You are on page 1of 17

International Journal of Mineral Processing, 11 (1983) 203--219 203

Elsevier Science Publishers B.V., Amsterdam -- Printed in The Netherlands

THE EFFECT OF AIR FLOW RATE ON THE KINETICS OF FLOTATION.


PART 1: THE T R A N S F E R OF MATERIAL FROM THE S L U R R Y TO THE
FROTH

A.R. LAPLANTE 1, J.M. TOGURI 2 and H.W. SMITH 2


Department Mining and Metallurgical Engineering, McGill University, Montreal, Queb.
H3A 2A 7 (Canada)
2Department Metallurgy and Materials Science, University of Toronto, Toronto, Ont.
M5S 1A4 (Canada)
(Received January 10, 1981; revised and accepted November 30, 1982)

ABSTRACT

Laplante, A.R., Toguri, J.M. and Smith, H.W., 1983. The effect of air flow rate on the
kinetics of flotation. Part 1 : The transfer of material from the slurry to the froth. Int.
J. Miner. Process., 11 : 203--219.

Bubble size distributions and flotation rates were determined as a function of air flow
rate and frother concentration using a specially designed batch flotation cell. This cell
permitted the unambiguous determination of the flotation rate from the slurry to the
froth.
Flotation rate constants were determined for different size classes of silica and galena.
The flotation rate constants increased to a maximum and then decreased as air flow rate
was increased. This maximum was predicted by a model which considered the effect of
bubble size on both the total bubble surface area and the bubble-particle collision effi-
ciency. This work shows that collision efficiency effects, shown to exist in single-bubble/
single-particle systems, are also present in flotation systems where many bubbles and
particles interact.
A second model for hindered flotation is proposed which assumes that the particle-
capturing bubble surface differs from the particle-retaining surface. This model predicts
a sharp transition from hindered to free flotation. Experimental results are presented
which agree well with those derived from the model.

INTRODUCTION

Air flow rate (AFR) is a variable of major importance in flotation kinetics.


This fact is now well established, and AFR is used as a control or optimiza-
tion variable in some Finnish (Niemi and Paakkinen, 1969; King, 1973a) and
Australian (McKee et al., 1976; Manlapig and Spottiswood, 1978) flotation
plants. However, the actual mechanisms by which AFR controls flotation
kinetics are still incompletely understood. No flotation model to date offers
a comprehensive description of the effect of AFR on all the transfer reac-
tions present in flotation systems. The major transport processes are the fol-

0301-7516/83/$03.00 © 1983 Elsevier Science Publishers B.V.


204

lowing: transfer of hydrophobic particles from the slurry to the froth, trans-
fer of material from the froth over the cell lip, drop-back of material from
the froth to the slurry, and nonselective recovery of fine particles in the con-
centrate by hydraulic entrainment. Generally the overall kinetics are mea-
sured. Consequently, the effect of A F R on each sub-process is n o t known.
Despite this problem, a few flotation models incorporating A F R as a variable
have been developed. One of these is that of King (1972, 1973b), who pro-
posed the following description of the flotation rate constant (FRC):
g = K* A Z ~(D) (1)

where: K = flotation rate constant (FRC) (s-l); K* = mass transfer rate con-
stant (cm s-l); A = bubble surface area per volume of pulp (cm-1); S = frac-
tion of the bubble surface n o t covered by adhering particles (dimensionless);
~(D) = particle size correction factor (dimensionless); and:
A = ~ ~ G/V (2)
where ~ = bubble surface area per unit volume, or specific bubble surface
(cm-1); r = bubble average residence time in the slurry (s); G = air flow rate
(cm 3 s-l); V = cell volume (cm3).
According to this model, A F R enters because of its influence on A. How-
ever, A F R may also influence o. King also assumed that only the free bubble
surface is effective for capture, an assumption c o m m o n to Sutherland
(1948), Pogorely (1962) and Sastry and Fuerstenau {1970).
The objective of the present study is to determine the rate of transfer of
material from the slurry to the froth as a function of AFR, and thereby test
King's model. For this purpose it is necessary to have data on the bubble
size distribution as a function of A F R and surface active reagent concentra-
tion. Bubble size distribution is necessary not only for calculating bubble
specific surface but for determining the collision efficiency (Flint and
Howarth, 1971; Reay and Ratcliff, 1973, 1975; Anfruns and Kitchener,
1976).

