Paper 13

Bubble Surface Area Flux: A Parameter To Characterize

Flotation Cells

Jim Finch'', Cesar Gomez', Colin Hardie', G. Leichtle',

R. Filippone2,and Donald Leroux2

'McGill University
Department of Mining and Metallurgical Engineering
M. H. Wong Building
36 10 University
Montreal, QC
H3A 2B2

'Telephone: 5 14 398-4755
Facsimile: 5 14 398-4492
e-mail: [email protected]

2NorandaTechnology Centre
240, Hymus Boulevard
Pointe-Claire, QC
H9R 1G5

Keywords: Flotation, bubble size, bubble surface area flux, gas rate, flotation cell

Presented 20 January 1999

199
 BUBBLE SURFACE AREA FLUX:
A PARAMETER TO CHARACTERIZE FLOTATION CELLS

J. A. Finch, C. 0. Gomez, C . Hardie, and G. Leichtle,

McGill University, Department of Mining and Metallurgical Engineering, Montreal, Canada

R. Filippone and D. Leroux

Noranda Technology Centre, Montreal, Canada

ABSTRACT

Bubble surface area flux (Sb) is the surface area of bubbles per unit time per unit cross-sectional
area of flotation machine. The units are: (m2 bubble surface/s)/(m2 cell) or s -'.
For bubbles of
mean size db (m), s b = 6-Qg/A.dbwhere Qg is the gas rate (m3/s), and A the cell cross-section
(m'). This parameter is now receiving wide attention as the possible key machine variable in
flotation diagnosis and modelling. The concept is described and the potential use of sb is
illustrated by work on flotation columns and mechanical cells. In column flotation of recycled
paper, the dependence of ink recovery on s b was the same regardless of size of column or type of
sparger. In mechanical cells, it has been established that the rate constant is related to Sb
independent of impeller type.
The use of s b is limited by the problem of measurement, primarily that of estimating db. A
prototype probe designed to estimate db (and hence sb) based on conductivity measurements is
described. Exploratory work with the probe in a mechanical cell suggests the promise of making
estimation of s b routine.

INTRODUCTION

Flotation is the result of interaction among variables associated with the feed particles,
chemistry and the machine (flotation cell). Trying to isolate the effects of each aids in process
diagnosis and is at the centre of attempts to model flotation. The machine variables that have
been investigated include the air flow number, the power number and specific power (Harris,
1976). None have survived as the basis for predicting flotation response perhaps because they
did not include directly the properties of the bubble swarm (or dispersion). Recently, Gorain
(1997) explored the relationship between the flotation rate constant and various "gas dispersion"
parameters notably: superficial air rate (Jg = Qg/A, where Qg is the volumetric gas rate and A is
cross-sectional area of cell), bubble size (db), gas holdup (Eg, i.e., the volumetric fraction of a cell
occupied by gas) and bubble surface area flux (sb = 6.Jg/db). Of them all, Gorain et al. (1997)
conclude that the latter, the surface area flux, correlated best with the flotation rate constant.
This concept was adopted from column flotation where it was used to estimate, among other
factors, the maximum amount of gas that could be injected (Finch and Dobby), 1990). Let's start
by reviewing the concept. Following that some examples of where the concept has been applied
will be described to illustrate the current interest and finally we will consider how to measure the
surface area flux.

200
 CONCEPT

The formal definition is the surface area of bubbles per unit time per unit cross-section of
a flotation cell. The concept is illustrated in Figure 1: air entering a cell at rate Qg(m3/s) is
divided by some bubble generating device into n bubbleskec each of surface area S (m2) and
they pass through an area of cell A (m2).

Figure 1. Illustration of the bubble surface area flux concept.

The derivation is then:

n-S
Sb=-
A

Assuming equal sized spherical bubbles, substituting for n (= 6 Qg/IT )d: and S (= n dz) gives

Since, by definition J, = Q,/A, then

The assumption of equal sized spheres simplifies the mathematics but does not invalidate
the concept. For a size distribution of bubbles the Sauter mean diameter should be substituted
for db in Eq. (3).

