Coarse particle concentration using the

HydroFloat Separator
J.N. Kohmuench and G.H. Luttrell
Department of Mining and Minerals Engineering, Virginia Polytechnic Institute and
State University, Blacksburg, Virginia

M.J. Mankosa
Manager technical development, Eriez Magnetics, Erie, Pennsylvania

Abstract
A new separator, known as the HydroFloat cell, was del'eloped to overcome the limitations oj'traditional
flotation cells in recovering coarse particles. The HydroFloat cell separates particles based 011 differences
in mass after the selective attachment of air hubbies to the hydrophobic component ofthefeed streal/l. proor
oj-concept testing, utili::.ing phosphate matrix, hem'Y mineral sands and anthracite slag, indicated that
sign(fkant improvements in separation efficiency cal/ be achiel'ed using this air-assisted teeter-hed s\'Stem.
Based on the encouraging laboratOlY results, a pilot-scale HydroFlo(/f cell wasfclbricated and installed at
a central Florida phosphate plant. Test data collected using the pilot-scale unit suggest that the BPL reC(}\'ery
qfthe +35-mesh plantfeed can be increased by more than4()CIc through the application oj'thi.1 nell' technolog\'.
This improvement is particularly beneficial due to the I'cry high grade (-7.FIc BPL) oj'the + Jo-mesh !}(lrticle.1
in the HvdroFloat concentrate. This article describes the theoretiml basis for the dl'l'e/opment oj' the
HydroFl(Jat cell and provides an overview of recent laboratory and pilot-scale test results.
Key words: Flotation, Course particle concentration, HydroFloat separator

Introduction the case of the HydroFloat separator. the teeter bed is continu-
Hindered-bed separators are commonly used in the minerals ously aerated by injecting compressed air and a small amount
industry for particle classification. These units can also be of frothing agent into the fluidization water. The gas is
employed for mineral concentration if the particle size range dispersed into small air bubbles by circulating the water
and density differences are within acceptable limits. How- through a high-shear mixer in a closed-loop configuration
ever, these separators often suffer from the misplacement of with a centrifugal pump. The air bubbles become attached to
low-density coarse particles to the high-density undertlow. the hydrophobic particles within the teeter bed. thereby.
This shortcoming is due to the accumulation of coarse low- reducing their effective density. The particles may be natu-
density particles at the top of the teeter bed. These particles are rally hydrophobic or made hydrophobic through the addition
too light to penetrate the teeter bed, but are too heavy to be of flotation collectors. The lighter bubble-particle aggregates
carried by the rising water into the overflow launder. These rise to the top of the denser teeter bed and overflow the top of
particles are eventually forced by mass action downward to the separation chamber. Unlike flotation. the bubble-particle
the discharge as more particles accumulate at the top of the agglomerates do not need to have sufficient buoyancy to rise
teeter bed. This inherent inefficiency can be partially cor- to the top of the cell. Instead. the teetering effect of the
rected by increasing the teeter water velocity to convey the hindered bed forces the low-density agglomerates to overflow
coarse, low-density solids to the overflow. Unfortunately, the into the product launder. Hydrophilic particles that do not
higher water rates will cause fine, high-density solids to be attach to the air bubbles continue to move down through the
misplaced to the overflow launder, thereby, reducing the teeter bed and eventually settle into the dewatering cone.
separation efficiency. These particles are discharged as a high solids stream (e.g ..
To overcome the shortcomings of traditional hindered-bed 75o/c solids) through a control valve at the bottom of the
separators, a novel device known as the HydroFloat separator separator. The valve is actuated in response to a control signal
was developed. As shown in Fig. I, the HydroFloat unit provided by a pressure transducer mounted to the side of the
consists of a rectangular tank subdivided into an upper sepa- separation chamber. This configuration allows a constant
ration chamber and a lower dewatering cone. The device effective density to be maintained within the teeter bed.
operates much like a traditional hindered-bed separator with Theoretically. the HydroFloat separator can be applied to
the feed settling against an upward current of fluidization any system where differences in apparent density can be
water. The fluidization (teeter) water is supplied through a created by the selective attachment of air bubbles. Although
network of pipes that extend across the bottom of the entire not a requirement. the preferred mode of operation would be
cross-sectional area of the separation chamber. However, in to make the low-density component hydrophobic. so that the

Preprint number 00-100, presented at the SME Annual Meeting, February 28-March 1, 2000, Salt Lake City, Utah.
Manuscript accepted for publication August 2000. Discussion of this peer-reviewed and approved paper is invited and must
be submitted to SME Publications Dept. prior to Nov. 30. 2001. Copyright 2001, Society for Mining, Metallurgy, and
Exploration, Inc.

