CFD Modelling of Bubble-Particle Attachments in Flotation Cells

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Minerals Engineering 19 (2006) 619–626

This article is also available online at:


www.elsevier.com/locate/mineng

CFD modelling of bubble–particle attachments in flotation cells


P.T.L. Koh *, M.P. Schwarz
CSIRO Division of Minerals, Bayview Avenue, Box 312, Clayton South, Victoria 3169, Australia

Received 12 July 2005; accepted 6 September 2005


Available online 19 October 2005

Abstract

In recent years, computational fluid dynamic (CFD) modelling of mechanically stirred flotation cells has been used to study the com-
plexity of the flow within the cells. In CFD modelling, the flotation cell is discretized into individual finite volumes where local values of
flow properties are calculated. The flotation effect is studied as three sub-processes including collision, attachment and detachment. In the
present work, these sub-processes are modelled in a laboratory flotation cell. The flotation kinetics involving a population balance for
particles in a semi-batch process has been developed.
From turbulent collision models, the local rates of bubble–particle encounters have been estimated from the local turbulent velocities.
The probabilities of collision, adhesion and stabilization have been calculated at each location in the flotation cell. The net rate of attach-
ment, after accounting for detachments, has been used in the kinetic model involving transient CFD simulations with removal of bubble–
particle aggregates to the froth layer.
Comparison of the predicted fraction of particles remaining in the cell and the fraction of free particles to the total number of particles
remaining in the cell indicates that the particle recovery rate to the pulp–froth interface is much slower than the net attachment rates. For
the case studied, the results indicate that the bubbles are loaded with particles quite quickly, and that the bubble surface area flux is the
limiting factor in the recovery rate at the froth interface. This explains why the relationship between flotation rate and bubble surface
area flux is generally used as a criterion for designing flotation cells. The predicted flotation rate constants also indicate that fine and large
particles do not float as well as intermediate sized particles of 120–240 lm range. This is consistent with the flotation recovery generally
observed in flotation practice. The magnitude of the flotation rate constants obtained by CFD modelling indicates that transport rates of
the bubble–particle aggregates to the froth layer contribute quite significantly to the overall flotation rate and this is likely to be the case
especially in plant-scale equipment.
 2005 Elsevier Ltd. All rights reserved.

Keywords: Flotation bubbles; Flotation kinetics; Flotation machines; Modelling

1. Introduction local values of flow properties are calculated. The detailed


understanding of flow gained using this approach allows
Researchers have recently started to use computational modification to existing equipment and operation to im-
fluid dynamics (CFD) for modelling mechanically stirred prove flotation performance.
flotation cells to study the complexity of three-phase (air– The flotation effect is modelled as three sub-processes
water–solids) flows within the cells (Koh and Schwarz, involving collision, attachment and detachment. A turbu-
2003). Flotation cells are conventionally designed using lent collision model is used to estimate the rate of bub-
empirically derived relations. In CFD modelling, the flota- ble–particle encounters, employing the local turbulent
tion cell is discretized into individual finite volumes where velocity, and the size and number concentrations of bub-
bles and particles in different parts of the cell. The proba-
bilities of collision, adhesion and stabilization are also
*
Corresponding author. calculated such that attachment rates can now be esti-
E-mail address: Peter.Koh@csiro.au (P.T.L. Koh). mated. The detachment rates are also estimated from the

0892-6875/$ - see front matter  2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mineng.2005.09.013
620 P.T.L. Koh, M.P. Schwarz / Minerals Engineering 19 (2006) 619–626