20 1
 The concept has immediate appeal in flotation. Particles are carried on the surface of
bubbles hence the production rate (e.g., t o m e s h ) from a cell must be related to the amount of
bubble surface area passing through the cell per unit time (or flux, Sb). The number of bubble-
particle collisions per unit time, which is related to the rate constant, depends on the number and
size of bubbles generated per unit time and the derivation shows these two factors are combined
in s b . Particle recovery by entrainment should also be related to S b as entrainment is related to
the amount of water transported by the bubbles.

EXAMPLES OF APPLICATION

The examples are selected to show that the surface area flux appears to be a fundamental
parameter; that is, regardless of how the sb is generated the results are the same. The work is
taken from Gorain et al. (1997) on sulphide mineral flotation in mechanical cells and our own
work on de-inking of paper using flotation columns. For mechanical cells, Figure 2 compares
the flotation rate constant as a b c t i o n of surface area flux for four impeller types. The
observation is that there is no difference between the impeller types. Figure 3 shows three
comparisons from work on de-inking: a). Columns of different diameter; b). Different internal
sparger types; and, c). Internal (porous) vs. external (static in-line mixer) spargers. All three
comparisons indicate that flotation perfonnance can be uniquely related to sb. Taken together
with the observations on mechanical cells there is strong evidence that the surface area flux is the
relevant machine variable. This is in large measure the reason that the parameter has excited
such interest.

n
I
I 1
E
*i
1
c, 0.8
E
c,
a
0.6
s
Q)
* 0.4
2
o

c,
3
n 0.2
a I 0 Don-Oliver
c,
0
E 0
0 50 100 150 200 250

Bubble Surface Area Flux (s-l)

Figure 2. Flotation rate constant as a function of bubble surface area flux: comparison of four impeller
types in a 2.8 m3 mechanical flotation cell (after Gorain et al., 1997).

202
 100

s-
I

90
80
~

w 70
a4
60
;* 50
8 I
2 40
,A A ! I ,

-
&
E
30
20
I - - - -
10 - j
7
0

Pilot Column
I
~

I
0 A

Bubble Surface Area Flux (8)

Figure 3a. Ink recovery as a function of bubble surface area flux: comparison of a 10 cm diameter
laboratory flotation column and a 50 cm diameter pilot flotation column (Leichtle, 1998).

100

-$ 90
80
70
60
’8
aJ
50

-
40
4A4 30
E
20
I
0 Jetting
;
I -
I

0 20 40 60 80 100 120 140 160

BubbIe Surface Area Flux (s-l)

Figure 3b. Ink recovery as a function of bubble surface area flux: comparison of internal sparger types
(S.S. = porous stainless steel, Jetting = MinnovEX variable gap sparger) in a 50 cm diameter pilot
flotation column (Leichtle, 1998).

203
 100 1 I I

I
1
1

2 40
30--
=o
I

0 S.S. Porous Sparger -

E 20 I I I
I I
i + Static Mixer Sparger
I I
-
I
I I I 1 1

0- I II 1

0 20 40 60 80 100 120 140

Bubble Surface Area Flux (8)

Figure 3c. Ink recovery as a function of bubble surface area flux: comparison of a internal (porous
stainless steel) sparger and an external (static in-line mixer) sparger in a 10 cm diameter laboratory
flotation column (Hardie, 1998).