MINERALS & METALLURGICAL PROCESSING 61 VOl. 18. NO.2. MAY 2001

 After a particle contacts a bubble, the particle is swept over
t?e bubble surface for a finite period known as the sliding
time. During this period, the thin liquid film separating the
bubble and particle must rupture if particle adhesion is to
occur. This "sliding time" is a reflection of the hydrodynamics
of the system and is primarily a function of the particle and
bubble sizes. On the contrary, the length of time required for
the liquid film to thin sufficiently so that rupture occurs is a
measure of the chemistry of the flotation system and is
commonly referred to as the induction time. The induction
e HlahOcftJlty time is small for hydrophobic particles (e.g., I msec) and may
® Low Dcnuty approach infinity for extremely hydrophilic particles.
o e"bb1c ..
Utilizing the induction time concept, Yoon and Luttrell
(1989) derived the following analytical expression for the
probability of bubble-particle adhesion (Pa)

(2)

where
I Dp is the particle diameter,
+
flo.1
RCJcct
Db is the bubble diameter,
Figure 1 - Schematic illustration of the HydroFloat ti is the induction time,
separator. Ub is the differential velocity between the bubble and
particle and
greatest difference in specific gravity would be achieved. B is a constant that varies depending on the particular flow
Compared to traditional froth flotation processes, the regime (as dictated by Reynolds number).
HydroFioat separator offers several important advantages
for treating coarser material. These include enhanced bubble- In most cases, Ub is simply assumed to be the terminal rise
particle contacting, better control of particle residence time, velocity of the bubble. Because Eq. (2) is expressed as a sine
lower axial mixing/cell turbulence, and reduced air con- function, the calculated value of P will always fall between
sumption. zero and unity, the correct limits for probabilities. Figure 2
shows a typical set of calculated P values as a function of
Process theory p~icle size for different levels of i;duction time (hydropho-
The reaction, or flotation, rate for a process is indicati ve of the ~IClt~). As expected, Pa increases sharply as the induction
speed at which the separation will proceed. In mineral flota- time IS reduced from 5 to 1 msec. However, for a given value
tion, the reaction rate is controlled by several probabilities, of t i , Pa decreases steadily as the particle size increases. The
e.g., collision, adhesion and detachment. The attachment of reduced Pa value is due to the fact that larger particles tend to
particles to air bubbles is the underlying principle upon which slid~ more rapidly over the bubble surface because they
all flotation processes are based. This phenomenon takes project further out into the high-velocity region of the stream-
place via bubble-particle collision followed by the selective lines that pass over the bubble surface. The differential veloc-
attachment of hydrophobic particles to the bubble surface. ity between bubbles and particles (Ub ) is greatly reduced by
Particles may also detach if the resultant bubble-particle the hindered settling/rise conditions within the teeter bed of
aggregate is thermodynamically unstable. According to the HydroFloat cell. As dictated by Eq. (2), the reduced
Sutherland (1948), the attachment process may be described velocit~ will incr~ase the probability of adhesion (Pa)' thereby,
by a series of mathematical probabilities given by enhanCing flotation recovery. As shown in Fig. 2, this phe-
nomenon is particularly important for coarse particles that
P=P/>JI-Pd ) (I) tend to suffer from low Pa values.
where Hindered-bed separators also operate as low-turbulence
Pc is the probability of collision, devices. Therefore, particle detachment is minimized due to a
Pa is the probability of adhesion and reduction in localized turbulence. Studies conducted by
Pd is the probability of detachment. Woodburn et al. (1971) suggest that
Pd =(DID/Y (3)
The attachment and detachment probabilities are con-
trolled by the process surface chemistry and cell hydrodynam- where
Dp is the particle diameter to be floated,
ics, respectively. In an open (free-settling) system, the colli-
sion probability is quite low due to the low particle concentra- D[/is the maximum floatable particle diameter and
tion. However, at higher concentrations, the crowding effect x IS an experimental constant (typically 312).
within the hindered bed increases the probability of collision.
This phenomenon is due to the compression of the fluid Factors that influence the magnitude of D • include pulp
streamlines around the bubbles as they rise through the teeter chemistry (surface tension and contact angle), physical par-
bed. The increased probability of collision can result in ticle properties (size, density, composition and shape) and cell
reaction rates that are several orders of magnitude higher than agitation intensity. Theoretical D • values have been calcu-
found in conventional flotation. lated by Schulze (1984) from thl tensile and shear stresses