fluid turbulence. The net rate of attachment, after account- that are available for attachment (Nb) is also related to
ing for detachments, is used in the combined CFD kinetic the total number of bubbles as follows:
model involving transient population-balance simulations N b ¼ ð1  bÞN bT ð4Þ
with removal of bubble–particle aggregates to the froth
layer. So, the attachment–detachment kinetics in Eq. (1) can be
In this paper, the local turbulent energy dissipation rates re-written as follows:
used in the kinetic model are obtained by CFD modelling dN p1
of the cell. This is a significant improvement over other ¼ k 1 N p1 N bT ð1  bÞ þ k 2 N bT b ð5Þ
dt
models where assumptions based on an average dissipation
An interpretation of the new equation is that the kinetics is
rate or an assumed Gaussian distribution for turbulent
now based on the free surface area available for particle
velocities for the whole cell (e.g. Bloom and Heindel,
attachment with a bubble loading parameter b such that
2003) have been used. CFD modelling provides a realistic
b = 0 for clean bubbles and b = 1 for fully loaded bubbles.
approach to flotation models without the additional
The number of particles that can be attached to a single
assumptions on turbulent energy dissipation rates.
bubble is given by S which is the ratio of the total surface
Flotation in a cylindrical tank fitted with a Rushton
area of the bubble to the particle projected area as follows:
turbine impeller is studied in this paper since commercial
 2
flotation equipment are mostly of this type fitted with ro- db
tor–stator system. For comparison, a laboratory flotation S¼4 ð6Þ
dp
cell designed by CSIRO Minerals is also modelled.
where dp is the particle diameter and db is the bubble diam-
eter. Realistically, it is not possible for the particles to cov-
2. Flotation kinetics er the whole bubble surface due to packing, shape and
other factors. As a first approximation, it is assumed that
Flotation is generally modelled as a first-order rate pro- the attached particles occupy about half of the total bubble
cess with respect to the number of particles and number of surfaces when the bubbles are fully loaded. The maximum
bubbles in the attachment step, and a detachment process number of particles per bubble ratio (Smax) is then de-
with respect to the number of bubble–particle aggregates. scribed by
The kinetic equation for the bubble–particle encounters is  2
described by the rate of removal of the number of particles db N p2
S max ¼ 0:5S ¼ 2 ¼ ð7Þ
in a given volume as follows: dp bN bT
dN p1 By rearranging the above equation, the bubble loading
¼ k 1 N p1 N b þ k 2 N a ð1Þ
dt parameter b can be obtained from
where Np1 is the number concentration (m3) of free parti- N p2
cles, Nb is the number concentration of bubbles available b¼ ð8Þ
S max N bT
for attachment, Na is the number concentration of bub-
ble–particle aggregates, k1 is the particle–bubble attach- The present method is a significant improvement over pre-
ment rate constant, and k2 is the particle–bubble vious models in modelling the number of particles that can
detachment rate constant. In a given volume within the flo- be attached to a bubble. The first model proposed by
tation cell, the total number of particles consists of free Bloom and Heindel (1997) had assumed that only bubbles
particles (Np1) and particles that are attached to bubbles which do not already have a particle attached to them are
(Np2), both as functions of time capable of picking up a particle, and that the average num-
ber of particles on a bubble was equal to 1.0. This assump-
N pT ¼ N p1 þ N p2 ð2Þ tion was replaced in their latest model (Bloom and Heindel,
with NpTo as the initial particle concentration. In general, 2003) by an average loading based on total numbers of par-
bubbles will have any number of particles attached as they ticles and bubbles in the flotation cell.
move up toward the froth layer, with the number of parti- In the kinetic equation, the particle–bubble attachment
cles per bubble varying with time and position, as well as rate constant k1 (m3/s) is defined by
from bubble to bubble. To simplify the situation, it is as- k1 ¼ Z 1P cP aP s ð9Þ
sumed in this paper that certain bubbles are fully loaded
with particles while the remaining bubbles are clean. The and the particle–bubble detachment rate constant k2 (1/s) is
number of bubble–particle aggregates (Na) is then related defined by
to the total number of bubbles (NbT) by an average loading k 2 ¼ Z 2 P d ¼ Z 2 ð1  P s Þ ð10Þ
parameter (b) as follows:
where Pc, Pa and Ps are the probabilities of particle–bubble
N a ¼ bN bT ð3Þ collision, adhesion and stabilisation against external forces.
The average loading parameter (b) in a given volume varies Pd is the probability a bubble–particle aggregate will be-
with time and position in the cell. The number of bubbles come unstable and is assumed to be equal to (1  Ps). Ps
P.T.L. Koh, M.P. Schwarz / Minerals Engineering 19 (2006) 619–626 621