MEASUREMENT

The insight represented by the concept is neither particularly profound nor particularly
new. Its use has been slow to take off, however, because of the difficulty of measurement, which
is primarily the problem of estimating bubble size. Gorain et al. opted for direct measurement on
samples withdrawn from the system and determined the Sauter mean size from the distribution.
Another approach is to estimate db from measurements of gas holdup using drift flux analysis
(Finch and Dobby, 1990). This offers the prospect of an automated technique and is the one we
are pursuing. It was used for example, in determining db in the column de-inking work.
Exploiting the approach in that case is aided by the convenient geometry of the column and the
low solids content (ca. 1% by wt) of paper pulps which means that the suspension density is for
all practical purposes that of water and pressure can be used to measure gas holdup. In mineral
systems, accurate gas holdup estimation from pressure measurements requires the density of the
pulp, a variable difficult to measure in-situ or to estimate from theoretical considerations.
Experience has demonstrated that conductivity measurements are a good alternative to measure
gas holdup.
The technique being tested is based on estimating values for the gas holdup and air flow
rate from conductivity measurements collected using a device which includes two conductivity
cells. One, the open cell, is designed to measure the conductivity of the pulphubble dispersion
( k p b ) and the other, the syphon cell, of the pulp alone (kp). These two values are used in an

204
equation based on Maxwell's model to calculate the volumetric fraction of gas in the dispersion
(Uribe-Salas et al., 1992):

I - - kPb
kP
Eg = (4)
1 +- 0.5-kPb
kP

The gas flow rates are determined from the time it takes to displace the free surface of the
pulp between two electrodes separated a known distance by the air bubbles accumulating at the
top of the open cell. Gas holdup and gas flow rate are used to estimate bubble size from drift
flux analysis, and gas flow rate and bubble size are subsequently combined to calculate bubble
surface area flux (Equation 3).
The approach has been established in flotation columns where the flow regime is suited
to drift flux analysis. However, the proposed sensor permits extension to agitated mechanical
cells as the design imposes the required flow conditions. The cell measuring the dispersion
conductivity is in the form of a vertical cylindrical tube open at both ends; therefore, once the
bubbles enter the cell through the bottom they rise vertically with no radial velocity component,
a situation which is accurately described by drift flux analysis.

THE SENSOR AND ITS OPERATION

A schematic of the sensor, which was made entirely of PVC, is presented in Figure 4.
The two conductivity cells in the form of parallel vertical cylinders are located at the bottom end.
Both cells have three stainless steel ring electrodes separated a distance of 0.1 m and flush
mounted to the internal wall of the PVC tube. The purpose of this three-ring configuration is to
limit the electric field to the region between the electrodes by maintaining the end electrodes at
the same potential. The open cell, used to measure the conductivity of the pulphubble
dispersion, is 0.1 m in internal diameter (4"). This cell is fully open at its lower end, which is
kept below the froth, to allow for the free flow of bubble and pulp. The cylinder is extended and
emerges from the flotation cell to permit manual operation of a valve.
The syphon cell, used to measure the conductivity of the bubble-free pulp, is 0.075 m in
internal diameter (3"). This cell is completely immersed in the pulp zone (below the fioth layer)
and is fully open at the top but ends in a small orifice at the bottom end. This geometry prevents
rising bubbles from moving into the cell which then fills only with pulp entering via the top. A
downward flow of pulp develops as a consequence of the difference in hydrostatic pressure
inside (only pulp) and outside (pulp and bubbles) the cell (hence, the description "syphonncell).
To try to ensure that the pulp and bubbles entering the open cell are not affected by this pulp
exiting the syphon cell the bottom of the syphon cell is located about 0.1 m above the bottom end
of the open cell.

205
 isolating valve
-
-
c

open cell syphon cell

(10 cm i.d.) (7.5 cm i.d.)

Figure 4. Schematic of the bubble surface area flux probe.

With the sensor in position, a froth layer develops within the open cell and grows to
overflow through the open valve. Conductivity readings are collected from both cells using an
electronic circuit that sequentially connects both cells to a conductivity meter. This circuit also
digitizes the output of the conductivity meter which can then be transmitted to a computer and
gas holdup is calculated. To measure the gas flow rate, the valve is closed and the conductivity
of the open cell is monitored. As gas accumulates at the top of the open cell, the froth/pulp
interface is pushed down the cell and the variation of the conductivity with time is recorded. The
conductivity decreases from a high value when the cell is filled with pulp and bubbles to a low
value when the cell is filled only with gas (and froth). Because of the three-electrode
configuration, the resulting curve is s-shaped (Figure 5). As the volume of the test section is
known (internal diameter and distance between electrodes) the gas flow rate is calculated from
the interface displacement velocity determined from the conductivity/time curve.