EXPERIMENTAL

(a) F l o t a t i o n cell

The flotation cell required for the present study must serve the following
purposes: quick removal o f the concentrate; quiescent slurry surface; similar-
ity to existing industrial machines; and ease of bubble photography. Thus
the design criteria were: (a) high length to froth surface ratio, to favour
quick removal of the froth phase; (b) high cell height to froth surface ratio,
to obtain a quiescent slurry surface, facilitate froth thickness control and in-
crease froth stability; and (c) a rotor-stator assembly of the sub-aeration
type, similar to that of most flotation cells.
205

Figure 1 shows the plexiglass cell of 5.5-1itre capacity constructed to meet


the above criteria. The inside dimensions are width, 22.9 cm; height to the
cell lip, 25.4 cm and depth, 10.2 cm. Two rotor-stator assemblies were used
(Fig. 2) to satisfy the first t w o design criteria while retaining sufficient
bubble dispersion and adequate suspension. The impellers were rotated in
opposite directions to increase flow turbulence and hence agitation effective-
ness. Each four-bladed impeller of the rotor-stator assembly was surrounded
by a shroud of baffles held in place by a metallic disc (Fig. 2). Four holes
were drilled in each disc to suck the slurry into the rotor path. Low-pressure
air was introduced into the shafts covering the rotor axles and directed above
the rotors, which were operated at 1180 RPM. The A F R was measured with
a wet test flow meter in paraUel with a U-tube manometer.

Fig. 1. Plexiglass flotation cell designed for this work.

(b) Bubble size distributions

To determine bubble size distributions, two series of experiments were car-


ried out, in the absence o f a mineral phase. In one, A F R was varied from 0
to 8 litres per minute, at methylisobutylcarbinol (MIBC) concentrations of
0 and 0.98 X 10 -4 molar. In a second series of tests, MIBC concentration was
varied from 0 to 3 X 10 -4 molar, at an A F R of 4.0 litres per minute. To mea-
sure bubble size, photographs were taken of the front t o p half of the flota-
tion cell, using a black plate immersed in the cell as background. This plate
206

Fig. 2. The two-rotor-stator assembly.

limited the thickness of the cell volume under investigation and provided a
contrasting background for the bubbles. A flash was used at 45 degrees on
both sides of the cell. A milimetric scale was secured to the front of the cell
to provide the necessary scale. Bubble diameters were measured either visual-
ly or, for the smaller bubbles, under a low-magnification microscope. For
each set of experimental conditions, bubble diameters on 2 to 4 photographs
were measured.

(c) Flotation rates

The high cell-height-to-froth-surface ratio creates two distinct zones with-


in the liquid phase. The first one, confined to the lower part of the cell, is
the active zone, where the mineral particles are suspended. Above this is the
second zone, which is only weakly agitated to provide a quiescent slurry--
froth interface. By adding a continuous flow of washing water, a liquid film
flowing over the cell lip was created on which the froth was quickly re-
moved. Preliminary experiments indicated that varying the washing water
flowrate between 1.8 and 3.6 litres per minute had no effect on the concen-
trate removal rate. All subsequent tests were performed at 2.7 litres per
minute. The mineral laden froth was recovered in pans replaced at measured
time intervals. The last concentrate product was collected over a period of
several minutes to ensure the recovery of all hydrophobic material. The
slurry, being confined to the lower part of the cell, was n o t entrained over
the cell lip by the washing water stream. The floated products were filtered,
dried, weighed, and kept for size analysis.
207

For the first series of tests, a quartz sand from Ottawa, Illinois, was sieved
and the +150 - 2 1 0 and +74 - 1 0 5 ~ m size fractions were saved. A third size
fraction was obtained by wet grinding, desliming and wet sieving to - 7 4 ~m.
Its size distribution is shown in Table I.