MAY 2001 • VOL. 18 NO.2 62 MINERALS & METALLURGICAL PROCESSING

acting on bubble-particle aggregates under homogenous tur- 1.0 -r---------------,
bulence. The degree of turbulence was quantified in terms of
the induced root mean square velocity (RMSV). A typical set I~= 1 mml
c
of results obtained by Schulze (1984) is shown in Fig. 3 for .2 0.8
different contact-angle values. According to this theoretical (II

study, the maximum size of particles that may be recovered by G.I

s:. \
,

-
flotation increases by more than an order of magnitude when ~ 0.6
changing from high to low turbulence. According to Barbery
(1984), the optimum conditions for coarse particle flotation
occur when cell agitation intensity is reduced to a point just
sufficient to maintain the particles in suspension. Thus, a
o
~ 0.4
..c
,
\ ',
,,
1 msec

teeter bed is an ideal environment for minimizing particle 1"11 \ ..

,
detachment. ..c 2 msec
In froth flotation, the bubble-particle aggregate must have
o.... 0.2 \
sufficient buoyancy to be lifted to the surface of the pulp. 0.. \ 5 ms~-" ... ........
Mathematically, the maximum particle diameter (Dmax) that 0.0 -1-_....::........._ _---.,~--_+_--.-;;~
- . -- --
may be floated may be estimated from
0.0 0.5 1.0 1.5 2.0
Particle Diameter (mm)
(4)
Figure 2 - Effect of particle size on the probability of
where bubble-particle adhesion for induction time values of 1, 2
and 5 msec.
Pp and Pi are the densities of the particle and fluid,
respectively. 10 ~ ____________________________I

This expression suggests that I-mm-diam bubbles are Low Turbulence

- - ... - -
_ ... -- - .
capable of carrying particles up to approximately 0.85 mm (SIalic Condition) ._ - -
before the critical buoyancy limit is exceeded (r = 2.5 g/cm 3).
However, unlike traditional processes, the HydroFloat cell is
both a flotation device and a density separator. The use of a
-.5.
E
,. '
-.
~

Medium Turbulence
teeter bed makes it possible to achieve separations based on G.I
small differences between the density of free suspended .!:::! (RMSV = 0.2 m/sec)
UJ
particles and the density of bubble-particle aggregates. There-
~ ~
fore, separations can be achieved even if the buoyancy of the u
bubble-particle aggregate is too small to lift the particle load. :e1"11 0. 1
In other words, the density of the bubble-particle aggregate 0..
need only be smaller than the effecti ve density of the teeter bed ~
High Turbulence
to achieve a separation. This capability eliminates the buoy-
(RMSV = 1.0 mlsec)
ancy limitation described by Eq. (4). This feature is important
for very large particles that are difficult to carry to the top of 0.01
a conventional flotation pulp. o 20 40 60 80 100
The hindered-bed environment also influences particle Contact Angle (0)
retention time, and hence, particle recovery. The mixers-in-
series model provides a convenient framework for analyzing Figure 3 - Influence of turbulence on the maximum
this phenomenon (Arbiter and Harris, 1962; Bull, 1966). particle size that may be recovered by froth flotation (after
According to this model, the cumulative fractional recovery Schulze, 1984).
(R) of a given particle species can be determined using the
expression: lected using solid or liquid tracers. Details related to this
(5) procedure have been described elsewhere (Mankosa et aI.,
1992).
where The HydroFloat cell operates under nearly plug-flow con-
k is the flotation rate constant, ditions because of the low degree of axial mixing afforded by
rp is the particle residence time and the uniform distribution of particles across the teeter bed.
n is the number of equivalent mixers. Consequently, the cell operates as if it were comprised of a
large number of cells in series (i.e., high value ofn) . As shown
Figure 4 shows recovery determined from Eq. (5) for in Fig. 4, this characteristic allows a single unit to achieve the
different values of n as a function ofthe dimensionless product same recovery as a multi cell bank of conventional cells (all
kr . In most cases, n is assumed to be equal to the number of other conditions equal). In other words, the HydroFloat cell
cells in the flotation bank. This assumption is generally valid makes more effective use of the available cell volume than
for a cell-to-cell flotation bank. However, the magnitude of n well-mixed conventional cells or open columns.
is typically smaller for flow-through flotation banks that have In column flotation, particles settle vertically through the
a significant amount of intermixing. The appropriate value of cell either with the fluid flow (cocurrent) or opposite to it
n can be readily estimated for any cell configuration using (countercurrent). A countercurrent arrangement has obvious
residence time distribution (RTD) data that have been col- advantages because the settling velocity is reduced by the