206
 I
I i I

I
3 I
I
01
I
I I
€ 4 !
! I I i
I
I
I
I

a z
m
E
G 1

0 20 40 60 80 100 120 140

Time (sec)

Figure 5. Data used to determine gas flow rate.

PLANT TESTING

A prototype probe was evaluated in plant trials at Brunswick Mines (Division of Noranda
Inc.). The tests were run in the second cell of a Zinc cleaning flotation bank (Denver cells,
nominal volume 7 m3). Vertical baffles separate the cells within the bank; each cell has a central
orifice (about 0.6 m diameter) for the slurry to flow between cells. The cell has a standard stator
and rotor-impeller assembly. Manually-controlled air is injected into the stator-rotor gap using a
common source for all the cells in the bank; no gas flow rate measurements are made on
individual cells.
The sensor was installed near the middle position between the rotor and the froth launder,
and with its lower end at 1.5 m depth. The air flow rate was increased from the lowest level to
- -
still yield froth overflow (Jg 0.25 c d s ) to the point where the froth layer is not stable (J, 2
c d s ) . The results are presented in Figure 6 which gives S b as a function of superficial air
velocity, J , with Sb calculated assuming water or slurry properties. The trend obtained was
expected based on results obtained in columns and mechanical cells using other measuring
techniques: as the gas rate increases so does s b , and a maximum s b is reached. The results
demonstrated the sensor was reliable and consistent: the results of Figure 6 were obtained by
repeatedly increasing and decreasing the gas rate. The results also showed that increasing the
gas flow rate above ca.1 cm/s does not generate more bubble surface flux, and the maximum is
about 100 s'l (assuming water properties), a value similar to those reported in flotation columns
and for mechanical cells (Finch and Dobby, 1990; Gorain et al., 1997).
Figure 6 illustrates a decision that must be made, chosing the liquid properties (density
and viscosity) used to calculate bubble size (and hence s b ) in drift flux analysis. The slurry

207
properties were measured for one set of conditions (density = 1.22 g/cm3, and viscosity = 0.016
g/cm s). As Figure 6 emphasizes the trend is the same whether water or slurry properties are
used. The choice of water is the simple, more practical one.

120 1 I 1

100

80

60

40

20

0
0 0.5 1 1.5 2

Superficial Gas Rate (cm/s)

Figure 6. Bubble surface area rate as a function of gas rate. (Sbcalculated using slurry or water
properties).

CONCLUSIONS

Bubble surface area flux is emerging as the machine gas dispersion property of choice.
Some of the evidence giving rise to this is illustrated. The main problem is measurement. A
technique based on conductivity is described.

REFERENCES

Finch, J.A. and Dobby, G.S. (1990) Column Flotation, Pergamon Press

Gorain, B.K., Franzidis, J.P., and Manlapig, E.V. (1997) "Studies on Impeller Type, Impeller
Speed and Air Flow Rate in an Industrial Scale Flotation Cell. Part 4: Effect of Bubble
Surface Area Flux on Flotation Performance" Minerah Engineering. Vol. 10 No. 4. pp.
367-379.

Hardie, C.A. (1998) Masters Thesis, McGill University, Canada,

208
Harris, C.C. (1976) "FlotationMachines'' In: Flotation, A.M Gaudin Memorial Volume (M.C.
Fuerstenau ed.) SME, New York, Chapter 27 pp. 753-815.

Leichtle, G.F. (1998) Masters Thesis, McGill University, Canada

Uribe-Salas, A., Gomez, C.O., and Finch, J.A. (1994) "A Conductivity Technique for Gas and
Solid Holdup Determination in Three-Phase Reactors" Chemical Engineering Science.
Vol. 49 NO.1, pp. 1-10.

209