TABLE I

Size analysis of the - 7 4 ~m quartz

Size (#m) Weight (%)

--74 +53 8.2


--53 +37 29.9
--37 +27 49.5
--27 +19 11.5
--19 +13.5 0.8

For each flotation test, a 300 g sample of the previously conditioned and
floated quartz was cleaned for 10 minutes in a 2-1itre WEMCO laboratory
flotation cell with 2 litres of a 0.60 molar HC1 solution. The sample was
filtered, washed, and transferred to the experimental flotation cell, then
conditioned for 8 minutes at a pH of 12 + 0.2 (controlled by the addition
of a NaOH saturated solution). The solution was stabilized by the addition
of 5.5 mL of a 0.20 molar solution of KNO3. After this preliminary con-
ditioning, 20 mL of a 1 wt.% solution of an industrial alkyl propylene col-
lector, Duomac-T, was added. Two minutes later, the washing water and air
were turned on.

Size Distributions

Feed Size ( v m ) Mass (%)


I Grinding +600 -1200 20.4
1200 g Galena Feed +300 -600 21.9
600 mL H20 +150 -300 24.7
pH ii.0 (NaOH) +74 -150 11.8
-74 21.2
I Total i00.0
20 min. grinding
25.4cm x 25.4 cm mill
50 2.5 cm dia. balls Flotation +29 17.9
Feed
I +22 -29 21.1
Flotation feed is +16 -22 16.6
filtered and split in +12 -16 18.7
three +9 -12 5.9
-9 19.8
I Total i00.0
Conditioning, pH Ii.0
5 mL KEX @ I Z w / w
2 mL MIBC @ I% w / w
I
Flotation
Washing water flowrate:
2.7 L/min; 0.20 minute
samples
I
Flotation products
are filtered, dried
and cyclosized

Fig. 3. Flowsheet of the experimental procedure of the galena flotation.


208

For a second series of tests, 30 kg of a high-grade galena sample (+90%


PbS) from Joplin, Missouri, was crushed, sieved and upgraded by Wilfley
tabling to over 99% PbS. The galena was then ground, filtered, conditioned
and floated, as shown in the flowsheet given in Fig. 3. The different flota-
tion products were then cyclosized.

RESULTS

(a) Bubble size distributions

A typical bubble diameter distribution is shown in Fig. 4 for an A F R of


5.63 litres per minute and a MIBC concentration of 0.98 X 10 -4 molar. The
bubble size distribution was represented by the 'average' bubble diameter
day , defined as the diameter of a bubble of specific surface A (day = 6 ds,
where ds is the Sauter mean diameter). Bubbles produced in the absence
of MIBC were large and irregularly shaped, as shown in Fig. 5. As expected,
the addition of MIBC reduced bubble size significantly, yielding spherical
bubbles (Fig. 6). Both with and without MIBC, total bubble surface in-
creased significantly as A F R increased (Fig. 8). Specific bubble surface de-
creased as A F R increased (Fig. 7). The effect of A F R and MIBC concentration
the average bubble radius ray (ray = 0.5 day ) are summarized in the follow-
ing relationship:
In (ray) = (0.0789 -+ 0.0070) G - (1.552 + 0.053) [MIBC] X 104
+ (0.343 + 0.019) [MIBC] 2 X 108 (3)
where the MIBC concentration is expressed in molar, ray in millimetres,
and G in litres per minute.

30

2O
t-
Z

10

O0 0.6 1.~2 118


BUBBLE DIAMETER (mm)
Fig. 4. Bubble diameter distribution (AFR: 5.63 L/rain; [MIBC] : 0.98 x 10 -4 molar).
209

Fig. 5. Bubbles produced in the absence of MIBC.

Fig. 6. Bubbles produced at a MIBC concentration of 0.98 x 10-' molar.


210

12 r
o /
,, l
,,, ! ~.2.0[
i \
i

W
~0E g l.o
,,,5i
.,_1
en ,
f

0 4 8
A F R (L/rain) AFR (L/min)

Fig. 7. B u b b l e specific surface as a f u n c t i o n o f A F R . A : [MIBC]= 0.98 x 10 -4 m o l a r ;


B: n o MIBC.
Fig. 8. T o t a l b u b b l e surface as a f u n c t i o n o f A F R . A : MIBC = 0.98 × 10-* molar; B:
n o MIBC.

