International Journal of Mineral Processing, 23 (1988) 279-292 279

Elsevier Science PublishersB.V., Amsterdam-- Printed in The Netherlands

Selectivity in Column Flotation Froths

J.B. YIANATOS,J.A. FINCH and A.R. LAPLANTE

Department o[ Mining and Metallurgical Engineering, McGill University, Montreal, P.Q.
H3A2A 7 (Canada)
(ReceivedJanuary 22, 1987; acceptedafter revision August 21, 1987)

ABSTRACT

Yianatos, J.B., Finch, J.A. and Laplante, A.R., 1988. Selectivityin columnflotation froths. Int.
J. Miner. Process., 23: 279-292.

Selectivity in the froth zone of two industrial flotation columns in a Mo-cleaningcircuit has
been studied and modelled.Froth depth is shownto be a significantfactor. Froths deeper than 1
m gave selective recovery of Mo, with upgrading of 10-15% absolute. Froths less than 50 cm
showed no selectivity.This differencebetween deep and shallow froths is attributed to mixing
caused by the wash water input. A modelbased on plug flowwas developedand shown to fit the
grade profilesthrough the froth.

INTRODUCTION

In flotation, selectivity or preferential recovery of minerals in the froth pro-

vides additional upgrading (Moys, 1978; Cutting et al., 1981 ). In mechanical
cells two ways of achieving selectivity are: drainage of entrained fines and co-
alescence which reduces bubble surface area causing rejection of the less firmly
attached particles. In flotation columns, the use of wash water stabilizes the
froth and largely eliminates hydraulic e n t r a i n m e n t at the p u l p / f r o t h interface,
particularly at gas rates smaller t h a n 1.5-2.0 c m / s (Yianatos et al., 1987).
Nevertheless deep froths (1-2 m) are often used and it is reasonable to ques-
tion if upgrading through the froth occurs. Basic studies by Yianatos et al.
(1986) have shown coalescence does occur in column froths which could pro-
mote some selectivity.
This paper reports local grade and percentage solids along a column froth in
a molybdenite cleaning circuit. This information, together with a mathemati-
cal model describing mass transport through the froth, is used to analyse min-
eral selectivity in a column froth.

0301-7516/88/$03.50 © 1988Elsevier Science Publishers B.V.

28O

MODELLING

Background

Moys (1978) developed a comprehensive model for flotation froth behav-

iour, using the concept of plug-flow introduced by Cooper (1966). The simu-
lation results were verified in a laboratory cell where transverse motion was
effectively eliminated. This approach represents a boundary condition for con-
ventional froth behaviour, since the actual process always has a significant
transverse component. More recently, Moys (1984) has described the froth
streamlines using a two-dimensional model. Measurements of residence time
distribution showed this approach constitutes a good description of the phys-
ical behaviour of froths.
In column flotation the froth transport most likely approximates plug-flow
for the following reasons. Firstly, vertical distances are significantly greater
than horizontal ones (unlike a mechanical cell) due to the large froth heights

Wash Water
l Ci =
Z¢ - - Lip Level
Conventional
Froth
l z.
FI (Z + dZ}
Countercurrent Ri(Z+dZ) Ei(Z +dZ)
Bubble Bed
Z +dZ t
dMi
, t
Ri(Z) F~IZ)
Ei(Z}

Z=O Interface
t Level
Ri(O) F~(O)
Collection Zone E,(O)

Fig. 1. Mass transport alongthe froth.

 281

achieved with wash water (Yianatos et al., 1986). Secondly, the froth is con-
tacted countercurrent with a wash water stream, evenly distributed through
the whole cross-section. And thirdly, the concentrate overflows along the whole
perimeter, usually via a thin (less than 10 cm thick) conventional froth layer
above the wash water entrance.
In this paper, a plug-flow model based on Moys' (1978), is developed to
describe the behaviour of a column froth.
Fig. I shows schematically the mass flowrates in the column froth.

Assumptions
(1) Mineral particles of species i attach to the bubbles in the collection zone
and enter the froth at a rate Fi(O) (g/s).
(2) Particles may detach from the bubble surface and enter the liquid in the
froth, depending on the strength of attachment and on the presence of mech-
anisms which encourage detachment, e.g. bubble coalescence and washing. The
rate of detachment is proportional to the concentration of the component in
question. Taking a mass balance over a differential element yields:
dFi(z) -kiFi(z)
- - - - (1)
dz V(z)
where ki is the detachment rate constant (s -1) and V(z) the bubble rise ve-
locity (cm/s) (see Notation). V(z) is given by:

V ( z ) - Jg (2)
tg (z)
where Jg is the superficial gas rate (cm/s) and tg(Z) is the gas holdup, which
is a function of z.
Combining eqs. I and 2 yields:
dFi (z.......~)_-kiFi (z)Q (z) (3)
dz J.
The solution of the above differential equation is:
Fi (z) =Fi (0) exp[-kiT(z) ] (4)
where T(z) represents the residence time of the bubbles in the froth at level z,
and is equal to:

it. (z) dz
T(z)= °
J.
(5)
282

NOTATION

n~: column cross-section (cm ~)

C, concentrate mass flowrate of mineral i (g/s)
Ei mass flowrate of entrained mineral i (g/s)
F, mass flowrate of mineral i (g/s)
Gi fractional grade of mineral i
J~ superficial gas rate (cm/s; positive upwards)
Jh superficial net downward rate (cm/s; positive downwards)
de. superficial concentrate rate (cm/s)
ki detachment rate constant of mineral i (1/min)
ki' washing rate constant of mineral i (1/min)
Mi mass of mineral i (g)
P~ solid mass percent in the froth
R, net downward flowrate of solids (g/s)
T bubble residence time in the froth (rain)
V bubble rise velocity (cm/s)
Z axial distance above the pulp/froth interface (cm)
Zc concentrate or lip level (cm)
Ze level at which 95% of the entrained particles are rejected (cm)
Zw wash water input level (cm)
fractional gas holdup
P liquid density (g/cm :~)
ps solid density (g/cm :~)

It is understood that particle reattachment may occur in the froth, but this
phenomenon is not considered explicitly in the model. The detachment rate
constant ki, therefore, represents the net loss of minerals from the ascending
bubbles.
(3) Bubbles entering the froth carry a thin layer of water which preferen-
tially entrains fine particles. Entrained minerals rise at a rate Ei (z) and veloc-
ity V(z) towards the top of the froth and are washed out into the downward-
flowing stream at a rate proportional to their concentration.
An analogous analysis to that for detachment yields:

Ei (z) -~Ei ( 0 ) exp[ - k i ' T ( z ) ] (6)

where ki' is the "washing" rate constant for entrained particles.

(4) Solid particles in the liquid (produced by detachment and washing ) flow
downward at a net velocity Jb (cm/s) and rate Ri(z) (g/s) given by:

Ri (z) =Fi (z) +El (z) - Ci (7)

where Ci is the mass flowrate of the species i at the concentrate level zc, i.e.
Fi(zc).
(5) Given the flowrates and flow velocities at any point, mineral mass frac-
 283

tions in the froth can be calculated by noting that the mass of a particular
mineral in the froth between heights z and z + dz is given by:
dMi(z)--{ [Fi(z)+Ei(z)]{~g(Z) Ri(z)[1-eg(Z)]}dz (8)
Jg ÷ Jb
and the mass fraction (grade) of mineral i is obtained from:
d M i (Z) (9)
Vi (z) - ~ dMj (z)
J
(6) Solid mass percent in the froth can be calculated:
100 ~ dMj (z)
P~(z)- (10)
(1-p/p,) ~dMj (z) +Acp[1 - ~g (Z) ]dz
J

Simulation
The simulation is an iterative procedure where the main assumption is the
shape of the gas holdup profile. The assumed shape is validated if it can repro-
duce the measured percent solid profile.
(A) Data input and assumptions
(1) From each experimental plant test, the following input data were ob-
tained: Jb, Jc, Jg, Ac, Ci, Zw, z¢, p, Ps.
(2) A first guess for Gi(0) was obtained from the experimental grade profiles.
(3) A first guess for Ps (0) was obtained from the experimental solid % profile.
(4) A gas holdup profile was assumed considering: (a) the gas holdup at the
top of the froth is known:

{~g (Zc) - - - - (li)

J~+Jg
(b) for bubble sizes and flow rates used in column flotation, there is a narrow
range for Q(O), typically 0.55-0.60 (Yianatos et al., 1986); (c) the froth can
be divided into two different sections, viz., the countercurrent bubble bed and
the naturally drained froth; the depth of each section is Zwand z c - Zw, respec-
tively, as shown in Fig. 1; (d) the shape of the gas holdup profiles is near
parabolic (Yianatos et al., 1986) and is fitted by second order models.
(B ) Data processing
(1) ~ d M j (0) is calculated as a function of P~(0) and eg(0), using eq. 10.
J
(2) The mass per component at the interface level is calculated as:
(0) (0) dMj (12)
J
284

