Minerals Engineering, Vol. 8, No. 6, pp.

615--635, 1995
Copyright © 1995 Elsevier Science Ltd
Pergamon Printed in Great Britain. All rights reserved
0892--6875(95)00025-9 0892-6875/95 $9.50+0.00

STUDIES ON IMPELLER TYPE, IMPELLER SPEED AND AIR FLOW RATE

IN AN INDUSTRIAL SCALE FLOTATION CELL--
PART 1: EFFECT ON BUBBLE SIZE DISTRIBUTION

B.K. GORAIN§, J.-P. FRANZIDIS~f * and E.V. MANLAPIG§

§ Julius Kruttschnitt Mineral Research Centre, Isles Road, Indooroopilly, Australia

1" Department of Chemical Engineering, University of Cape Town, South Africa
* Presently at Julius Kruttschnitt Mineral Research Centre, Australia
(Received 24 October 1994; accepted I December 1994)

ABSTRACT

Bubble size distributions were measured at different locations in a 2.8 m 3 portable

industrial scale sub-aeration flotation cell, treating zinc cleaner feed in the Hellyer
Concentrator in Tasmania, Australia. The cell was fitted in turn with four different
impeller-stator systems, and operated over a range of airflow rates and impeller speeds.
The mean bubble size was found to increase with increase in air flow rate at different
locations in the cell, for all four impellers, and to decrease with increase in impeller
speed. The mean bubble size was largest close to the impeller shaft and smallest at the
impeller discharge point, for all the impellers. The shape of the bubble size distribution
also changed with location in the cell. The "global mean" bubble size calculated by
simple arithmetic average of the values at six locations in the cell coincided remarkably
well with tJie mean bubble size measured halfway between the impeller shaft and the side
of the cell, at the top of the pulp. In general, the impellers produced "global mean"
values of 1.0 mm or less at the manufacturer's recommended impeller speed.

Keywords
Industrial flotation cell, Outokumpu impeller, Dorr--Oliver impeller, Agitair Chile-X
impeller, Agitair Pipsa impeller, Bubble size distribution, aeration rate, impeller speed

INTRODUC~ON

The importance of the bubble size in flotation cells has been appreciated since the very early days of froth
flotation, when Nevett [1] noted that "it is essential that the air should be completely atomised in the pulp
and not allowed to be distributed through the pulp in the form of comparatively large bubbles". Later
Bennett et al. [2] studied the effect of bubble size in the flotation of coal in a laboratory flotation cell, using
direct weighing and high-speed cinematography, and concluded that for a given rate of air supply coal
flotation rate would be increased by dividing the air into the largest possible number of small bubbles, i.e.
by reducing the bubble size and providing more bubbles.

Since then, many experimental and theoretical investigations into the effect of bubble size on particle
flotation have been carried out. But, as pointed out recently by Diaz-Penafiel and Dobby [3], much of the
data reported was obtained under highly idealized conditions of dilute slurries and single bubbles, on
account of the difficulties in determining bubble size in real flotation slurries. Moreover, these laboratory
investigations have generally employed bubbles smaller than 1.0 mm (frequently smaller than 0.5 mm) while
in practice, in column and sub-aeration cells, bubble sizes range between 0.5 and 2.0 mm.

615
616 B.K. GORAINet al.

The advent in the 1980's of column cells, pneumatic cells, the Jameson cell, and the air-sparged
hydrocyclone, all of which employ methods other than impellers to generate air bubbles, has heightened
interest in the effect of bubble size on flotation performance. The method of drift flux analysis, by which
an average bubble size may be calculated from air flow rate and air hold-up determinations, has been
applied to column cells and found to correlate well with photographic measurements [4]. Using bubble size
values obtained from drift flux analysis, investigations with column cells have confirmed [3] that smaller
bubbles deliver a higher rate constant. However this work has yet to be extended to industrial scale
conventional sub-aeration flotation cells.

The first step in the determination of the effect of bubble size on flotation in industrial scale sub-aeration
cells is the measurement of bubble size distributions in these cells under various conditions. Physical factors
that would be expected to affect bubble size (among others) are impeller type and speed, as well as air flow
rate. The bubble size would probably vary with location in the cell according to the level of turbulence and
the flow patterns prevailing, but it would be useful to find some "global mean" value to represent the
overall conditions in the cell in terms of bubble size.

The development of the U.C.T bubble size analyser, with its associated bubble sampler, allows rapid and
accurate determination of bubble size distributions in three phase (solid-water-gas) systems. The method
relies on direct capture of bubbles in a capillary [5], and has been used to measure bubble size distributions
in slurries containing up to 40% solids [6]. With the use of one of these instruments, bubble size
distributions have been measured in an industrial scale(2.8 m 3) test flotation cell operated in the zinc cleaner
circuit at Hellyer Mine operated by Aberfoyle Resources Limited in Tasmania.

The cell was fitted in turn with four different impellers commonly used in flotation cells, and operated at
various combinations of air flow rate and impeller speed, at values around those recommended by the
manufacturers. Measurements were made at six different locations in the cell. It is believed that this is the
first time that such measurements have been made in an industrial scale sub-aeration flotation cell, operating
on-site, using run-of-plant ore. This paper presents the results of the investigation. A subsequent publication
will consider the effect of the measured bubble size distributions on the flotation performance in the cell.

It should be emphasised that these experiments were not conducted to compare the performance or
characteristics of the different impellers used. The impellers were not necessarily operated under optimum
(recommended) conditions, or in a cell of optimum design, and were utilised solely to provide a range of
hydrodynamic conditions for the study.

EXPERIMENTAL

Flotation cell, impellers and associated instrumentation

The 2.8 m 3 test flotation cell was designed at the JKMRC, and manufactured out of mild steel by Newfab
constructions in Newcastle, Australia. A schematic diagram of the cell is shown in Figure 1. The cell is of
standard design used in the industry, with provision to fit in different impellers and stators. A 11 kW A.C.
motor was mounted on the top of the cell, and connected to the impeller shaft with a pulley, with a variable
speed drive to facilitate altering the impeller speed. The rig incorporates a feed box, a discharge box, and
a tailings sump, as well as a ladder and walkway on one side to enable an operator to take samples or make
measurements within the cell (of e.g. bubble size distributions or air hold-up). Though large, the unit was
deliberately made to be portable, so as to enable it to be installed in situ in different locations for test
purposes.

The impellers used in this testwork and their dimensions are listed in Table 1. Figure 2 shows a schematic
diagram of the different impellers and stators. The Chile-X and Pipsa impellers (both of the Agitair family)
are of the same design, with a flat disc and fingers attached around the bottom of the disc, but the Pipsa
impeller has an additional pumping mechanism attached on to the top of what is a Chile--X impeller. Both
of these impellers use the same stator. The Outokumpu and the Dorr-Oliver impellers both have downward
tapered blades but in the case of the Outokumpu impeller separate slots are provided for air and slurry. A
 Bubble size distribution 617

multimix stator was used for the Outokumpu impeller. All of the impellers were purchased from the
respective manufacturers in Australia.
.m

0.61 t 1.68 m
Q___.

(a)

©
Q ----D~
®

®
o_.