MINERALS & METALLURGICAL PROCESSING 63 VOL. 18, NO.2. MAY 2001

 100 parameter can be mathematically described by
n=8
K=1/4PS h (9)
80 where
P is the probability of bubble particle attachment and
Sh is the superficial bubble surface area rate.
~ 60
~ The latter can be calculated directly using
GI
>
0
0
GI (10)
0::

20 where
QIi is the volumetric gas flow rate,
Dh is the bubble diameter and
0 Ac the cell cross-sectional area.
0 3 4
Equations (9) and (10) suggest that the same flotation rate
constant (k) can be maintained at a lower overall gas rate (QIi)'
if the attachment probability (P) increases accordingly. Fur-
Figure 4 - Relationship between flotation recovery and thermore, because the HydroFloat cell provides a longer
the dimensionless quantitytfor different numbers of mixers
in series. particle residence time (. ,), a lower gas rate (lower rate
constant) can usually be tolerated without sacrificing recov-
ery. In some cases, the aeration rate may be only 10% of that
upward flow of liquid resulting in a higher retention time.
employed by mechanical cells or columns.
Hindered settling, as previously explained, provides an envi-
ronment in which the particles never achieve their terminal
free-fall velocity. Therefore, the effective particle velocity
Proof-of-concept testing
through the cell is greatly reduced providing a significant Several laboratory-scale tests were conducted to evaluate the
increase in retention time as compared to a free-settling potential of the HydroFloat separator for upgrading several
types of mineral samples. These samples included mineral
system.
The particle residence time (. ) can be estimated from the sands, phosphate matrix and carbonaceous slag. Conven-
liquid residence time (.,) using the expression: tional flotation tests, whose results were used as baseline data
in some investigations, were conducted using a laboratory
(6) flotation cell (Denver Model 0-12). Aerated teeter-bed inves-
tigations were conducted with a laboratory-scale HydroFloat
where
V is the volume of each cell (corrected for gas holdup), cell operated in two different modes, i.e., batch and continu-
Q is the volumetric flow rate of slurry and ous. The test unit was fabricated from Plexiglas with an open
KS is a correction factor to account for particle settling area of approximately 0.1 m2 .
behavior.
Testing of phosphate matrix. A sample of run-of-mine
For well-mixed cells, K is approximately equal to one and phosphate matrix (5 mm x 65 mesh) was scalped at 3 mm to
may be ignored. For vertical-flow cells, such as columns, K remove the extreme oversize and debris material. The feed ore
may be estimated from was classified to remove the -35-mesh fines and conditioned
with a fatty acid-diesel fuel mixture at a dosage of approxi-
K=--='---
Qs mately 0.50 kg/t (active fatty acid). In all tests, ammonium
{cocurrent mode} (7) hydroxide was used for pH control and a polyglycol frother
V"A+Qs
was added to stabilize the bubble suspension. Batch HydroFloat
tests were conducted and compared against conventional
flotation results.
{countercurrent mode} (8) Figure 5 compares the results of the conventional flotation
tests with those obtained using the HydroFloat cell operated in
where the batch mode. In this series of tests, the HydroFloat cell
V is the particle settling velocity and achieved a BPL recovery of nearly 95% compared to less than
A ~s the cross-sectional area of the flotation cell. 79% for the conventional cell. This represents an increase in
BPL recovery of more than 20%. Although the conventional
In most flotation processes, feed particles move with the cell was floated to exhaustion, higher recoveries were not
fluid flow towards the discharge point (cocurrent mode). In possible because many of the coarser particles remained
contrast, particles move in the opposite direction to the fluid unfloatable. Furthermore, the recovery improvement was
flow within the HydroFloat cell (countercurrent mode). As achieved while maintaining a slighter higher concentrate
dictated by Eqs. (6) and (8), the fluidization water within a grade (67.0% vs. 65.7% BPL). The insol content of the
hindered settling regime provides a significant increase in the HydroFloat concentrate was also lower (7.2% vs. 7.7%). In
particle retention time. The longer retention time allows good fact, the test data show that the batch HydroFloat cell pro-
recoveries to be maintained without increasing cell volume. duced concentrates with a lower insol content over the entire
Finally, the rate constant (k) is the most important term in range ofBPL recoveries. The unexpected improvement in the
determining flotation performance. Studies indicate that this quality ofthe concentrate produced by the HydroFloat cell has