(b ) Flotation rates

Typical results are shown in Figs. 9 and 10, where the natural logarithm
of the percentage mass of hydrophobic material in the cell is plotted as a
function of time. Curves a, b and c (Fig. 9) show the effect of AFRs of 1.32,
7.14 and 4.32 litres per minute, respectively, on the flotation rate of +74
--105 g m quartz. As suggested b y Cooper (1966), Reay and Ratcliff (1973),
Imaizumi and Inoue (1963) and many others, the transfer of material from
the slurry to the froth was assumed to follow first order kinetics with re-
spect to the mineral phase. The FRC is then the slope of the straight section
of the natural logarithm of the mass of hydrophobic mineral vs. time curve.
This slope was obtained by linear regression. Typical results are shown in
Fig. 11 for +74 - 1 0 5 p m and - 7 4 p m quartz, and in Fig. 12 for unclassified
and +22 - 2 9 p m galena. All results are summarized in Table II.
The FRC increases as A F R increases to reach a maximum at around 4 to 5
litres per minute, after which further increases in A F R affect flotation kinet-
ics adversely. For quartz, the +74 --105 um and +150 --210 pm fractions,
which exhibit the highest FRC, were also the most adversely affected by ex-
cessive AFRs. The - 7 4 ~m fraction, which floats most slowly, was the least
adversely affected. Galena behaved similarly, as its slowest floating size frac-
tion ( - 9 pm) was the least adversely affected by excessively high AFRs.
211

I00

10

0 .3 .6 .9 1.2 1.5
TIME (min)

Fig. 9. P e r c e n t mass o f silica in t h e slurry as a f u n c t i o n o f t i m e (air t u r n e d o n a t t - 0;


particle size: + 7 4 t o - - 1 0 5 pro). a: A F R ,. 1.32 L / r a i n ; b: A F R ,, 7.14 L / r a i n ; c : A F R ffi
4 . 3 2 L / r a i n ; d : h i n d e r e d f l o t a t i o n , e n c o u n t e r e d at l o w A F R a n d h i g h h y d r o p h o b i c particle
c o n c e n t r a t i o n ; e: free f l o t a t i o n ; p o r t i o n o f t h e curve used t o m e a s u r e t h e F R C .

100

10

u)

0 ,
0.5 I
1.0 1.'5
TIME (m in.)
Fig. 10. P e r c e n t mass o f galena in t h e slurry as a f u n c t i o n o f t i m e . a: t e s t 10, A F R ffi 0.96
L / r a i n ; b : t e s t 14, A F R ffi 4.67 L / m i n ; c: t e s t 16, A F R ffi 9.01 L / m i n .
212

5 "•

E
= X\
°
3
i • //

o
AFR (L/min)

Fig. 11. F R C as a f u n c t i o n o f A F R . A: + 7 4 to - 1 0 5 ~ m SiO2; B: - 7 4 u m SiO2.

2 • •

0 t
0 4 8 12
AFR (L/min)

Fig. 12. F R C as a f u n c t i o n o f A F R . A: unclassified PbS; B: + 2 2 to - - 2 9 ~ m Pbs.


213

TABLE II

Effect of A F R o n t h e FRC, experimental results


Cm
FRC = k G (L/min) exp(--c ! G) ffi G exp(1 -- G/Gm)
Qm
Particle No. of k S~. cl Sc 1 b Cm Gm
size tests (L-I) (%) (rain L-!) (%) (min -1 ) (L min -z )
(,m)
Quartz
+74 - 1 0 5 38 3.57 6 0.275 4 4.78 3.64
+148 - 2 1 0 15 3.48 11 0.271 8 4.72 3.69
-74 13 1.81 11 0.207 9 3.22 4.83
Galena
Overall 15 1.46 7 0.216 6 2.48 4.63
+29 11 2.01 12 0.219 8 3.38 4.57
+22 --29 11 2.10 7 0.225 5 3.43 4.44
+16 --22 11 1.64 12 0.215 9 2.81 4.65
+12 --16 11 2.28 10 0.233 7 3.60 4.29
+9 --12 11 1.23 11 0.216 8 2.09 4.63
--9 11 0.97 11 0.209 9 1.71 4.78

aStandard deviation of k; b standard deviation of c 2.

