed and cell aspect ratio.
The RCS5, however, resulted in poorer flotation performance when
compared to the RCS3 because the overall flotation rate and froth re-
covery decreased.
Comparing the metallurgical performance of the cell with the differ-
ent hydrodynamic conditions it was found that the collection zone flo-
tation rate was not directly related to overall energy dissipation, as is
commonly observed at laboratory scale, where energy is usually only
changed by varying impeller tip speed. Results suggest that it is the
size of the turbulent zone, rather than just energy input, that affects flo-
tation recovery.
Acknowledgements
Professor Dr. Heinrich Schubert and his team are gratefully acknowl-
edged for laying the foundation for the understanding of the role of cell
hydrodynamics on flotation performance. The authors would also like
to acknowledge Metso Process Technology and Innovation for funding
this research and the assistance of personnel involved in the testing
campaign and during sample preparation and data analysis: Jaclyn
McMaster, Robert Crosbie, Kevin Cummins and Rae de Rusett. Rio
Tinto's Northparkes operation is also gratefully acknowledged for
assisting during the research work and assay analysis.
References
Ahmed, N., Jameson, G.J., 1985. The effect of bubble size on the rate of flotation of fine par-
ticles. Int. J. Miner. Process. 14 (3), 195–215.
Amini, E., Xie, W., Bradshaw, D., 2015. Enhancement of the precision and accuracy of the
flotation scale up process by introducing turbulence parameters to the AMIRA P9
model. Proceedings of the 7th International Flotation Conference (Flotation '15),
Cape Town, South Africa.
Cheng, Y.H., Mikhail, M.W., Salama, A.I.A., 1995. Effect of turbulence on the flotation of
fine coals. In: Laskowski, J.S.P., G.W. (Eds.), 1st UBC-McGill Bi-annual International
Symposium on Fundamentals of Mineral Processing, Vancouver, pp. 425–434.
Deglon, D.A., Sawyerr, F., O'Connor, C.T., 1999. A model to relate the flotation rate constant
and the bubble surface area flux in mechanical flotation cells. Miner. Eng. 12 (6),
599–608.
Deglon, D.A., Egya-mensah, D., Franzidis, J.P., 2000. Review of hydrodynamics and gas dis-
persion in flotation cells on South African platinum concentrators. Miner. Eng. 13 (3),
235–244.
Dobby, G.S., 1984. A fundamental flotation model and flotation column scale-up. Depart-
ment of Mining and Metallurgical Engineering. McGill University, Montreal, p. 259
PhD thesis.
Duan, J., Fornasiero, D., Ralston, J., 2003. Calculation of the flotation rate constant of chal-
copyrite particles in an ore. Int. J. Miner. Process. 72 (1–4), 227–237.
View publication stats
107
Feteris, S.M., Frew, J.A., Jowett, A., 1987. Modelling the effect of froth depth in flotation.
Int. J. Miner. Process. 20, 121–135.
Finch, J.A., Dobby, G.S., 1990. Column Flotation. Firth ed. Pergamon Press, Oxford.
Gorain, B.K., 1997. The Effect of Bubble Size on the Kinetics of Flotation and Its Relevance
to Scale-up PhD Thesis University of Queensland, Julius Kruttschnitt Mineral Research
Centre, Australia.
Gorain, B.K., Napier-Munn, T.J., Franzidis, J.P., Manlapig, E.V., 1998. Studies on impeller
type, impeller speed and air flow rate in an industrial scale flotation cell. Part 5: val-
idation of k-Sb relationship and effect of froth depth. Miner. Eng. 11 (7), 615–626.
Hall, G.A., Dell, C.C., 1989. An experimental study of the role of turbulence in flotation cell
design. Mineral Processing in the United Kingdom. Proceedings of a Meeting Orga-
nized by the Institution of Mining and Metallurgy. 4–5 April, 1989, pp. 149–164.
Jameson, G.J., Ahmed, N., 1983. Improving the rate of flotation of fine particles. Chemeca
83: The Eleventh Australian Chemical Engineering Conference. Brisbane, Australia,
pp. 203–208.
Jordan, C.E., Spears, D.R., 1990. Evaluation of a turbulent flow model for fine-bubble and
fine particle flotation. Miner. Metall. Process. 65–73 May.
Koh, P.T.L., Manickam, M., Schwarz, M.P., 2000. CDF simulation of bubble-particle colli-
sions in mineral flotation cells. Miner. Eng. 13 (14–15), 1455–1463.
Meng, J., Xie, W., Tabosa, E., Runge, K., Bradshaw, D., 2015. Turbulence modelling for flo-
tation cells based on piezoelectric sensor measurement data. Proceedings of the 7th
International Flotation Conference (Flotation '15), Cape Town, South Africa.
Morris, R.M., Matthesius, G.A., 1988. Froth flotation of coal fines - the influence of turbu-
lence on cell performance. J. South. Afr. Inst. Min. Metall. 88 (12), 385–391.