MAY 2001 • VOL. 18 NO.2 64 MINERALS & METALLURGICAL PROCESSING

 100
BPL

--
Recovery
(%)
80
~
to
~
60
Grade >
~
u
(% BPL) ~

=: 40
....;!

=
Q..

20 Ll Conventional
Insol (%) • ConventDnal o HydroFloat
o HydroFloat
o
o 20 40 60
0.5 1.0 1.5 2.0 2 .5
80 100
Concentration Ratio
Figure 5 - Comparison of separation data for batch test FigureS-Comparison of separation data from continuous
units for a phosphate matrix. test units for a phosphate matrix.

been attributed to the incremental recovery of very high-grade 100

coarse patticles that could not be floated by the conventional
flotation technique. ,-.
In light of the promising results obtained in the batch tests, 0~ 80
several follow-up experiments were conducted by operating
'-"
t' o 0
the HydroFloat cell in the continuous mode. Several different GI

o
~
combinations of reagent dosages, water rates and teeter bed Q
c; 60
levels were examined in this effort. Due to slight variations in =:"
the BPL head assay, it was necessary to compare the test =
,.,
Q
.&I 40
results based on a concentration ratio . This parameter is
OIl
mathematically defined as the ratio of the concentrate grade U
(BPL%) divided by the feed grade (BPL%). As shown in Fig. 'C
0With Air
">< 20
6, the continuous unit produced consistently higher BPL
fi:
recoveries than the conventional cell. The average recovery DWithout Air
for the HydroFloat was 93.8 %, compared to only 78.7% for 0
the conventional test. The best continuous HydroFloat test
result provided a BPL recovery of 92.4% at a concentration 0 20 40 60 80 100
ratio of 2.12. Fixed Carbon (%)
Testing of anthracite slag. An industrial slag sample (nomi-
nally 6 mm x 200 mesh) was screened at 6.35 mm to remove Figure 7 - Testing of anthracite slag using the HydroFloat
separator (6.35 mm by 200-mesh).
oversize tramp material prior to testing. The feed sample
contained", 12% fixed carbon in the form of anthracite coal
and ",30% Fe 20 3 and ",26% Ti0 2 . This sample was ideally Testing of heavy mineral sands. Two samples of heavy
suited for HydroFloat treatment due to the inherent hydropho- mineral sands were tested using the HydroFloat separator.
bicity of the low-density component (anthracite). The objec- The first sample contained unwanted carbonaceous matter as
tive of these tests was to recover the remaining fixed carbon well as an undesirably high pyrite content (average 0.7X lf(
at a product quality greater than 80% . Only seven tests were sulfur). To process this sample. the feed was first treated with
conducted on this sample due to the small amount of available sodium isopropyl xanthate to make the pyrite hydrophobic
material. Four tests were conducted utilizing the laboratory- and then passed through the HydroFloat unit. As shown in Fig .
scale HydroFloat with full teeter-bed aeration. Three were 8, the HydroFloat separator achieved sulfur and carbon rejec-
conducted with the HydroFloat operating as a traditional tions of up to 8()C/c and 55 * . respectively . These rejections
hindered-bed separator (i.e. , no teeter-bed air was employed). were maintained at a high product yield of approximately
Figure 7 shows the product grade and recovery plot for the 97 %. The second sample also contained an unacceptably high
+28-mesh fraction of the feed material. Without aeration, the carbon content (average O.92 ~' fixed carbon). However. no
hindered-bed separator was able to make a separation, al- xanthate was added for this sample because the sulfur content
though a product quality greater than 71 % fixed carbon could was already within product specifications . Figure 9 shows that
not be achieved. In contrast, aquality of more than 80% could the HydroFloat separator was also effective in treating this
be achieved when the teeter bed was aerated . Aeration al- sample. More than 81 * of the carbon was removed from the
lowed the product recovery to be increased by an overage of feed material at a product yield of nearly 95 ct{. The lower
10% to 15%, while simultaneously improving the fixed car- sulfur rejections retlect the low feed sulfur content (O .OY.4 ) of
bon content of the product. this particular sample.