Q m Q

(~-..--....4P-

Q -.=--ID-
k\\"J

(b)
Fig.1 Test flotation cell a: Top view; b: Elevation
1) Discharge box, 2) Concentrate launder, 3) Feed box, 4) Setup box, 5) Pulley torque transducer, 6) Drive pulley with guard, 7)
11 kW A.C motor with :;uppon bracket, 8) Air inlet pipe with control valve, 9) Concentrate launder discharge point, 10) Impeller
shaft, 11) Tailings discharge box, 12) Base support for the flotation cell, 13) Sump discharge point for
combined concentrate and tails.
618 B. K. GORAIN et al.

TABLE 1 Dimensions and settings of impellers used in the testwork

Impeller type Impeller~ameter Impeller Height (an) Bottom

(cm) dearance(cm)

Chlle-X 68.6 17.5 11.4

Plpsa 68.6 27.0 11.4

Outokumpu 50.0 30.0 10.0

Dorr-Oliver 37.5 25.0 4.5

(a) (b)

(¢) (d)
Fig.2 Impellers/stators used for the test work
a: Pipsa; b: Chile-X; c: DoE-Oliver; d: Outokumpu
 Bubble size distribution 619

The test cell was fitted with a rotameter on the feed air pipeline to regulate and measure the air flow rate.
Air was delivered to the call at approximately 12 kPa from the plant air line. A compact unit for measuring
the power input, impeller speed and torque electronically was fitted on the top of the pulley. The torque was
measured between the rotating shaft and the pulley using a high precision torque transducer. Power input
was given by the multiple of impeller speed and torque.

Feed and operating conditions

The test rig was installed in the HeUyer Concentrator1 in Tasmania, operated by Aberfoyle Resources
Limited. The rig was located in a place where the feed to the test cell could be taken easily from the feed
pipe to the zinc clearier circuit, and the concentrate and tails of the cell could be easily recycled back to the
same circuit. The zinc cleaner feed material at Hellyer has already passed through copper and lead flotation
stages, and is of rea~;onably constant composition. Feed solids concentration is of the order of 35%. The
particle size is rather fine, with a Ps0 of 20 to 25 micron. The stream contains frother known commercially
as "Interfroth 56", added in the copper rougher circuit at the rate of 46 g/t; in addition, potassium amyl
xanthate(PAX) colle~:tor is added in the zinc rougher circuit together with lime and copper sulphate. PAX
is also added to the :first and third cells of the first zinc cleaner bank at a low rate.

The chemical conditions in the test cell were kept identical to the zinc cleaner feed stream, i.e. no chemicals
were added to the test feed during any of the experiments. Bubble size measurements were made at four
different air flow rates at each of four impeller speeds, i.e. a total of 16 different operating conditions for
each of the four impellers. The same four air flow rates were used for each impeller, viz. 16.5, 28.3, 42.5
and 56.7 1/sec (corresponding to 35, 60, 90 and 120 cubic feet per minute, respectively). The impeller
speeds investigated for each impeller are shown in Table 2, and were selected around the values
recommended by each respective manufacturer for the particular cell size and application. These values are
also indicated in Table 2. The slurry feed rate to the test cell was constant throughout the tests, and the
distance between the pulp-froth interface and the overflow lip was maintained at a constant 36 cm.

TABLE 2 Impeller speeds investigated in the test work.

Impeller type Impeller speeds(rpm)

Chile-X 140, 160", 180, 200

Pipsa 10.5, 120", 140, 160

Ontokumpu 18,5,205*, 225, 245

Dorr-Oliver 235, 255*, 275, 295

Note : The asterisk mark (*) indicates the manufacturers recommended speed

Bubble size measurement

Bubble size distributions were measured using a U.C.T. bubble size analyser fitted with a bubble sampler,
as described by Tucker et al. [6]. With the aid of a long handle, the unit could be submersed fairly deeply
(approximately 1.0 m) into the cell to take samples at different locations. In

la process;descriptionof the HellyerConcentratorand a flowsheetmay be foundin Richmond[7]

620 B.K. GORAINet al.

between each measurement, the bubble sampler was filled with water containing Interfroth 56 frother, to
reduce the probability of coalescence while drawing bubbles from the slurry. Each reading was taken over
a period of 5 minutes, during which about 4000 bubbles were counted (per test). Software supplied with
the equipment was used to calculate the mean bubble size diameter and standard deviation for each
measured bubble size distribution.

Bubble size distributions were measured at six different locations in the test cell for each of the 64 different
operating conditions tested. Measurements were made in a vertical plane as shown in Figures 3a and 3b.
Locations 1 and 2 are closest to the impeller, with location 2 vertically below location 1. Locations 3
(above) and 4 (below) are located in the middle of the cell, while locations 5 (above) and 6 (below) are
closest to the wall. These locations were selected in the same way as Cochrane [8]; they follow
approximately a possible basic flow pattern (circulation loop) generated by the impellers in the cell, as
indicated in Figure 3b.

TOP VIEW

3,4

,I I
(a)

ELEVATION

I , i
2 0 cm 3 0 crn Z5 o n

(b)
Fig.3 Locations of measurement in the cell,with possible flow paRern (circulationloop).
a: Top view; b: Elevation
 Bubble size distribution 621

Tests were conducted with each impeller in turn over a period of several weeks in April and May 1994. The
procedure involved keeping the impeller speed fixed, setting the air flow rate to the desired value, allowing
20 minutes to elapse for conditions in the cell to stabilise and measuring the bubble size distribution at the
six different locations in the cell. Each set of six measurements usually took about 45 minutes, after which
the air flow rate w~ls set to a new value, and the process repeated. Once measurements had been taken at
all four air flow rates, the impeller speed was changed to the new desired value and the whole procedure
repeated from the ~eginning. For each impeller, the bubble size measurements for all 16 conditions were
carried out on two consecutive days.

Reproducibility of bubble size measurements

Table 3 shows the r,~sults of reproducibility tests carried out with the bubble size analyser at the start of the
test program. Four measurements were made at each of the six locations in the cell, and the mean bubble
size and standard deviation of each of the distributions were calculated. The standard deviation at each
location is an indication of the spread of the bubble size distribution measured at that location. The
reproducibility is excellent except at locations 2 and 4 which are in the turbulent zone of the cell. At these
locations the reproducibility is good, though not as good as at the other locations. The good reproducibility
may be attributed at least in part to the large number of bubbles (approximately 4000) sampled by the
bubble size analyser per reading.

T A B L E 3 Reproducibility of bubble size measurements at different cell locations in the test ceil.

Run 1 Run 2 Run3 Run 4

Mean size s.d Mean size s.d Mean size s.d Mean size s.d
Location 1 1.09 0.32 1.11 0.32 1.09 0.32 1.07 0.32

Location 2 1.43 0.41 1.37 0.44 1.45 0.44 1.33 0.42

Location 3 0.79 0.26 0.81 0.25 0.82 0.25 0.83 02.7

Location 4 0.59 0.20 0.66 0.21 0.60 0.19 0.57 0.17

Location S 0.73 0.22 0.78 02.4 0 76 0.23 0.77 02.3

Location 6 0.60 0.24 0.66 0.21 0.66 0.22 0.65 0.22

RESULTS

Overall results

The results of all the bubble size measurements obtained during this investigation are listed in Tables 4a
to 4d. Each Table presents, for a particular impeller type, the mean bubble size in mm measured at each
of the six locations for each of the 16 operating conditions tested.