pffiffiffiffiffiffi 1=3
is included in both the particle–bubble attachment rate C1e
Z2 ¼ ð15Þ
constant k1 and in the detachment rate constant k2 because ðd p þ d b Þ2=3
the processes involve different turbulent eddies acting inde-
pendently of each other. The eddy that facilitates attach- where C1 is an empirical constant with a value of 2 as sug-
ment can also cause detachment of the aggregate during gested by Bloom and Heindel (2003).
the attachment process. Z1 is related to the particle–bubble The collision efficiency Pc accounts for the tendency of
collision frequency dependent on the size of the particles particles to follow the fluid streamlines around the bubble
and bubbles, and hydrodynamics of the flotation pulp. Z2 and avoid actual contact, which is especially true for parti-
is the detachment frequency of particles from bubbles. cles that are much smaller than the bubble. The equation
Based on work of Abrahamson (1975), Schubert and for the collision probability proposed by Yoon and Luttrell
Bischofberger (1979) applied the following equation for (1989) for intermediate bubble Reynolds number in the
the number of particle–bubble collisions per unit time range of 0.2 < Reb < 100 is applied:
and volume in flotation cells. The equation is valid only   2
in turbulent flows where inertial effects are the primary 4 0:72 d p
P c ¼ 1:5 þ Reb ð16Þ
cause of collisions 15 d 2b
 2
dp þ db  2 1=2 where the bubble Reynolds number based on Reb = dbUb/m
Z 1 ¼ 5:0 U p þ U 2b ð11Þ
2 is used. The equation is valid for particles smaller than
100 lm and bubbles smaller than 1 mm with immobile sur-
where Up is the turbulent (rms) fluctuating velocity of the faces due to adsorbed surfactants.
particle relative to the fluid, and Ub is the turbulent (rms) The probability of adhesion Pa is determined by the slid-
fluctuating velocity of the bubble relative to the fluid. In ing time of the particles on bubble surfaces and the induc-
typical flotation processes, these velocities (Ui = Up or tion time for rupture of the disjoining film between the
Ub) are a function of the local turbulent energy dissipation particle and bubble. If the sliding time is longer than the
rate as follows (Leipe and Möckel, 1976): induction time, adhesion is likely. Yoon and Luttrell
7=9  2=3 (1989) derived an expression for the adhesion probability
0:4e4=9 d i qi  qf dependent on the particle and bubble sizes, the bubble Rey-
Ui ¼ ð12Þ
m1=3 qf nolds number and the induction time as follows:
where e is the turbulent energy dissipation rate per unit   
2 ð45 þ 8Re0:72b ÞU b t ind
mass, m is the kinematic viscosity, qf is the fluid density, P a ¼ sin 2 arctan exp ð17Þ
15d b ðd b =d p þ 1Þ
and qi is the density of the particle (p) or bubble (b). The
condition for use of the above model with independent where tind is the induction time and Ub is the bubble rela-
bubble or particle velocities is that the diameter of the par- tive velocity. The induction time is a function of the parti-
ticle or bubble must be greater than the critical diameter, cle size and contact angle which can be determined by
dcrit in the following equation: experiment and correlated in the form
15lf U 2f tind ¼ Ad Bp ð18Þ
d 2i > d 2crit ¼ ð13Þ
qi e
where parameters A and B are independent of particle size.
where lf is the fluid viscosity and Uf is the mean fluid veloc- From measurements with particles and bubbles of various
ity deviation. In applying the case to bubbles, the critical sizes in solutions of varying ionic strength, Dai et al. (1999)
diameter is based on the virtual mass which is estimated found that parameter B is constant with a value of 0.6, and
by qi = 0.5qf using the fluid density. The criterion is used parameter A is inversely proportional to the particle con-
to determine the applicability in terms of particle and bub- tact angle h. Based on these findings, the following equa-
ble diameters in various turbulent regimes within the flota- tion was constructed by the present authors and applied
tion cell. in the CFD model:
In regions where velocities for the particle and bubbles
75 0:6
are not independent, the collision equation by Saffman tind ¼ d ð19Þ
h p
and Turner (1956) is applicable for fine particles and bub-
bles confined within eddies in low turbulent dissipation re- where tind is given in s, h in degrees and dp in m. The prob-
gions as follows: ability of adhesion can now be calculated for given values
rffiffiffiffiffiffi 3 of bubble size, particle size and contact angle.
8p d p þ d b e 1=2 The probability of stability Ps accounts for the stabiliza-
Z1 ¼ ð14Þ
15 2 m tion or destabilization of the bubble–particle aggregate.
Schulze (1993) proposed a form as follows:
The bubble–particle detachment frequency Z2 is depen-  
1
dent on the relative velocity between the particle–bubble P s ¼ 1  exp 1   ð20Þ
aggregate and the surrounding fluid, and is estimated by Bo
622 P.T.L. Koh, M.P. Schwarz / Minerals Engineering 19 (2006) 619–626