Newell, R., Grano, S., 2006. Hydrodynamics and scale up in Rushton turbine flotation cells:
part 2. Flotation scale-up for laboratory and pilot cells. Int. J. Miner. Process. 81 (2),
65–78.
Newell, R., Grano, S., 2007. Hydrodynamics and scale up in Rushton turbine flotation cells:
part 1 - cell hydrodynamics. Int. J. Miner. Process. 81 (4), 224–236.
Ogilvie, J.F., 1984. A Monte-Carlo approach to error propagation. Comput. Chem. 8 (3),
205–207.
Pyke, B.L., Duan, J., Fornasiero, D., Ralston, J., 2002. From turbulence and collision to at-
tachment and detachment: a general flotation model. In: Ralston, J., Miller, J.D.,
Rubio, J. (Eds.), Flotation + Flocculation: From Fundamentals to Applications,
pp. 77–89 Hawaii.
Pyke, B., Fornasiero, D., Ralston, J., 2003. Bubble particle heterocoagulation under turbu-
lent conditions. J. Colloid Interface Sci. 265, 141–151.
Runge, K.C., Seaman, B., Seaman, D., 2007. Optimising down the bank flotation froth per-
formance in Metso's 3 m3 RCS™ test cell. Flotation '07. Minerals Engineering Confer-
ence, Cape Town, South Africa.
Rutherford, K., Mahmoudi, S.M.S., Lee, K.C., Yianneskis, M., 1996. The influence of rushton
impeller blade and disk thickness on the mixing characteristics of stirred vessels.
Chem. Eng. Res. Des. 74 (3), 369–378.
Schubert, H., 1985. On some aspects of the hydrodynamics of flotation processes. In:
Forssberg, E. (Ed.), Flotation of Sulphide Minerals, pp. 337–352.
Schubert, H., 1986. On the hydrodynamics and the scale-up of flotation processes. In:
Somasundaran, P. (Ed.), Advances in Mineral Processing. Soc. Min. Eng., Littleton
(636–349).
Schubert, H., 1999. On the turbulence-controlled microprocesses in flotation machines.
Int. J. Miner. Process. 56, 257–276.
Schubert, H., 2008. On the optimization of hydrodynamics in fine particle flotation. Miner.
Eng. 21 (12–14), 930–936.
Schubert, H., Bischofberger, C., 1978. On the hydrodynamics of flotation machines. Int.
J. Miner. Process. 5 (2), 131–142.
Schubert, H., Bischofberger, C., 1979. On the optimization of hydrodynamics in flotation
processes. XIII International Mineral Processing Congress, Part B, pp. 353–375.
Schubert, H., Bischofberger, C., 1998. On the microprocesses air dispersion and particle-
bubble attachment in flotation machines as well as consequences for the scale-up
of macroprocesses. Int. J. Miner. Process. 52 (4), 245–259.
Schubert, H., Bischofberger, C., Koch, P., 1982. On the influence of the hydrodynamics in
flotation processes. XIV International Mineral Processing Congress. Toronto, Canada,
pp. IV-15.1–IV-15.17.
Seaman, D.R., Manlapig, E.V., Franzidis, J.P., 2006. Selective transport of attached particles
across the pulp–froth interface. Miner. Eng. 19 (6–8), 841–851.
Spears, D.R., Jordan, C.E., 1989. The effect of turbulence on the flotation of galena when
using fine bubbles. Advances in Coal and Mineral Processing Using Flotation, pp. 77–84.
Tabosa, E., 2012. The Effect of Cell Hydrodynamics on Flotation Kinetics PhD Thesis Julius
Kruttschnitt Mineral Research Centre, The University of Queensland, Brisbane,
Australia.
Tabosa, E., Runge, K., Holtham, P., 2012. Development and application of a technique for
evaluating turbulence in a flotation cell. Proceedings of the XXVI International Mineral
Processing Congress, IMPC 2012, New Delhi, India, 24–28 September, pp. 5377–5390.
Tabosa, E., Runge, K., Duffy, K., 2013. Strategies for increasing coarse particle flotation in
conventional flotation cells. Flotation '13. Minerals Engineering Conference, Cape
Town, South Africa.
Vera, M.A., Franzidis, J.P., Manlapig, E.V., 1999. Simultaneous determination of collection
zone rate constant and froth zone recovery in a mechanical flotation environment.
Miner. Eng. 12, 1163–1176.
Weber, A., Walker, C., Redden, L., Lelinski, D.S., Ware, S., 1999. Scale-up and design of large
scale flotation equipment. In: Parekh, B.K., Miller, J.D. (Eds.), Advances in Flotation
Technology. SME, Littleton, CO, pp. 353–370.
Weiss, T., Schubert, H., 1988. The effects of fine particles on the hydrodynamics of flota-
tion processes. In: Forssberg, E. (Ed.), XVI International Mineral Processing Congress,
pp. 807–818.
Wu, H., Patterson, G.K., 1989. Laser-Doppler measurements of turbulent-flow parameters
in a stirred mixer. Chem. Eng. Sci. 44 (10), 2207–2221.