It is apparent from even a cursory glance at all the results together that the values differ from impeller to
impeller, and vary with air flow rate, impeller speed and location within the flotation cell. Various patterns
are discernible, and will be discussed below. The mean bubble sizes ranged from 0.4 mm to 2.0 ram, which
is typical of the values generally accepted as occurring within industrial sub-aeration flotation cells. What
makes the data different from those reported previously is that each value in the Table is the mean of
approximately 4000 bubbles measured; the entire sample population represents a count in excess of 1.5
million bubbles. Tl~is is a very big sample, and was made possible by the technique of bubble size
measurement used.
622 B. K. GORAIN et al.

T A B L E 4a Bubble size measured at different locations and operating conditions

(impeller speed, air flow rate): Pipsa impeller.

Operatin8 Mean bubble size (nun) "Global Input

conditions Mean" power (kW)
(nun)
Location Location Location Location Location Location
1 2 $ 4 S 6
105 rpm
16.5 l/sec 0.93 0.97 0.89 0.74 0.81 0.79 0.86 1.73
28.3 l/see 1.28 1.34 1.22 0.97 1.09 1.04 1.16 1.70
42.5 l/sec 1.47 1.56 1.31 1.01 1.19 1.10 1.27 1.69
56.7 l/sec 1.89 2.18 1.75 1.45 1.63 1.58 1.75 1.67
120 rpm
16.5 1 / ~ 0.85 0.91 0.80 0.69 0.76 0.73 0.79 2.95
28.3 l/sec 0.99 1.09 0.95 0.72 0.88 0.82 0.91 2.77
42.5 l/see 1.20 133 1.12 0.89 1.03 0.96 1.09 2.70
56.7 l/sec 1.59 1.99 1.41 1.09 1.33 1.26 1.45 235
140 rpm
16.5 l/sec 0.77 0.82 0.75 0.61 0.71 0.67 0.72 4.46
28.3 l/sec 1.01 1.05 0.92 0.70 0.87 0.80 0.89 4.15
42.5 l/sec 1.09 1.17 0.97 0.76 0.88 0.84 0.95 3.94
56.7 l/sec 1.50 1.78 1.38 1.09 1.28 1.20 1.37 3.81
160 rpm
16.5 l/sec 0.70 0.75 0.68 0.6O 0.65 0.63 0.67 5 .55
283 l/sec 0.90 0.94 0.85 0.67 0.79 0.72 0.81 4.65
42.5 l/see 0.93 0.98 0.86 0.71 0.82 0.81 0.85 4.12
56.7 l/sec 1.18 1.33 0.99 0.80 0.92 0.88 1.02 3.64

T A B L E 4b Bubble size measured at different locations and operating conditions

(impeller speed, air flow rate): Chile--X impeller.

Operating Mean bubble size (nun) "Global Input

Conditions ntean" power
(nmO (kW}
Location Location Location Location Location Location
1 2 3 4 S 6
140 rpm
16.5 l/sec 0.87 0.92 0.83 0.67 0.76 0.70 0.79 2.82
28.3 l/sec 1.16 1.26 1.08 0.82 0.95 0.89 1.03 2.70
42.5 l/sec 1.32 1.36 1.26 0.97 1.19 1.12 1.20 2.62
56.7 l/sec 1.46 1.51 1.36 1.15 1.28 1.21 1.33 2.55
160 rpm
165 l/sec 0.78 0.84 0.70 0.57 0.66 0.64 0.70 3.20
283 l/sec 0.83 0.86 0.76 0.63 0.72 0.67 0.75 3.08
42.5 l/sec 1.23 1.32 0.99 0.71 0.90 0.80 0.99 2.95
56.7 l/sec 1.24 1.41 1.11 0.80 1.02 0.85 1.07 2.82
180 rpm
16.5 l/sec 0.71 0.76 0.66 0.53 0.63 0.59 0.65 3.82
28.3 l/sec 0.82 0.88 0.75 0.57 0.71 0.65 0.73 3.78
423 l/sec 1.01 1.16 0.85 0.65 0.76 0.70 0.86 3.67
56.7 I/sec 1.12 1.26 1.01 0.74 0.92 0.81 0.98 3.51
200 rpm
16.5 l/see 0.67 0.70 0.60 0.45 0.56 0.52 0.59
28.3 l/sec O.66 0.72 0,66 0.45 0.54 0,.52 0.59 4,o
423 l/sec 0.76 0,81 0.66 0.53 0.65 0.58 0.67 4.32
56.7 l/sec 1.08 1.16 1.00 0.69 0~6 0.77 0.93 4.!8
 Bubble size distribution 623

T A B L E 4c Bubble size measured at different locations and operating conditions

(impeller speed, air flow rate): Dorr-Oliver impeller.

Operaan8 Mean bubble size (ram) "Global Input

ConditionJ mean" power
(ram) (kW')
LoC|tiOfl Location Location Location Location Location
1 2 3 4 5 6
235 rpm
16.5 l/see 0..92 1.14 0.85 0.56 0.66 0.76 0.82 1.97
28-3 l/see 095 1.13 0.88 0.63 0.71 0.82 0.85 1.72
425 l/see 111 1.03 0.93 0.70 0.85 0.75 0.90 1.53
56.7 i/see 1,,36 1.21 1.12 0.82 1.07 0.97 1.09 1.36
255 rpm
16.5 l/sec 0.84 0.79 0.68 O.48 0.53 0.59 0.65 2.76
28-3 l/see 0,¸92 0.84 0.90 0.56 0.69 0.77 0.82 2.42
42.5 1/see 1,16 0.97 0.87 0.66 0.72 0.78 0.86 2.12
56.7 l/see 1.25 1.15 1.04 0.74 0.96 0.84 1.08 1.78
275 rpm
165 l/sec 0.63 0.75 0.58 0.43 0.49 0.51 0.57 2.98
28-3 l/sec 0.78 0.92 0.70 0.46 0.53 0.60 0.67 2.72
425 l/sec 1.08 0.93 0.83 0.52 0.76 0.65 0.80 2.26
56.7 l/sec 1.12 1.07 0.86 0.59 0.79 0.71 0.86 1.99
295 rpm
16.5 l/see 0.67 0.77 0.64 0.41 0.49 0.57 0.59 3.32
28.3 l/see 0.71 0.85 0.67 0.43 0.49 0.60 0.63 2.93
42.5 l/sec 0.95 0.80 0.75 0.53 0,70 0.55 0.71 2.50
56.7 l/sec 1.14 0.94 0.89 0.59 0.78 0.67 0.84 2.19

T A B L E 4d Bubble size measured at different locations and operating conditions

(impeller speed, air flow rate): Outokumpu impeller.