- T h e experimental error was found to be -+ 15% (-+1 standard deviation).


The most significant source of error was the variation in the conditioning of
the feed. Such errors are known to inhibit particle hydrophobicity more
often than to enhance it, thus causing negatively biased errors in the FRC.
These negative errors, especially at the maximum of the A F R - F R C curve,
will reduce the sharpness of this maximum, as shown in Fig. 12. In the
case of fine particles, entrainment at high AFRs may cause positively biased
errors in the FRC, which will decrease the adverse effect of high A F R on the
FRC (Fig. 11, curve B).

DISCUSSION

(a) Development of the model

Based on theoretical and experimental work by Anfruns and Kitchener


(1976), Jameson et al. (1977) derived the following relationship:
E e cc dp2 dr), .~, (4)
where E e = collision efficiency, i.e. the fraction of the particles in a bubble's
path which actually collide with the bubble (dimensionless); dp = particle
diameter; and db ffi bubble diameter.
Anfruns' and Kitchener's work was limited to a simple one-bubble system
in a very dilute mineral suspension. They used strongly methylated quartz as
the mineral phase. Consequently, nearly all the bubble-particle collisions re-
214

sulted in attachment. The main differences between their experimental sys-


tem and the present work (as well as industrial flotation) are the following:
(1) Hydrodynamic forces; whenever mineral slurries are vigorously stirred,
particle-bubble associations are subjected to various shearing and centrifugal
forces (in excess of 60 g). Thus the probability of detachment of the cap-
tured particles from bubbles is increased.
(2) Trajectories; bubble and particle trajectories in a turbulent fluid are
erratic, as opposed to the well determined vertical path of the bubbles in
Anfruns' and Kitchener'swork.
(3) Bubble-bubble or particle-particle interactions; slurries encountered in
the present work and in industry have bubble and particle concentration in
excess of 102/cm 3 and 105/cm 3 respectively. One of the major effects of such
high concentrations is to change bubble rise velocities, which can be higher or
lower than those of individual bubbles of the same diameter (Van Krevelin
and Hoftijzer, 1950). Another major effect is the saturation of bubble surface
with captured particles, which decreases flotation rates (Pogorely, 1962).
(4) Hydrophobicity; particles used in the present work and in industrial
systems are usually less hydrophobic than the surface methylated quartz par-
ticles used by Anfruns and Kitchener.
In view of these differences, eq. 5 can be applied to the present work only
if the following assumptions are valid:
(1) Attachment, following collision, and detachment are independent of
bubble size.
(2) The effect of bubble-bubble interaction is independent of bubble size.
(3) Bubble surface coverage with particles does n o t affect collision effi-
ciency (otherwise first order kinetics with respect to the mineral phase will
not be observed).
(4) Bubble size distributions can be characterized b y their average diam-
eter, day, to determine the effect of bubble size on collision efficiency.
Thus:
FRC (s-1) = c N ( A F R , day) E(dav) (5)
where: c = proportionality constant, which includes the effect of cell geom-
etry, hydrophobicity and solution chemistry on the FRC (ram-2); N (ARF,
day ) = rate of bubble production (s-i) = G / v , where v = volume of one
bubble (mm 3) (= (~/6 da3v); E(dav) = absolute collision efficiency, (mm2),
defined by Anfruns and Kitchener as the collision efficiency multiplied b y
the bubble cross-section.
From Anfruns and Kitchener (1976), E c = K* da½69, where K* is a pro-
portionality constant. Thus, eq. 5 can be rewritten:
FRC = 1.5 c K* G arab69 (6)
In the experimental system of the present work, day was characterized as a
function of A F R . For a MIBC concentration of 0,98 × 10 -s molar, eq. 3
yields:
215