where the modified Bond number (Bo*) is defined as the ra- conical shroud similar to a Denver-type mechanism. Air is
tio of detachment to attachment forces, and is given by injected at a rate of 8 l/min through a nozzle located within

  1=3   
2=3 d p

d 2p Dqp g þ 1:9qp e 2
þ db
2
þ 1:5d p
4r
db
 d b qf g sin2 p  h2
Bo ¼

6r sin p  h sin p þ h

ð21Þ
2 2

where g is the acceleration due to gravity, r is the surface the stator and directed downwards at the impeller which is
tension, and Dqp = (qp  qf). In the derivation by Schulze, stirring at 1200 rpm.
the acceleration (force) that determines the detachment of a
particle from the bubble is dependent on the intensity of 4. Results and discussion
turbulence in the flow. It is assumed that turbulent vortices
of dimensions corresponding to those of the bubble–parti- CFD predictions indicate a complex flow field within the
cle aggregate cause the detachment. From their validation flotation cell and much of the action occurs in the impeller
experiments, Bloom and Heindel (2003) suggested a modi- region. The flow around the cell is important as it deter-
fied form as follows: mines the paths taken by the bubble–particle aggregates to-
   wards the froth layer. Profiles of the flow field obtained
1
P s ¼ 1  exp As 1   ð22Þ include turbulent energy dissipation, volumetric fraction
Bo
of air phase and particle–bubble collision rate.
where As is an empirical constant with a value of 0.5. The cumulative distribution of the turbulent energy dis-
sipation rates in the stirred cell have been reported (Koh
3. Model description et al., 2000) from which two regions in the flotation cell
are represented: the impeller region having higher values
Multi-phase flow equations for the conservation of of the turbulent energy dissipation rate and the remainder
mass, momentum and turbulence quantities have been of the cell having lower dissipation values. In the popula-
solved using an Eulerian–Eulerian approach in which the tion balance, the bubble number concentration is obtained
pulp (liquid–solid) and the gas phases are treated as inter- by dividing the local gas holdup by the bubble volume.
penetrating continua. Two mesh blocks have been gener- The probabilities of collision, adhesion and stabilization
ated for modelling the rotating impeller and the have been calculated using values of turbulent energy dissi-
stationary stator within the tank. Transport equations have pation rates found in the flotation cell. Two contact angles
been solved using the multiple frames of reference tech- have been used in the calculation for comparison. The
nique in the CFX-4.4 (2001) computer code. Source terms attachment rate, (Z1PcPaPsNbT) is plotted as a function
include buoyancy for the gas phase. The froth layer is not of particle diameter in Fig. 1. This rate represents the initial
included in the computational domain; only the pulp zone rate of contact between particles and clean bubbles.
is simulated. At the froth–pulp interface, gas bubbles with
attached particles are transferred from the pulp zone to the
froth layer at the rate bubbles arrive at the interface. For
the flotation kinetics, the transfer of particles between the
pulp and bubbles is achieved by applying source/sink terms
in the population-balance equation in Eq. (1). The tran-
sient simulation uses adjustable/variable time steps such
that the mass error is less than 0.1% for each time step.
A stirred tank cell used in flotation experiments has been
simulated. A flat-bottomed cylindrical tank of diameter
T = 0.195 m was fitted with a standard Rushton turbine
with a diameter D = T/3 stirring at 840 rpm (Hui and
Ahmed, 1999). The tank includes four baffles of width T/
10. An average bubble diameter of 1 mm, and monosized
particles of density 2600 kg/m3 in a pulp of 32% by weight
have been applied in the model. This is equivalent to a par-
ticle volume fraction of 0.153.
For comparison, a laboratory flotation cell designed by
CSIRO Minerals has also been modelled. The CSIRO flo- Fig. 1. Attachment rate (Z1PcPaPsNbT) for a bubble of diameter 1.0 mm
tation cell, with a volume of 3.78 l, has an eight-bladed 72- plotted as a function of particle diameter at four turbulent energy
mm diameter impeller driven from below and enclosed in a dissipation rates and two contact angles.
P.T.L. Koh, M.P. Schwarz / Minerals Engineering 19 (2006) 619–626 623