Operating Mean bubble size (nun) "Global Input

Conditions mean" Power
(mm) (kW)
Location Location Location Location Location Location
1 2 3 4 5 6
185 rpm
165 l/see (3.96 0.92 0.85 0.65 0.72 0.77 0.81 4.31
28-3 l/see 1.07 0.97 0.91 0.68 0.77 0.83 0.87 3.29
42.5 l/see 11.13 1.01 0.95 0.61 0.87 0.78 0.89 2.96
56.7 l/sec 11.32 1.11 0.97 0.72 0.88 0.76 0.96 2.64
205 rpm
16.5 l/see 0.73 0.89 0.57 0.42 0.66 0.49 0.63 5.47
28,3 l/see 0.97 0.84 0.72 0.47 0.53 0.60 0.69 4.34
425 l/see 1.12 0.92 0.84 0.57 0.78 0.67 0.82 3 .57
56.7 l/see 1.21 0.98 0.89 0.68 0.83 0.77 0.89 3.28
225 rpm
165 l/sec 0.61 0.68 0.53 0.41 0.47 0.49 0.53 6.80
28,3 l/sec 0.75 0.66 0.61 0.42 0.49 0.51 0.57 5.35
425 l/sec 0.79 0.83 0.69 0.49 0.63 0.54 0.66 4.50
56.7 l/see 0.99 0.89 0.83 0.60 0.74 0.69 0.79 3.90
245 rpm
16.5 l/sec 0.65 0.70 0.56 0.40 0.43 0.49 0.54 8.41
i 28-3 l/sec 0.73 0.79 0.60 0.43 0.47 0.51 0.59 6.,54
42.5 i/see L04 0.92 0.88 0.62 0.81 0.73 0.83 5.56
56.7 l/sec IL.30 1.13 0.99 0.88 0.92 0.88 1.02 4.58
624 B.K. GORAINet al.

Effect of air flow rate

The variation of mean bubble size with air flow rate at various impeller speeds, at each of the six measuring
locations in the test cell, is shown in Figures 4a to 4d for the Pipsa, Chile-X, Don'-Oliver and Outokumpu
impellers, respectively. For all the impellers at all the locations, the mean bubble size generally increased
with increase in air flow rate, with the manner of the increase depending on the impeller type and speed.
At the lowest air flow rate employed (16.5 l/see), all four impellers were able to disperse the air adequately
with a small mean bubble size (<l.0 mm) at all the locations even at the lowest impeller speeds selected
for the test work. As the air flow rate was increased, however, the ability of the impellers to disperse the
air into small bubbles varied from impeller to impeller.

LOCATION 1 LOCATION 2
2.3 • 2.3
A 2.1" 2.1
1.9" 1.9
1.7"
1.5 ~
1.3"
1.7
1.5'
1.3'
......Siiiii:
1.1' `*`* ,ms~--''41r ." . • 1.1'
0.9" 0.9'
0.7" 0.7'
~= 0.5" 0.5'
0.3 0.3
2'o 3'0 ,.0 5.0 50 10 2'0 3'0 ,'o 5'0 60
Air flaw rate (I/see) Air flow rate (lies©)

LOCATION 3 LOCATION 4
2.3" 2.3'
2.1' 2.1'
1.9- 1.9"
1.7" o,JB 1.7"
1.5" 1.5' ,AI
1.3 1.3 • `*"
1.1' 1.1 • Jb ...... 4'`* . ~',,~
o.9~ = ~ . ~ : ~ ...... .~--- 0.9" ,- ,.4" ,.-
0.7 ~
o.5~
i 0.7-
0.5'
~;_-;;;O-= ===-=It''"

0.3' 0.3
10 2'o 3'0 ,'0 5'0 50 10 2'0 3'0 4; 5.0 50
Air flaw rate (I/no) Air flew rate (line)

LOCATION 5 LOCATION 6
2.3 • 2.3
2.1 2.1
i 1.9 1.9
1.7 .~ 1.7
,,,41
1.5 .,`* 1.5
s S
`*s
,- .,:J
a 1.1 " - 1.1 ~¢4l "~'- ~ `*
J~ 0.9 _. "" "*"•" "''''" "4"'*. .`*. .s `**-- - ~1) 0.9
; 07 |=~.-.-$::::==:lr 0.7 ..w,,.~.'_-=_-==,~ .....
• 0.5 0:5
0.3 0.3
10 2'0 3.0 ,.0 5.0 ,o 10 2'0 3.0 ,.0 5'0 so
Air flaw rate (I/see) Air flaw rate (I/aae)

Fig.4a Variation of mean bubble size with air flow rate and impeller speed
at each of the six measuring locations. Pipsa impeller
[] 105 rpm; 635 120 rpm; .- • 140 rpm; 0 160 rpm
 Bubble size distribution 625

LOCATION 1 LOCATION 2
2.3" 2.3 ,
2.1" 2.1 -I
1.9: 1.9"t
1.7 17-(
"3 1.5;
_e 1.3 ' ...... ==:-_'_'_ 1.3
1'51 ...---;:?-.:: ....
-''-'-'-"
1.1 1.1 .*"" ,.4."" ,,"""
$
0.9 0.9 [_:..-: .-.-r'"
o 0.7- 0.7
:i 0.5- 0.5 -I
0.3- 0.3 I
0 2'0 3'0 ,'0 5'0 60 10 ,
20 3.O ,
40 5'0 80
Air flow rata (llsec) Air flow rate (llao¢)

LOCATION 3 LOCATION 4
2.3'
2"31
2.1 2.1'
1.9 1.9'
1.7 1.7'
! 1.5 1.5'
1.3 . . ~ r . . . . . . -41 1.3;
1.1 ...4r - - ~ " _..-.e 1.1'
0.9 w" ~.~'_..-e"" .." 0.9'
•= 0.7 I¢: .... * ...... -r 0.7'
0.5 0.5' i~_:--=::~:: . . . . . , . - - -
0.3 0.3
1o 2'0 3'0 4'0 5'0 50 10 2'o 3'0 ,'0 5.O 60
Air flow rate (llaec) Air flow rate (lilac)

LOCATION S LOCATION 8

i
23I
2.1
1.9
2.8
2.1
1.9
1.7
--ao 1.7
-- 1.5 1.5'
4 1.3'
• 1.3 "1 .....B
1.1'
.~. 0.9
1.1 1 .... 4r . ~ - ~ . . - - * ' = = - - . ._'.:.l- 4 ) 0.9'
0.7 iF,,, ' ' ' ' l r ....e'" 0.7' Ig"====¢=--:=---t:2 ....
0.5 e- . . . . . .o. . . . . 0.5' I1:: . . . . . . . . . . .
0.3 0.3
10 2'0 3'0 ,'0 5'0 60 10 2'o 3'o ,'0 5'0 50
Air flow rate (llloc) Air flow rate (llaoc)

Fig.4b Variation of mean bubble size with air flow rate and impeller speed
at each of the six measuring locations. Chile-X impeller
[3 .... 140 rpm; 635 160 rpm; • 180 rpm; 0 200 rpm

In the case of the Pipsa impeller (Figure 4a) the mean bubble sizes increased very rapidly with increasing
air flow rate at the lowest impeller speed (105 rpm). At each location, the mean bubble size was roughly
doubled in going from the lowest to the highest air flow rate. The biggest increase was in going from 42.5
l/sec to the'highest air flow rate of 56.7 1/sec: at this latter condition, the cell contents were visually
observed to be floc~ling with a boiling appearance near to the impeller, and the measured bubble sizes at
locations 1 and 2 were around 2 ram. At the other four locations, the values were between 1.4 mm and 1.8
mm. At the recommended impeller speed of 120 rpm, the increase in mean bubble size with increasing air
flow rate was not as marked, except at location 2. At the other locations, the increase in mean bubble size
ME $:6-0
626 B.K. GORAIN et al.

with air flow rate was nearly linear, with bubble size increasing by around 75% over the range of air flow
rates investigated. Except for locations 1 and 2, mean bubble diameters were between 1.1 nun and 1.4 m m
at the highest air flow rate. The results at 140 rpm were very similar to those at 120 rpm, with a slight
decrease in the mean bubble size at location 2, to around 1.8 mm. There was a notable difference at 160
rpm, however, in that the increase in mean bubble size with increasing air flow rate was much more
contained, at around 50% on average. The sharp increase in going from 42.5 l/sec to 56.7 1/sec was only
observed at locations 1 and 2: the maximum bubble size recorded at location 2 was 1.3 mm. A w a y from
the impeller, mean bubble size varied from 0.8 m m to 1.0 mm.