MINERALS & METALLURGICAL PROCESSING 65 VOL. 18. NO . 2. MAY 2001

 100

80 ~
-.. ~

--=
0~

0
60
0 X
...
~

:
III
40
\00
20 X Sulfur
o Carbon
o~--~--~--------~--~
0.00 0.10 0.20 0.30 0.40 0.50
Product Assay (%)
Figure 8 - HydroFloat results forthe removal of impurities
from mineral sands (Sample #1).

100 r--------------,
80
o
DO
~ 0
o
""'
~
~ 60
c

...........
.~

40
~

20
I)( X Sulfur
o Carbon
Figure 10 - Pilot-scale HydroFloat separator.
o~------------------~
0.00 0.05 0.10 0.15 0.20 0.25
Product Assay (%) conditioning tanks where flotation and pH reagents were
added. The product from the conditioners was then directed to
Figure 9 - HydroFloat results for the removal of impurities the HydroFloat separator.
from mineral sands (Sample #2). Table I provides a comparison of test data from the pilot-
scale HydroFloat unit with that typically achieved by the
existing full-scale conventional flotation circuit now in opera-
Table 1 - Comparison of typical plant data and pilot-scale tion at the phosphate plant. At present, the plant typically
HydroFloat test results. operates with a BPL recovery of approximately 35% for the
+ l6-mesh feed and approximately 60% for the 16 x 35-mesh
Particle Plant cells HydroFloat HydroFloat
recovery, recovery,
feed. In comparison, the HydroFloat unit achieved a BPL
Size, grade,
mesh % % %BPL recovery of more than 60% for the + 16-mesh feed and nearly
85% for the 16 x 35 mesh feed. This represents an increase in
+16 = 35 61.4 72 .8 recovery of more than 40%. It is also interesting to note that
the +16-mesh fraction had a very high BPL content (72.8%
16 x 35 = 60 84.7 54.6
BPL) and a very low insol content (4.0%). The combined (i.e.,
Total = 50 80.5 56.8 +35 mesh) concentrate from the HydroFioat cell represented
at a total recovery of more than 80% with a BPL grade of
56.8%. This result compares very favorably with the existing
Pilot-scale testing plant recoveries of 80% to 85% normally achieved for the
After completing the proof-of-concept evaluations, sufficient finer 35 x I 50-mesh feed.
data were obtained to justify testing a pilot-scale HydroFloat
separator. To this end, a 0.6-m-square, 2.0-m-tall test unit was Summary and conclusions
fabricated and installed at a central Florida phosphate opera- A new separator, known as the HydroFioat unit, was devel-
tion (Fig. 10). Circuit feed was supplied at a rate of 20 to 30 oped to overcome some of the shortcomings associated with
t/h from a port located in the feed line to a bank of plant traditional flotation machines in recovering coarse particles.
dewatering cyclones used to prepare feed for existing conven- The novel characteristic of this separator is the formation of a
tional flotation cells. The feed slurry was passed through a hindered "teeter" bed of fluidized solids into which small air
hydroclassification circuit to produce a +0.6-mm underflow bubbles are introduced. The bubbles attach to hydrophobic
product. This material flowed by gravity into a bank of particles and create light bubble-particle aggregates that can