(3) The initial mass rates are estimated from eqs. 7 and 8.
Two extreme conditions can be distinguished: (a) no entrainment; Ei (0) = 0;
and (b) no detachment, Fi (0) = Ci (if entrainment is completely eliminated
along the froth). It has been shown (Yianatos et al., 1987) that normal oper-
ation of a column froth is close to Ei(O)= O. Under this condition the initial
mass rate equals Fi (0). Alternatively, entrainment can be assigned a fraction
of the total mass rate entering the froth.
(4) Detachment rate constants are calculated from eqs. 4 and 5.
Similarly, washing rate constants can be estimated by:
- l n [El (Ze)/Ei (0)]
ki' - (13)
T(ze)
where Ze is the froth depth at which the entrained particles have been washed
out (taken here as 95% entrained particle rejection, estimated from previous
experiments (Yianatos et al., 1987).
Grade profiles are only marginally dependent on the shape of the gas holdup
profile. Thus the iteration to fit the grade profiles can proceed even if the pro-
file shape assumed is not the best one. Solid percentage, however, is strongly
dependent on gas holdup. The profile shape of the gas holdup is then used to
adjust the solid percentage profile.

PLANT STUDY

Experiments to measure grade and percentage solid profiles through the froth
zone were performed on two flotation columns in a MoS2 cleaning circuit at
Mines Gasp~, Quebec (Fig. 2). Both columns are 12 m in height, one is 0.45 m
(18 in.) square and the other 0.91 m (36 in.) square. Operation and perfor-
mance of the columns have been previously described (Coffin and Miszczak,
1982). Operating conditions at the time of the tests for the two columns are
listed in Table I.

Test procedure

Samples at different levels in the froth and collection zone, and of the feed,
concentrate and tailings were taken while the columns operated at steady state.
Samples from the froth and collection zone were taken from the top of the
column using a sampler consisting of four independent tubes (1 cm inner di-
ameter) of different length, each connected to a 1-1 Erlenmeyer flask and to a
central vacuum system. Samples were taken by opening lines simultaneously.
After sampling, the percentage solid of the slurry was determined and X-ray
fluorescence analysis was performed for copper, molybdenum, iron and silicon.
These were converted to mineral assays assuming the minerals present are
 285

~_ water water

•MOLY
CONC.

air
Fig. 2. Columnflotationcircuitat Mines Gasp~.
TABLE I

Column operating conditions

TestNo.: G1 G2 G3 G4 G5 G6

Date March 1986 March 1986 July 1985 July 1 9 8 5 March 1986 March 1986
Columnsize (ca) 45x45×1200 45X45×1200 45X45×1200 45×45×1200 91X91X1200 91X91X1200
Froth depth (ca) 58.00 145.00 150.00 150.00 45.00 130.00
Sup. gas rate (ca/s) 1.16 1.27 0.95 0.95 1.45 2.09
Sup. bias rate (ca/s) 0.60 0.60 0.46 0.46 0.38 0.38
Sup. feed rate (ca/s) 0.85 0.75 1.54 1.60 0.94 1.05
Residence time (rain) 11.30 12.10 10.30 10.20 12.80 10.20

chalcopyrite (CuFeS2), molybdenite (MoS2),pyrite (FeS2)and quartz (SiO2).

The quartz assay is used merely to represent non-sulphide gangue, thus min-
eral totals do not equal 100%.
Mass balance adjustments were performed using a least square algorithm
(Smith and Ichiyen, 1973), and the solid split between concentrate and tails
as well as mineral recoveries were estimated.

RESULTS AND DISCUSSION

General observation

Fig. 3 shows the experimental grade profile from the top of the column down
to 65 cm below the interface (i.e. into the collection zone), for test G5. In this
test the 91 cm column was operated with a relatively shallow froth (45 cm
depth).
The grades up to the interface are relatively constant, illustrating that the
collection zone is well mixed (Yianatos et al., 1987). At the interface a sharp
286

6O
I A

J r T E S T G5

Chalcopyrite
4O Pyrite

Mol¥

Silica

I.
2O
I

lo I
• • • • • •

t I I t I I
20 40 60 80 100 120
DISTANCE B E L O W THE LIP, cm

Fig. 3. Gradeprofilesaroundthe interface.Columnsize 91 × 91 × 1200cm {test G5).

change in grade occurs which is related in part to the suppression of entrain-

ment. Through the froth relatively little further change in grade is observed.
In deeper froths, however, upgrading through the froth did occur, as shown in
Fig. 4. This effect of froth depth was consistent for both columns. It reflects
the mixing imparted by the wash water. Previous tests on these columns had
shown wash water addition causes mixing in the froth to a depth of about 50
cm (Yianatos et al., 1987).