LOCATION 1 LOCATION 2
2.3 2.3 ,
2.1 2.1

i 1.7
1.0 1.7
1.9
1.5 1.5
1.3 F~q.~ 1.3
~..-e" ~ _ ..-.11 1.1 ..ll,_.-.-m ' ' l l
1.1 =___ _ _ . ~ . . : ":~.-.-e- " " ....~, o.s I ~_-... .-. . . : : 4: ". .. .-. .-. ~
. a. ~. -. . . . - ' ' e o
0.9 ..-' , - .~.~..---.41'"
.= o.7 |... e o.7 ~==::7; .....
=I 0.5 0.5
0.3 , 0.3 ,
10 2'0 3`0 ,0 5`0 ~0 10 2`0 ~0 ,'0 5'0
Air flow rote (I/no) Air flow rate (lleo¢)

LOCATION 3 LOCATION 4
2.3 2.3 ,
.i
2.1 "~
Z 1.91
2"1 1.9-I
= 1.7 1.7-1
"E 1.5 1.5"~
1"3
1.1t
i_e 1.1
1,31
o, . ..... °---_~z-=-':::~---" --1
s 0.7 .... "!"_;-.';---"'~ . . . . 0.7
0"91
• 0.5 I'I; . . . . 0.5
0.3 11.3
1o 2'0 3'0 4'0 5'0 60 10 2`0 3'0 ,'0 s'0 50
Air flow rate (llsoo) Air flow rote (llaec)

LOCATION 5 LOCATION 6
2.3" 2.3'
2.1" 2.1'
E~ 1.9. 1.9'
1.7" 1.7'
_a
t 1.5: 1.5'
1.3: 1.3'
.o
.n 1.1 1.1"
$ o., 0.0'
c o17: 0.7'
•I 0.S ~==,==;~ -" 0.S" u,d=::~.-'=" .... -w . . . . .
0.3 0.3
10 2~ 3`0 ,`0 5`0 .o 10 2'o 3'0 ;0 5~ .0 . . . .

Air flow rate (llseo) Air flow rate (I/ooo)

L :

Fig.4c Variation of the mean bubble size with air flow rate and impeller speed
at each of the six measuring locations. Dorr-Oliver impeller ~~ :
I"1 235 rpm; 635 255 rpm; - 275 rpm; 0 295 rpm
 Bubble size distribution 627

LOCATION 1 LOCATION 2
2.3 • 2.3'
2,1" 2.1'
1.9'
s 1.9"
1,7' 1.7'
_m
II 1.5' 1.5"
_e 1,3' 1.3"
,O
,O 1.1"
• . . . . . . . . . . :::$
.= 1.1'
0.9' t----==4, .,~ ~,O" 0.9'
0.7' 0.7'
0,5' 0.5'
0,3" 0.3
10 2'0 3'0 ,'0 ,'o 6o 10 20 3'0 ,'o ,'0 60
Air flow rate (flee©) Air flow rate (lleec)

LOCATION 3 LOCATION 4
2.3"
2.1"
E
.EE 1.9"
1.9,-I
1.7" 1.7 -I

e
l~ 1.5" 1.5 "t
1.3 "1
e 1.3"
a
a~ 1.1. ~4
...-.=...e 11..i
•~ 0.9. 0.9"!
' - =: . . . .
~, 0.7. ......
0.5, Ill;--- ,ll, "''
0.3.0 ,
20
•
30 4'0 50 60
0.3 I
10 2'0 3.0 4.0 5.0 60
Air flow rate (llaec) Air flow rate (lleec)

LOCATION 5 LOCATION 6
2.3 2.3

i 2.1
1.9 t 2,1
1.9 1
v 1.7 "] 1,7
1,5 1,5

1
1,3 1,3
1.1 .,41 1.1
~' 0.9
...... ..,,-".-'". •-::::.= ..... . . :__.-_-'_:- -- .";
t¢ 0,7 •--.-~. ;AF-~;£,-r'ff ~lF'" 0,7
0"91
" 0.5 ,:'----,"'" o.5 Ir===-'~ ...... -~--"
0.3 i 0.3 , i
t0 2'0 30 4'0 5'0 80 10 20 3'0 4'0 50 60
Air flow rata (I/lee) Air flow rata (I/leo)

Fig.4d Variation of the mean bubble size with air flow rate and impeller speed
at each of the six measuring locations. Outokumpu impeller
[1 185 rpm; 635 205 rpm; • 225 rpm; 0 245 rpm

With the Chile-X impeller (Figure 4b), which is similar to the Pipsa but does not have the additional
pumping mechanism, the effect of air flow rate on bubble size was not as marked as with the Pipsa
impeller. At the lowest impeller speed employed (140 rpm), mean bubble size at each location increased
by about 60%--70% on going from the lowest to the highest air flow rate. There was no big increase at the
highest air flow late of 56.7 1/see: the mean bubble size at locations 1 and 2 was around 1.5 nun, and
between L1 'and L4 mm at the other locations. These values are considerably smaller than those obtained
with the Pipsa impeller at the lowest impeller speed. On increasing the impeller speed to the recommended
level of 160 rpm, the mean bubble diameters away from the impeller varied from 0.8 mm to 1.1 turn at the
628 B.K. GORAINet al.

highest air flow rate. When the impeller speed was increased further to a maximum of 200 rpm, the
influence of air flow rate on the mean bubble size only became apparent when the air flow rate was set at
42.5 l/sec and over. At 56.7 l/see the maximum bubble size recorded at location 2 was 1.2 mm, while away
from the impeller the mean bubble size varied from 0.7 mm to 1.0 mm (very similar to values obtained with
the Pipsa).

The Dorr-Oliver (Figure 4c) impeller produced small bubbles at all impeller speeds, so the trend of
increasing bubble size with increasing air flow rate, though present, is not as marked. At the lowest impeller
speed of 235 rpm for the Dorr-Oliver impeller, mean bubble size increased by between 6% and 62% going
from the lowest to the highest air flow rate, depending on location. The biggest relative increases were in
locations 1 and 5, a trend repeated at all the impeller speeds. However, the absolute differences in bubble
diameter were small, as the mean bubble size was generally less than 1.0 mm away from the impeller.