MAY 2001 • VOL. 18 NO.2 66 MINERALS & METALLURGICAL PROCESSING

be separated from hydrophilic particles based on the principal References
of differential density. Benefits of this new separator include Arbiter. N .. and Harris. C.C., 1962, "Flotation kinetics," Froth Flotation - 50th
enhanced bubble-particle contacting, better control of particle Anniversary Volume, Chapter 8, D.W. Fuerstenau, ed .. American Institute
of Mining, Metallurgy and Petroleum Engineers, Inc. New York. pp. 215-
residence time, lower axial mixing/cell turbulence, and re- 246.
duced air consumption. Barbery, G., 1984, "Engineering aspects of flotation in the minerals industry:
Laboratory tests were conducted with both batch and Flotation machines, circuits and their simulation," The Scientific Basis of
Flotation, K.J. Ives, ed., NATO Advanced Institute Services. Series E:
continuous HydroFloat cells to evaluate the potential of this Applied Sciences, No. 25, Martinus Nijhoff Publishers, Boston, pp. 289-
new technology for the upgrading mineral samples from 348.
various sources (e.g., phosphate matrix, anthracite slag and Bull, W.R, 1966, "The rates of flotation of mineral particles in sulphide ores,
mineral sands). The test data indicate that the HydroFloat cell Proceedings, Australian Institute of Mining and Metallurgy, No. 220, De-
cember 1966, pp. 69-78.
is capable of increasing coarse particle recoveries by 20% King, RP., Hatton, T.A., and Hulbert, D.G., 1974, "Bubble loading during
over conventional flotation. Furthermore, the concentrate flotation," (technical note), Transactions of the Institute of Mining and
grades were also improved in some cases due to a reduction in Metallurgy, Section C, March, p. C9.
Mankosa, M.J., Luttrell, G.H., Adel, G.T., and Yoon, R.-H., 1992, "A study of
coarse particle misplacement. axial mixing in column flotation," International Journal of Mineral Process-
In light of promising laboratory data, a pilot-scale ing, Vol. 35, pp. 51-64.
HydroFloat cell was installed in an industrial phosphate plant. Schulze, H.J., 1984, "Physico-chemical elementary processes in flotation,"
Test data obtained with this unit to date suggest that the BPL Developments in Mineral Processing, Vol. 4, Chapter 5, Elsevier, New
York, pp. 238-253.
recovery of the + 16-mesh feed can be nearly doubled (i.e., Schulze, H.J., and Gottschalk, 1981, Proceedings of the 13th International
increased from approximately 3S% to more than 6S%) through Mineral Processing Congress, Warsaw, Poland, June 1979, PartA, Elsevier,
the application of this new technology. In addition, the data New York, p. 63
Sutherland, K.L., 1948, "Kinetics of the flotation process," Journal of Physical
suggests that the HydroFloat cell may be used to increase the
Chemistry, Vol. 52, p. 394.
BPL recovery of the 16 x 3S-mesh material from about 60% Woodburn, E.T., King, R.P., and Colborn, RP., 1971, "The eHect of particle
to nearly 8S%. Therefore, the combined recovery of +3S- size distribution on the performance of a phosphate flotation process,"
mesh product from the HydroFloat cell compares very favor- Metallurgical Transactions, Vol. 2, pp. 354-362.
Yoon, R.H., and Luttrell, G.H., 1989, "The effect of bubble size on fine particle
ably to the existing plant recoveries of 80% to 8S%, which are flotation," Mineral Processing and Extractive Metallurgy Review, Vol. 5,
normally achieved for the 16 x ISO-mesh feed. Gordon and Breach Science Publishers, Inc., Great Britain, pp. 101-122.

MINERALS & METALLURGICAL PROCESSING 67 VOL. 18, NO.2. MAY 2001