In (day) = - 0 . 4 9 8 + (0.079 + 0.007) G (7)

where day is in millimetres and G in litres per minute. Substituting in eq. 7,


and setting k = 1.5 c K* exp[--2.69 (--0.498)] :
FRC = k G exp [(--0.213 +- 0.019) G] (8)
Experimental data was fitted to the more general expression:
FRC = k G e x p ( - c l G) (9)
King's model can also be used to predict the effect of AFR on the FRC, if
K* is assumed independent of AFR. Using day to represent the bubble size
distribution:
6 6
a --- - exp(0.498) exp[--(0.079 + 0.007)G] (mm -1) (10)
7rdav ~r
Then, from eqs. 1 and 2:
FRC = k* G e x p [ ( - 0 . 0 7 9 + 0.007)G] (11)
The proportionality constant k* is equal to:
C2 K2* r ¢ ( D ) S V -1 (12)
where C2 = (60/~) • exp(0.498) (cm -l) .

(b) Evaluation o f the proposed model

By differentiating eq. 8, it can be shown that the maximum FRC occurs


at an AFR given by:
Gm = 1/cl (13)
Table II shows the experimental values of Gm for quartz and galena. The
galena values are in good agreement with the value predicted by eq. 13 which
is 4.7 + 0.4 litres per minute. The quartz values are slightly lower, except for
the - 7 4 ~ m size class. Experimental results would also suggest that the maxi-
m u m FRC is not very clearly defined for the finest size classes. This effect is
not predicted by the model, but could be explained by mechanical entrain-
m e n t of fine particles into the froth phase. Although the experimental cell is
designed to minimize this effect, it could be significant for very fine particles
because of their slow flotation rate. Equation 11, derived from King's model,
predicts that Gm equals 12.7 + 1.1 litres per minute. This is much higher
than experimentally observed values, and indicates that the mass transfer
rate constant (K*)is a function of AFR, because of its effect on bubble size.

(c) Effect of bubble surface coverage with particles

Pogorely's assumption that the FRC will be proportional to the average


216

fraction of the bubble surface not covered by adhering particles was also in-
vestigated. Tomlinson and Flemming (1963) argued that " t h e degree of in-
hibition is not simply proportional to the percentage of bubble surface
covered by particles, for the particles collected on the advancing front of a
rising bubble are swiftly swept to the b o t t o m of the bubble..." Brown
(1965) presents photographic evidence to support this last statement. To
further investigate this point, Pogorely's model is compared to a new model
for which the capturing bubble surface is assumed completely different from
the surface where the collected particles are eventually swept. The following
notation is used:
M = non-floated mineral mass in the slurry (g)
G = A F R ( c m as -1)
o = bubble specific surface (cm-1)
r = uniform bubble retention time in the slurry (s)
y = m a x i m u m bubble load per unit bubble surface ( g c m -2)
w = m a x i m u m flotation rate -- saturated bubble surface (g s-1)
K = FRC for free flotation -- essentially a free bubble surface (s- 1)
S = fraction of the bubble surface covered by adhering particles
X = mass transfer rate coefficient -- mass of solids captured per mass of
solids in the slurry per unit of time per unit of free bubble surface
( c m - : s -1 )
In the free flotation mode, for which S - 0:
K =X G o ~ (14)
In the completely hindered flotation mode, in which the bubble surface is
completely saturated with solids after its stay in the slurry:
w=Goy (15)
The intermediate flotation mode is that in which the bubble surface is only
partially covered with solids after its stay in the slurry. For this mode,
Pogorely (1962) derived the following equation:
dM -X T M
- G a y [ 1 - exp( )] (16)
dt y
which yields, upon integration:
1 [. 1 -- e x p ( - K 2 M ) ] =
(M--Mo)+K: In L.~- ~ i . ] --Kit (17)

where: K1 = Gay = w; K2 = X r /y = K / w ; Mo = mass of hydrophobic solids in


the slurry at t = 0 (g); and t = flotation time (s).
The first term on the left side of eq. 17 is dominant in the hindered flota-
tion mode, whereas the second term is d o m i n a n t for free flotation. The FRC
(K) and the m a x i m u m flotation rate (w) can easily be measured in the com-
pletely free and hindered modes.
217