The case of the Outokumpu impeller (Figure 4d) is interesting, as the lines in the Figures cross each other
at various places, indicating that the effect of impeller speed on bubble size is more complicated than is the
case with the Pipsa and Chile-X impellers. This may be on account of the test cell having a flat bottom
instead of a curved one as in the conventional Outokumpu cell, resulting in different flow patterns and
effects. Contrary to the other impellers tested, the influence of air flow rate on mean bubble size increased
with increasing impeller speed, with values at some locations more than doubling on going from the lowest
to the highest air rate at the maximum impeller speed, whereas the average increase at the lowest impeller
speed (185 rpm) was only around 20%. Nonetheless, the bubble sizes produced by the Outokumpu impeller
under all conditions were small: around a third of the values in Table 4d are 0.6 mm or less.

Figures 5a and 5b illustrate the effect of aeration rate on bubble size d i s t r i b u t i o n . The values are for the
Outokumpu impeller at location 6, at a speed of 225 rpm. At 16.5 l/see air flow rate, the bubble size
distribution was narrow, with a mean of 0.5 mm and a standard deviation of 0.16 mm, demonstrating that
the impeller had sufficient energy available to disperse all the air into small

Ood~, 0.30

0.30.
g M
£0.20

~0.20-
L t
i
q4 ~0.L0.
a a
O.lO.

0.% o'.6 t'.a I'.9

o.oo.
0 i0
Bubble Oianeter ( ~ ) ~able aiNta¢ (aa~

(a) (b)
Fig.5 Effect of air flow rate on bubble size distribution
(Outokumpu impeller, speed=- 225 rpm, location 6)
a: Low aeration rate: 16.5 1/see; b: High aeration rate: 56.7 l/see
 Bubble size distribution 629

bubbles. At 56.7 b'sec air flow rate, a much wider distribution of bubble sizes was found, with a mean of
0.7 mm and a standard deviation of 0.33 ram, suggesting that'the energy input was no longer sufficient for
uniform air dispersion. Thus the increase in the air flow rate not only increased the mean bubble size but
significantly altered the shape of the distribution, producing a much wider range of bubble sizes at the
higher aeration rate. Similar results were found for the other impellers. The effect of this on metallurgical
performance is the subject of continuing investigation by the authors.

It should be noted that the narrow bubble size distribution obtained at the low air flow rate might be the
result of the technique employed in this test work to measure the bubbles. The U.C.T bubble size analyser
captures bubbles in a glass capillary, and there is a physical limitation to how small a capillary can be
drawn. The technique relies upon the bubbles being deformed into cylinders in the capillary, where two
light-emitting diodes (LEDs) note the difference in refractive index between water and air as each bubble
passes, and convel~ the time intervals between pulses into bubble diameters [5]. Bubbles smaller than the
size of the capillary would typically not be deformed into cylinders, and would be absent from the results.
Thus there is a possibility that part of the left hand side of the distributions in Figure 5 might be missing.
However, the results of a recently completed investigation [9] showed that the bubble size distribution
obtained were very similar, and that the calculated mean diameters were generally within 10% of each other.
Thus the above re:mlts may be taken to be accurate.

Effect o f impeller speed

The effect of impeller speed on the mean bubble size generated by each of the four impellers at various
locations in the cell has already been noted above in the discussion of the effect of air flow rate. In general,
the mean bubble size was found to decrease with increase in impeller speed for all the impellers. This effect
was greatest for the Pipsa impeller (Figure 4a), which at the highest air flow rate was only able to disperse
the air uniformly into small bubbles throughout the cell if the impeller speed was increased to the maximum
level investigated. With the Chile-X impeller (Figure 4b), increasing the impeller speed from 140 rpm to
the recommended 160 rpm resulted in significantly smaller bubbles at locations away from the impeller (i.e.
excluding locations 1 and 2); further increases in impeller speed resulted in only marginal reductions in
bubble size.

The Dorr-Oliver and Outokumpu impellers produced very small bubbles even at the lowest impeller speeds
tested. For both of these impellers the decrease in mean bubble size with increase in impeller speed was
fairly uniform at all the aeration rates with anomalies at the high impeller speeds. The Outokumpu impeller
in particular show.,; a distinct reverse trend at the highest impeller speed investigated, i.e. the mean bubble
size increased when the impeller speed was increased from 225 rpm to 245 rpm, at all locations in the cell,
especially at the higher air flow rates. This may be attributed to the fact that the internal gas recirculation
increases dramatically at very high impeller speed, increasing the actual gas input rate handled by the
impeller, as evidenced by the sharp rise in air hold-up at this point [10]. The size of bubbles produced in
this situation is a strong function of aeration rate rather than impeller speed [8].

Effect o f location

It is quite evident from Figures 4a to 4,1 that the mean bubble size varies with location in the flotation cell.
Larger bubbles were found near to the impeller shaft (locations 1 and 2) while the smallest bubbles were
found near to the impeller discharge point at location 4. The mean bubble size seemed to increase as the
bubbles moved along the flow profile of the cell as illustrated in Figure 3b, probably as a result of
coalescence. The trend varied from impeller to impeller, and is considered in the discussion of the effect
of impeller type, t~elow.

Bubble size distributions were also found to vary with location in the cell. Typical bubble size distributions
at different locations in the cell are shown in Figure 6. The values were obtained with the Chile--X impeller,
at a constant air flow rate of 28.3 1/sec and impeller speed of 140 rpm. It can be seen that the bubble size
distribution at the impeller tip (location 4) is skewed towards the small bubble size classes, forming an
approximate gamvaa distribution. With increase in distance from the impeller around the circulation loop,
630 B.K. GORAINet al.

the distribution shifts to larger bubble classes and the mean bubble size correspondingly increases. This
suggests the slurry was following the flow pattern in the cell as indicated in Figure 3b.

Bubble Dimeter O/str/Imtim Bubble Oimeter Distribution

0.30 0,30

0,20

qq i
u
~.10
11
l

0.00
o'.6 1'.3 ,5 o~6 ,~3 1:9 2:s
Bubble OSmseter (m) Bubble Oiamter (m)

EUbble Dimeter Ofstr/butt~ Bubble Oimeter Olstrilmt/m

0.30 0.~,

~ O.ZO |o=o4

A
L
u

0.00
0 o'.6 ,'.3 ,'.9 2:, °'%!o o'.6 1'.~ 1'.9 2~s
Bubble Oianeter (nn) Bubble Diaseter (m)

BUbble O l m e t e r O i s t r i b u t i o n Bubble DiaMter Oistrfbutlm

0.30. 0.3).

t.

0.W
o'.6 1:=
EUbble Oimseter (m)
l'.g 2'.s o:o
jL 0:6 1'.~ 1'.9
Bubble DfaJmter (m)
2'.s

Fig.6 Bubble size distribution at six different locations for the Chile-X impeller
(speed=140 rpm, air rate=28.3 l/sec)
 Bubble size distribution 631

At locations 1 and 2, the bubble size distributions attain a normal distribution. Thus the bubble size
distributions change from a gamma-type distribution at locations 4 and 6 to a normal distribution to
locations 1 and 2, i.e. both the shape of the distribution and the mean bubble diameter vary with location
in the cell. Similar observations were made by Kawecki [11], Greaves and Kobbacy [12] and Cochrane [8],
and suggest that coalescence takes place along the streamlines. Flow patterns in the cell were not
discernable from this work, but a glass cell of exactly the same dimensions as the test flotation cell has been
constructed at the JIKMRC in Brisbane and will be used to investigate flow patterns produced by the
different impellers at the same combinations of air flow rate and impeller speed.