To obtain flotation rates, the mass transfer of particles


between the pulp and the bubbles in the flotation cell has
been simulated for a contact angle 38. The total number
of particles remaining in the batch cell is determined by
summing particles in all the finite volumes. The results
for the case of 60 lm particles are shown in Fig. 2 where
the predicted fractions of free, attached and floated parti-
cles are plotted as a function of time. These results may
be interpreted as produced by a ‘‘reaction in series’’ mech-
anism. The sum of free and attached particles remaining in
the cell is plotted against time in Fig. 3 (semi-log) for var-
ious particle sizes. This plot is similar to results usually ob-
tained from flotation tests. A more interesting plot is the
ratio of free particles to the total number of particles
remaining in the cell (free and attached) shown in Fig. 4.
A global ‘‘equilibrium’’ between free and attached particles
is observed in some of the cases in Fig. 4 due to attachment Fig. 4. CFD prediction of the ratio of free particles to the sum of free and
attached particles remaining in the cell plotted as a function of time for
various particle diameters.

and detachment processes occurring in different parts of


the cell. Assuming a first-order rate process, flotation rate
constants can be obtained from the half times t0.5 using
the following equation:
ln 2:0
k¼ ð23Þ
t0:5
The predicted rate constants are plotted as a function of
particle diameters in Fig. 5 showing that fine and large par-
ticles do not float as well as intermediate sized particles of
120–240 lm. This trend is generally observed with flotation
recovery (Trahar, 1981). Experimental data reported by
Pyke et al. (2003), Duan et al. (2003) and Ahmed and
Jameson (1985) are also consistent with this observation.
An initial comparison against experimental rate constant
Fig. 2. CFD prediction of the fractions of free, attached and floated
for quartz at 600 rpm by Ahmed and Jameson in Fig. 5
particles plotted as a function of time for particles of 60 lm diameter.

Fig. 3. CFD prediction of the sum of free and attached particles Fig. 5. CFD predicted flotation rate constant plotted as a function of
remaining in the cell plotted as a function of time for various particle particle diameter and comparison against experimental rate constant for
diameters. quartz by Ahmed and Jameson.
624 P.T.L. Koh, M.P. Schwarz / Minerals Engineering 19 (2006) 619–626

shows that the CFD predicted values are of the same order times is because the removal of attached particles from
of magnitude. Detailed comparison against experiment is the pulp is limiting the process. The distribution of bubble
planned in further work. loading in the cell is plotted in Fig. 6 showing that the bub-
In practice, the rate constants are usually correlated bles are almost fully loaded in the top part of the cell with b
with the bubble surface area flux Sb which can be obtained values closer to 1.0. The initial attachment rate is found to
from be very fast in comparison to the recovery rates of attached
6J g particles from the froth interface. Since the bubbles are
Sb ¼ ð24Þ loaded with particles quite quickly, the bubble surface area
db
flux is the limiting factor in the recovery rate from the froth
where Jg is the superficial gas velocity. Plant data indicate interface. This explains why the relationship in Eq. (25) be-
that Sb is dependent on the impeller speed and air flow rate. tween flotation rate and bubble surface area flux is gener-
Gorain et al. (1999) found that the flotation rate constant k ally observed in batch or plant-scale cells and why this
has a strong correlation with Sb as follows: relationship has been used as a criterion for designing flo-
k ¼ P k  S b  Rf ð25Þ tation cells.
The recovery rate is slower than the net attachment rate
where Pk is the flotation probability representing the float- because of transport times for bubble–particle aggregates
ability of the ore, and Rf is the froth recovery factor. The to move through the pulp to the froth interface. The pre-
flotation probability for various particle sizes have been dicted recovery rates are still quite large in comparison
calculated from CFD predicted k values, with Rf = 1.0 with the observed rates in the plant because the overall
and Jg = 0.0106 m/s (or Sb of 63.4 s1) in the cell. The cal- rates include transport within the froth layer. For example,
culated values of Pk of the order of 104 are shown in Ta- measured flotation rates of 0.1–0.25 min1 have been re-
ble 1. Pk is calculated using Eq. (25) based on the rate ported by Gorain et al. (1998) in 3 m3 commercial cells.
constant k predicted by CFD simulation. The relationship The transport times are quite significant in plant-scale flo-
between Pk and PcPaPs is not obvious since they are de- tation cells as the bubble–particle aggregates have to move
fined differently. The probabilities PcPaPs are involved dur- over large distances. This is one reason why flotation rates
ing a single particle–bubble encounter, while Pk is obtained are much faster in laboratory batch cells where the bubble–
from the rate of decrease of the total number of particles in particle aggregates have shorter distances to reach the froth
the flotation cell as a result of multiple encounters over layer.
time. The distribution of attachment rates in the stirred cell is
The attachment rates from previous simulations in a plotted in Fig. 7 where negative values represent detach-
laboratory flotation cell designed by CSIRO Minerals ment. The plot shows that detachment rates are large near
(Koh and Schwarz, 2003) are for bubble–particle attach- the impeller tip. The maximum attachment rates are not far
ments only, without the recovery of particles through the from the impeller. The two zones are quite close. The net
froth interface. In the previous work, the attached particles
were removed from the pulp phase as soon as a bubble–
particle aggregate was formed. The attachment rate con-
stant of the order of 300 s1 for 60 lm particles was ob-
tained and is much larger in comparison to the flotation
rate of 0.02 s1 (or 1.13 min1) in Table 1 as obtained in
the present work which includes transport to the froth
interface.
In Fig. 2, the fraction of free particles to the total num-
ber of particles remaining in the cell decreases quite rapidly
initially with the rate of decrease abating at longer times.
The reason for slower decrease in free particles at longer