The new model is derived using Tomlinson and Flemming's (1963) de-
scription of the capture mechanism. The capturing surface can be assumed
to be the advancing front of the rising bubbles, whereas the actual location
of the captured particles is at the bottom of the bubble. Completely
hindered flotation would then take place when the bottom of the bubble
becomes saturated with collected particles. Further collection of particles
would then result in the displacement of an equal mass of previously col-
lected particles, or the failure of the newly collected particles to remain at-
tached to the bubble. In either case, the net gain of attached particles on the
bubble would be zero. Then, in the free flotation mode:
K = XGA*r (18)
where K, X, G and r were defined for the previous model, and A* = bubble
specific collecting surface -- restricted to the advancing front of the bubble
(cm-').
In the completely hindered flotation mode:
w = GA'y (19)
where w, G and y were defined in the previous model, and A' -- bubble spe-
cific loading surface -- restricted to the bottom of the bubble (cm-~).
The transition from completely hindered to completely free flotation
would then be abrupt, and take place when the free flotation rate predicted
by eq. 18 is equal to the completely hindered flotation rate given by eq. 19:
w =KM, M = w/K

This transition would take place after a flotation time of t* = (Mo - M ) / w .


The mass of solids in the cell would be described by the following model:
M =M0 - w t 0 < t <~ t* (20)
w
M = --. e x p [ - K ( t - t*)] t* < t < oo (21)
K
Figures 13a and 13b show experimental results of the flotation of 1 kg
samples of 74--105 ~m quartz, according to the procedure already de-
scribed, as well as the predicted results from the two proposed models. Fig-
ure 13a shows the mass of quartz in the cell as a function of time, which
should exhibit a linear relationship for completely hindered flotation. Figure
13b shows the same relationship using a logarithmic vertical scale. The plot
is now linear for free flotation. The model based on different particle cap-
turing and storing surfaces is in very good agreement with experimental re-
sults. A series of tests with AFR ranging from 1.5 to 6.0 litres per minute
and initial quartz mass ranging from 600 to 1000 grams also displayed the
same abrupt transition from completely hindered to completely free flota-
tion. This would indicate that the capture mechanism is close to that de-
scribed by Tomlinson and Flemming (1963). Pogorely's model yields a
218

"~kl < B¸
I x
ly;. < •

O9 Ii, \
"x
O3

A B it k
10--

o~ k<q, o"
O9

% - ....... to......... T . . . . (] to 2 /'


TIM E (min} T I M E (rain)

Fig. 13. Mass of quartz in the cell as a function of time. 1 : Pogorely's model, eq. 17; 2:
Proposed model, eqs. 20 and 21; A and B: completely hindered and completely free flo-
tation for the proposed model, respectively; B' : completely free flotation for King's sim-
plified model; squares: experimental points; to: transition time for the proposed model.
a. Linear--linear scale, b. Log--linear scale.

longer transition zone than either the alternate model or the experimental
system.

CONCLUSIONS

The findings of the present study are in good agreement with a flotation
model derived from previous work on simpler bubble-particle systems, which
predicts a drop in the bubble-particle collision efficiency as bubble diameter
increases. As AFR increases, the total bubble surface also increases, which
should increase the flotation rate. However, AFR increase also increases the
average bubble diameter, which causes a decrease in the flotation rate. At
low AFR, the first effect dominates; the second one is dominant at higher
AFRs. A maximum FRC is obtained at an intermediate AFR, when the two
effects are of equal magnitude. Around this maximum, the effect of changes
in AFR is limited.
It should be noted that these results apply only to a system having no
froth phase. Other workers (Mehrotra and Kaput, 1974; Engelbrecht and
Woodburn, 1975) reported that, with froth present, flotation rate increases
with AFR over its full range. Thus the influence of AFR on the behaviour of
the froth appears to warrant investigation.
219