Effect of impeller ~pe

Figures 7a to 7d show the variation of mean bubble size with air flow rate and location in the cell for each
of the four impellers tested, at the manufacturer's recommended impeller speeds. There is a general trend
of increasing bubble size along the "location" axis. The locations are sequenced 4 - 6 - 5 - 3 - 1 - 2 ,
corresponding to the circulation loop in the cell illustrated in Figure 3b. Location 4 corresponds to the
impeller tip, where bubbles would be discharged [13] while location 2 is just above the impeller, close to
the shaft, at the point where slurry is drawn into the impeller.

Plpse Impeller Dorr-Oliver Impeller

2.3 • 2.9 '

2.1' 2.1"
I 1.9" 1.9"
SS
S
S 1.7"
Ne 1.7"
"g 1.5" 1.5"
_~.4P " ~ 4 T ~'4
_o 1.3" 1.3-
0, ,~lr ~" ~.IT ~
1.1' ,~ ,,,,,~ • ~41 ,~
~41.~ ~
. ~
o 1.1-
.Z o.o o.e-.
;" "---- .i'----- 41" o.7~
•~" 0.7 "i
• 0.5. 0.5"~ !
0.3 l I I I 0.3 i i , , /
4 6 5 3 1 4 6 5 3 1 2
location Location

Outokumpu Impeller Chile-X Impeller

2.3 2.3 m

2.1 -I
2.1 t
s 1.9 1.9 -I

'1::]
1.7-1
"g 1.5"1
_e
Jm
=1
1.1 ;1 ..'...,,m'-,~-,./ .
Jm
g=
o.o] .... ==~'.,,." ,,.:'4 o.oJ __~-:.-.t -° _..,
o.71..-:--.:::--•'--..,- ....- , o.7t-.::'_.,-__=.-..,::-:,:::-'--
0.5 0.3 ~
0.3 i i i
4 6 S 3 1 2 4 6 5 3 1
Location location

Fig.7 Variation of mean bubble size with air flow rate and location for the four impellers
at the manufacturer's recommended impeller speed (Pipsa 120 rpm, Chile-X 160 rpm,
Dorr-Oliver 255 rpm, Outokumpu 205 rpm).
13---- 16.5 l/see; 635 28.3 l/sec; • 42.5 l/sec; ¢ 56.7 1/sec

The Figures show clearly that the mean bubble size varies with location in a set pattern for the different
impellers. For the Pipsa impeller, there is a clear and steady increase in mean bubble size along the given
sequence of locations (around the circulation loop), with the values increasing for increasing air flow rates.
632 B.K. GORAIN et al.

At the highest air flow rate, the Pipsa impeller operated at the manufacturer's recommended rpm was clearly
unable to produce small bubbles under the other conditions prevailing in the cell. The same trend of
increasing bubble size around the circulation loop is discernable for the Chile-X impeller, but the mean
bubble sizes were generally smaller.

The Dorr-Oliver and Outokumpu impellers produced results that are similar to each other but somewhat
different from those obtained with the Pipsa and Chile-X. The trend of steady increase in mean bubble size
around the circulation loop is not as evident, with a sharp decrease in mean bubble size in going from
location 1 to location 2 for both Dorr-Oliver and Outokumpu impellers (in six of the eight experiments).
There is also a sharper increase in going from location 3 to location 1 and a slight but quite evident
decrease from location 6 to location 5, for the Dorr-Oliver impeller. This suggests the presence of a
secondary flow pattern within the outer circulation loop or of a different circulation loop completely. Thus
the differences in bubble size distribution observed between the impellers may be partially explained by the
relationship of the location of the sample point s to the turbulence or flow patterns for the individual
impellers. As mentioned above, a detailed investigation of these flow patterns will be carried out using a
2.8 m 3 glass flotation cell at the JKMRC.

"Global mean" bubble size

While it is apparent from the results presented above that the mean bubble size varies considerably with
location in the flotation cell, it is of interest to calculate a "global mean" bubble size to represent the
"overall" bubble size distribution in the cell. Because each bubble size distribution measured was based on
a different number of bubbles, and the distribution of air around the cell is not known, it is easiest to
calculate a "global mean" bubble size for each operating condition as
i--6

i=1
ai (1)
6

w h e r e ' d ' is the mean bubble size at a particular location, and 'i' is the location number.

"Global mean" bubble sizes calculated in this way are given in Tables 4a to 4d above, for each combination
of impeller type, speed and air flow rate investigated. Interestingly, each of these values may be seen to be
remarkably close t~ the corresponding value of mean bubble size measured at location 3. This holds good
for all the impellers, indicating that (for this test work at least) it was only necessary to measure the mean
bubble size at location 3 to obtain the arithmetic mean of the bubble sizes at locations 1 to 6. Whether this
finding is universal, or changes under different flotation conditions, still needs to be confirmed. It would
be very useful if a global "mean" bubble size could be determined from a measurement at only one point
in the cell. However, at this point, no physical significance can be attached to this value.

Table 5 shows the "global mean" bubble sizes calculated using equation (1) for each of the four impellers
at the recommended impeller speeds, at the four air flow rates investigated in this work. The Table shows
that at all the air flow rates, the smallest mean bubble sizes were produced by the Outokumpu impeller, and
the largest by the Pipsa impeller, with the Dorr-Oliver and Chile-X impellers falling between these two.
Except for three values in the Table, all the global bubble sizes produced at the manufacturers'
recommended speeds were 1.0 mm or smaller.

Another, and possibly more appropriate, basis for comparing the relative performance of the impellers in
bubble formation is the power input at the particular operating conditions of air flow rate and impeller
speed. The power drawn by the impeller in each of the tests carried out in this programme isigiven in the
last column in the Tables of results (Tables 4a to 4d). As may be seen, at each impeller speed, the power
draw decreased steadily with increase in air flow rate, for each impeller. The power input at the
manufacturer's recommended speed for each of the four impellers is given in TableS, for each of the air
flow rates examined. The Outokumpu impeller, which produced the smallest global mean bubble sizes,
 Bubble size distribution 633

also drew more power than any of the other impellers3; the Don--Oliver impeller, which was second to the
Outokumpu in dispersing air at the higher flow rates, drew the least power of all four impellers. To produce
a global mean bubble size of 1.0 mm at the maximum air flow rate of 56.7 1/see, the Pipsa impeller had
to be rotated at 160 :rpm, requiring a power input of 3.64 kW. To do the same, the Chile-X needed 180 rpm
and 3.51 kW, the Outokumpu 185 rpm and 2.64 kW, and the Dorr-Oliver 255 rpm and only 1.78 kW.