Table 1
CFD predicted flotation rate constant and flotation probability for
various particle diameters
dp (lm) t0.5 (s) k (s1) k (min1) Pk
7.5 205.9 0.0034 0.202 5.3 · 105
15 113.0 0.0061 0.368 9.7 · 105
30 61.3 0.0113 0.678 1.8 · 104
60 36.8 0.0188 1.130 3.0 · 104
120 19.9 0.0348 2.090 5.5 · 104
240 11.7 0.0592 3.555 9.3 · 104
Fig. 6. CFD prediction of the bubble loading parameter b after flotation
480 16.7 0.0416 2.496 6.6 · 104
time of 320 s for particles of 60 lm diameter in the stirred cell.
P.T.L. Koh, M.P. Schwarz / Minerals Engineering 19 (2006) 619–626 625

Fig. 7. CFD predicted net attachment rates (108 m3 s1) after flotation Fig. 9. CFD predicted volume fraction of attached particles after flotation
time of 320 s for particles of 60 lm diameter in the stirred cell. Negative time of 320 s for particles of 60 lm diameter in the stirred cell.
values indicate net detachment.

effects of the attachment rates are shown in the predicted


distribution of free particles remaining in the cell plotted
in Fig. 8, and in the distribution of attached particles plot-
ted in Fig. 9.
For comparison, the distribution of net attachment rates
in the CSIRO Denver cell is shown in Fig. 10. The plot
shows that detachment rates (with negative values) are also
large near the impeller tip, but the maximum attachment
rates are outside the impeller zone. The two zones are sep-
arated by the stator shroud. The negative regions are inev-

Fig. 10. CFD predicted net attachment rates (108 m3 s1) after flotation
time of 228 s for particles of 60 lm diameter in the CSIRO Denver cell.

itable because of the need to use a high impeller speed to


generate fine bubbles for the necessary bubble surface area
flux to operate. Flotation cell design should ideally mini-
mize the negative regions while maximizing the attachment
rates available in the cell.