REFERENCES

Anfruns, J.P. and Kitchener, J.A., 1976. The absolute rate of capture of single particles
by single bubbles. In: M.C. Fuerstenau (Editor), Flotation. A.M. Gaudin Memorial
Volume. Am. Inst. Min. Metall. Petrol. Eng., New York, N.Y., pp. 625--637.
Brown, D.J.A., 1965. A photographic study of froth flotation. Fuel Soc. J., 16: 22--34.
Cooper, H.R., 1966. Feedback process control of mineral flotation. Part 1: Development
of a model for froth flotation. Trans. Soc. Min. Eng., AIME, Dec., pp. 439--450.
Engelbrecht, J.A. and Woodburn, E.T., 1975. The effects of froth height, aeration rate
and gas precipitation on flotation. J. S. Aft. Inst. Min. Metal., Oct., pp. 125--131.
Flint, L.R. and Howarth, W.J., 1971. The collision efficiency of small particles with
spherical air bubbles. Chem. Eng. Sci., 25: 1155--1168.
Harris, C.C., 1974. Impeller speed, air, and power requirements in flotation machine
scale-up. Int. J. Miner. Process., 1: 51--64.
Harris, C.C., 1976. Flotation machines. In: M.C. Fuerstenau (Editor), Flotation. A.M.
Gaudin Memorial Volume. Am. Inst. Min. Met. Petrol. Eng., New York, N.Y., pp.
753--815.
Imaizumi, T. and Inoue, T., 1963. Kinetic considerations of froth flotation. In: A.
Roberts (Editor), Mineral Processing. Proc. 6th Int. Miner. Process. Congr., Cannes,
pp. 581--593.
Jameson, G.J., Nam, S. and Moo Young, M., 1977. Physical factors affecting recovery
rates in flotation. Miner. Sci. Eng., 9(3): 103--118.
King, R.P., 1972. Flotation research work of the N.I.M. Research Group, and the Depart-
ment of Chemical Engineering, University of Natal. J. S. Aft. Inst. Min. Metall., 72(4):
135--145.
King, R.P., 1973a. Computer-controlled flotation plants in Canada and Finland. National
Institute of Metallurgy Research Rep. No. 1517, p. 13.
King, R.P., 1973b. A computer programme for the simulation of the performance of a
flotation plant. Revised Report. N.I.M. Research Rep. No. 1436, p. 10.
Manlapig, E.V. and Spottiswood, D.J., 1978. Present practices in the computer control
of copper flotation plants. SME-AIME Fall Meet., Lake Buena Vista, Preprint No.
78-B-307, p. 12.
McKee, D.J., Fewings, J.H., Manlapig, E.V. and Lynch, A.J., 1976. Computer control of
chalcopyrite flotation at Mount Isa Mines Ltd. In: M.C. Fuerstenau (Editor), Flota-
tion. A.M. Gaudin Memorial Volume. Am. Inst. Min. Met. Petrol Eng., New
York, N.Y., pp. 994--1026.
Mehrotra, S.P. and Kapur, P.C., 1974. The effects of aeration rate particle size and pulp
density on the flotation rate distributions. Powder Technol., 9: 213--219.
Niemi, A. and Paakkinen, U., 1969. Simulation and control of flotation circuits. Auto-
matica, 5: 551--561.
Pogorely, A.D., 1962. Limits of the use of the kinetic equation proposed by K.F.
Beloglazov. Invest. Uysskikh Ucheb, Zavendenii. Tsvetn. Metall., 5: 33--50.
Reay, D. and Ratcliff, G.A., 1973. Removal of fine particles from water by dispersed air
flotation: Effects of bubble size and particle size on collision efficiency. Can. J. Chem.
Eng., 51: 178--185.
Reay, D. and Ratcliff, G.A., 1975. Experimental testing of the hydrodynamic collision
model of fine particle flotation. Can. J. Chem. Eng., 53: 481--486.
Sastry, K.V.S. and Fuerstenau, D.W., 1970. Theoretical analysis of a countercurrent flo-
tation column. Trans. AIME, 247: 46--52.
Sutherland, K.L., 1948. Kinetics of the flotation process. J. Phys. Chem., 52: 394--425.
Tomlinson, H.S. and Flemming, M.G., 1963. Flotation rate studies. In: A. Roberts
(Editor), Mineral Processing. Congr. of Miner. Process., Cannes, Pergamon Press, N.Y.,
pp. 563--579.
Van Krevelin, D.W. and Hoftijzer, P.J., 1950. Studies of gas-bubble formation. Chem.
Eng. Progr., 46(1): 29--35.

You might also like