T A B L E 5 Global mean bubble size (from equation (1))

and p o w e r input at the m a n u f a c t u r e r ' s r e c o m m e n d e d speed.
Bubble size in nun and power input in kW

Impell .~r t y p e

ALrflow Pipsa Chile,X Dorr-Oliver Outokumpu

rate(M..~

Bubble lower Bubble Power Bubble Power Bubble Power

size size size size

16.5 0.79 2.95 0.70 3.20 0.65 2.75 0.63 5.47

28.3 0.91 2.77 0.75 3.08 0.82 2.42 0.69 4.34

47.5 1.09 2.70 0.99 2.95 0.56 21'7 n.~ 3.57

56.7 1.~ 2.55 1.07 2.82 1.00 1.78 0.89 _'a_TR

CONCLUSIONS

The bubble size me~Lsurements reported in this study are not claimed to be universal for the impellers tested
or the operating conditions employed. It is acknowledged [14] that the bubble size in a flotation cell will
vary considerably with other factors, both chemical (especially frother dosage and water quality) and
physical (such as pulp viscosity, solids concentration and particle size distribution of feed). However, a
number of important conclusions may be derived, as follows:

a portable industrial scale sub-aeration flotation cell, with provision to fit in different impellers and
stators, has been constructed with the aim of conducting on-site metallurgical or other testwork using run-of-
plant ores. This test cell with its associated instrumentation offers a very powerful and convenient research
tool with which to study flotation conditions in a full-scale cell at different design (e.g. impeller type) and
operating condition,,;.

- - bubble size mea:mrements have been carried out in the test cell, located at the head of a zinc cleaner
circuit, using four typical impellers over a range of air flow rates and impeller speeds. The mean bubble
size was found to increase with increase in air flow rate at various locations in the cell, for all four
impellers, although the manner in which the increase took place depended on the impeller type and speed.
At constant impeller speed, increase in air flow rate increased not only the mean of the bubble size
distribution, but the spread as well.

3as noted elsewhere in this paper, the test cell has a fiat bottominstead of a round bottomlike the conventional Outokumpucell. The
different hydrodynamicsmight necessitate a greater power draw than would normallybe required.
634 B.K. GORAINet al.

- - the mean bubble size decreased with increase in impeller speed for all the impellers, but the effect was
greatest for the Pipsa followed by the Chile-X impeller. The effect was not as marked for the Dorr--Oliver
and Outokumpu impellers under the conditions tested, as the air was already well dispersed at the lower
impeller speeds. The trend was reversed with the Outokumpu impeller at the highest impeller speeds, i.e.
the bubble size increased when the impeller speed was increased. This was attributed to changing flow
patterns in the cell at the higher impeller speed.

- -bubble sizes varied with location in the flotation cell, being largest closest to the shaft above the impeller
and smallest at the impeller discharge point, for all the impellers. The shape of the distribution also changed
from a gamma distribution at the impeller discharge point to a normal distribution at points above the
impeller. Flow patterns postulated in the cell from the change in bubble size with location for each impeller
will be verified using a glass cell of the same dimensions operated at the same combinations of impeller
type, speed and air flow rate.

- - the global mean bubble size calculated by simple arithmetic average of the values obtained at six
locations in the cell was found to coincide remarkably with the mean bubble size measured half-way
between the impeller and the side of the cell, at the top of the pulp. With the exception of the Pipsa impeller
at the higher air flow rates, and the Chile-X impeller at the highest air flow rate tested, all the impellers
produced global mean bubble sizes of 1.0 mm or less at the manufacturers' recommended speeds.

The size of the bubbles produced by the impeller is not the only criterion of air dispersion in flotation cells.
Equally important is the air hold-up or volume fraction of air in the slurry, which determines how much
air is available for flotation, and which together with the bubble size distribution determines how much
bubble surface area is available for the particles being floated. Of even greater importance is the effect of
these values on the metallurgical performance obtained. These issues will be addressed in detail by the same
authors in subsequent publications.

ACKNOWLEDGEMENTS

The authors would like to thank Geoff Richmond, John Glen and other staff of Aberfoyle Resources Limited
for their generous help throughout the period of the test work. The financial support of the Australian
Mineral Industries Research Association and the Australian Research Council is also gratefully
acknowledged.

R E F E R E N C E S

. Nevett, R.D., Some controlling factors in flotation, Proc. Aust. Inst. Min. Metall., New Series; No
37, 55-72 (31 March 1920).
2. Bennet, A.J.R., Chapman, W.R. & Dell, C.C., Froth flotation of coal, 3rd Int. Coal Prep. Conf.,
Brussels-Liege, E2. 452-462 (1958).
3. Diaz-Penafiel, P. & Dobby, G.S., Kinetic studies in flotation columns: bubble size effect, Minerals
Engineering, 7(4), 465-478 (1994).
4. Dobby, G.S., Yianatos, J.B. & Finch, J.A., Estimation of bubble diameter in flotation columns from
drift flux analysis, Canadian Metallurgical Quarterly, 27(2), 85"90 (1988).
5. Randall, E.W., Goodall, C.M., Fairlamb, P.M., Dold, P.L. & O'Connor, C,T., A method of
measuring the sizes of bubbles in two and three phase systems, J. Phys. E: Sci. Instrum., 22,
827-833 (1989).
. Tucker, J.P., Deglon, D.A., Franzidis, J.-P., Harris, M.C. & O'Connor, C.T., An evaluation of a
direct method of bubble size distribution measurement in a laboratory batch flotation cell, Minerals
Engineering., 7(5/6), 667-680 (1994).
 Bubble size distribution 635

. Richmond, G.D., Zinc-lead ore concentration by Abeffoyle Limited, Hellyer, Tasmania

Australasian Instituteof Mining and Metallurgy, ed J.T Woodcock and J.K Hamilton, The Sir
Maurice Mc~wby Memorial Volume, 2nd edition, I, 530-535, Aus I.M.M Monograph No 19,
(1993).
. Cochrane, D.R., Hydrodynamics in aerated stirred tank, Ph.D Thesis, Department of Chemical
Engineering, University of Queensland, (1986).
9. Vanderkruk, C., Comparison of bubble sizing techniques in a laboratory scale flotation cell,
B.Sc.(Hons) Thesis, Department of Chemical Engineering, University of Queensland, (1994).
10. Gorain, B.K., Franzidis, J.-P. & Manlapig, E,V., Studies on impeller type, impeller speed, and air
flow rate in an industrial scale flotation cell. Part 2: Effect on gas hold-up and superficial gas
velocity, submitted to Minerals Engineering, 1994.
11. Kawecki, W., Reith, T., Van Heuven, J.W. & Beek, W.J., Bubble distribution in the impeller
region of a stirred vessel, Chem. Eng. Sci., 22, 1519-1522 (1967).
12. Greaves, M. & Kobaccy, K.A.H., Measurement of bubble size distributions in turbulent gas-liquid
dispersions, Chem. Eng. Res. Des., 62(3), 3-12 (1984).
13. Bruijn, W., Van't Riet, K. & Smith, J.M., Power consumption with aerated Rushton turbines,
Trans. Inst. Chem. Engrs., 52, 88-104 (1974).
14. O'Connor, C.T., Randall, E.W. & Goodall, C.M. Measurement of the effects of physical and
chemical variables on bubble size, Int. Jour. Min. Proc., 28, 139-149 (1990).