5. Conclusions

CFD modelling of a batch flotation cell has been per-


formed. Bubble–particle collision rates, detachment rates
and the probabilities of collision, adhesion and stabiliza-
tion have been calculated in different parts of the cell. Com-
Fig. 8. CFD predicted volume fraction of free particles after flotation time parison of the predicted fraction of particles remaining in
of 320 s for particles of 60 lm diameter. the cell and the fraction of free particles to the total number
626 P.T.L. Koh, M.P. Schwarz / Minerals Engineering 19 (2006) 619–626

of particles remaining in the cell indicates that the particle Duan, J., Fornasiero, D., Ralston, J., 2003. Calculation of the flotation
recovery rate to the pulp–froth interface is much slower rate constant of chalcopyrite particles in an ore. International Journal
of Mineral Processing 72, 227–237.
than the net attachment rates after accounting for detach- Gorain, B.K., Napier-Munn, T.J., Franzidis, J.P., Manlapig, E.V., 1998.
ments. For the case studied, it was found that the bubbles Studies on impeller type, impeller speed and air flow rate in an
are loaded with particles quite quickly, and that the bubble industrial scale flotation cell—Part 5: validation of k-Sb relationship
surface area flux is the limiting factor in the recovery rate at and effect of froth depth. Minerals Engineering 11 (7), 615–626.
the froth interface. This explains why the relationship be- Gorain, B.K., Franzidis, J.P., Manlapig, E.V., 1999. The empirical
prediction of bubble surface area flux in mechanical flotation cells
tween flotation rate and bubble surface area flux is gener- from cell design and operating data. Minerals Engineering 12, 309–
ally used as a criterion for designing flotation cells. The 322.
predicted flotation rate constants also showed that fine Hui, S., Ahmed, N., 1999. Hydrodynamics of coarse particle flotation. In:
and large particles do not float as well as intermediate sized Proceedings of the 27th Australian Chemical Engineering Conference,
particles of 120–240 lm range. This is consistent with the CHEMECA 99, Newcastle, NSW.
Koh, P.T.L., Schwarz, M.P., 2003. CFD modeling of bubble–particle
flotation recovery generally observed in flotation practice. collision rates and efficiencies in mineral flotation cells. Minerals
The magnitudes of the flotation rate constants obtained Engineering 16, 1055–1059.
by CFD modelling indicate that transport rates of the bub- Koh, P.T.L., Manickam, M., Schwarz, M.P., 2000. CFD simulation of
ble–particle aggregates to the froth layer contribute quite bubble–particle collisions in a flotation cell. Minerals Engineering 13
significantly to the overall flotation rate and this is likely (14–15), 1455–1463.
Leipe, F., Möckel, O.H., 1976. Untersuchungen zum stoffvereinigen in
to be the case especially in plant-scale equipment. flüssiger phase. Chemical Technology 30, 205–209.
Pyke, B., Fornasiero, D., Ralston, J., 2003. Bubble particle heterocoag-
References ulation under turbulent conditions. Journal Colloid and Interface
Science 265, 141–151.
Abrahamson, J., 1975. Collision rates of small particles in a vigorously Saffman, P.G., Turner, T.S., 1956. On the collision of drops in turbulent
turbulent fluid. Chemical Engineering Science 30, 1371–1379. clouds. Journal of Fluid Mechanics 1, 16–30.
Ahmed, N., Jameson, G.J., 1985. The effect of bubble size on the rate of Schubert, H., Bischofberger, C., 1979. On the optimization of hydrody-
flotation of fine particles. International Journal of Mineral Processing namics in flotation processes. In: Proceedings of the 13th International
14, 195–215. Mineral Processing Congress, vol. 2, Warsaw, pp. 1261–1285.
Bloom, F., Heindel, T.J., 1997. Mathematical modeling of the flotation Schulze, H.J., 1993. Flotation as a heterocoagulation process: possibilities
deinking process. Mathematical and Computer Modelling 25, 13–58. of calculating the probability of flotation. In: Dobias, B. (Ed.),
Bloom, F., Heindel, T.J., 2003. Modeling flotation separation in a semi- Coagulation and Flocculation. Marcel Dekker, New York, pp. 321–
batch process. Chemical Engineering Science 58, 353–365. 353.
CFX USER GUIDE, Release 4.4, 2001. Computational Fluid Dynamics Trahar, W.J., 1981. A rational interpretation of the role of particle size in
Services, AEA Industrial Technology, Harwell Laboratory, Oxford- flotation. International Journal of Mineral Processing 8, 289–327.
shire, UK. Yoon, R.H., Luttrell, G.H., 1989. The effect of bubble size on fine particle
Dai, Z., Fornasiero, D., Ralston, J., 1999. Particle–bubble attachment in flotation. Mineral Processing and Extractive Metallurgy Review 5,
mineral flotation. Journal of Colloid and Interface Science 217, 70–76. 101–122.

You might also like