Testing the model

From the above observations model testing was restricted to froths deeper
than i m for which the plug-flow assumption is more justified. Figs. 4, 5 and 6
(test G2) show the experimental profiles of mineral grade, solids percentage
and gas holdup, respectively, with the model simulation assuming no entrain-
ment (continuous line). Note that the data taken from inside the froth con-
verge to those taken from the concentrate (lip level). Figs. 7 and 8 show the
experimental and simulated grade profiles for tests G3 and G6, respectively.
These results show that a significant selective process occurs along the froth,
and the process can be simulated reasonably well by using the plug-flow model
without entrainment.
In order to assess particle entrainment, test G2 was simulated considering
15% of the total mass rate of minerals entered the froth by entrainment, and
ze is 50 cm (95% of entrained particles is rejected in the first 50 cm of the froth
 287

90

SO

70

TEST G2 I

Data eeAO I
60
Model ~ (no entrainment)
.... (15% entrainment)

30

20
T
t
i
I
10

CHALCOPYRITE JL
I SILICA tl
r ;
0 50 100 iS0
DISTANCE BELOW THE LIP, cm
Fig. 4. Grade profilesalongthe froth. Columnsize 45 ×45 X 1200cm (test G2).

column (Yianatos et al., 1987). Results are shown in Figs. 4 and 5 (dashed
lines). These results show there is no improvement in model fitting by intro-
ducing the entrainment terms. This was expected from previous testing (Yian-
atos et al., 1987), which showed that rejection of entrained particles does not
take place throughout the froth, but rather very close to the pulp/froth interface.
Table II summarizes total recovery ( R t ) , froth zone recovery (Rf) and the
derived collection zone recovery (Re) for tests G2, G3, G4 and G6. Molybde-
num recovery in the froth of the large column (test G6) was 81.1%, while for
the smaller column (tests G2, G3 and G4) it varied from 84.3% to 95.3%. Chal-
copyrite and pyrite (depressed with NariS and NaCN, respectively) were sig-
nificantly lower in recovery than molybdenite. In general, pyrite recovery
through the froth was higher than chalcopyrite. Information such as this can
be interfaced with models of the collection zone for column scale-up (Dobby
and Finch, 1986).
Table III shows the detachment rate constants ki for tests G2, G3, G4 and
G6. The range is 0.04-0.93 min-1. The less hydrophobic minerals have the
288

3o/ i
Data
TEST G2

Model
p- I
z
,~ 2c

_o
g,
I
I
~ tOJ
®
I
! I
0 50 100 160
DISTANCE BELOW THE LIP, cm

Fig. 5. Solidspercentage profile alongthe froth. Columnsize 45 × 45 X 1200cm (test G2).

~oo t
I
t
I
I TEST G2 I
I
Data O
.o ,ii_ Predicted
Profile
¢

80
O

70

60

S0 I I
50 100 150
DISTANCE BELOW THE LIP, cm

Fig. 6. Gas holdupprofilepredicted. Columnsize 45 × 45 × 1200 cm (test G2).

larger detachment rate constants and consequently are selectively rejected from
the froth.
There is little data on d e t a c h m e n t rate constants with which to compare the
present results. Moys (1978) estimated a detachment rate constant of ki = 5.1
min -1 for pyrite in the upper half of a batch conventional froth (20 cm total
depth). This, however, represents a boundary condition of a collapsing froth.
The relatively low values of h~here testify to the stability of deep column froths.
 289

80
I TEST G3 I

L
60- I

• 3O
I

=o I
I
I
I ~HALCOPYRITE & 1
I
i SILICA "~l ' i $ ql, i
0 SO 100 150
DISTANCE BELOW THE LIP, cm
Fig. 7. Grade profiles along the froth. Co|utah size 45 × 45 × 1200 cm (test G3 ).
so I t
I 'I TEST G6 I
I
i I Data .*,• [
.o,.,..____~----

# I

2o r ~ " I

• SILICA •
• • t I
I
+ , , +
GO 100
DISTANCE BELOW THE LIP cm
Fig. 8. Grade profiles along the froth. Column size 91 × 9 1 × ]200 cm (test G 6 ) .
290

'FABLE II

'Fotal, froth zone and (derived) collection zone recoveries i';, t

Test No.: G2 G:~ G4 G6

Recovery: R, Rf R,, R, R, R, R, & R,. R, R, R,

MoS_, 72.04 95.30 75.59 83.00 84.30 98.50 87.70 93.00 94.30 56.71 81.10 69.93
FeS~ 21.30 55.30 38.52 33.20 66.20 50.10 33.20 73.10 45.40 9.42 78.00 12.08
CuFeS~ 12.90 65.10 19.82 23.80 56.20 42.30 21.50 60.80 35.30 3.36 50.46 6.66
SiO~ 14.34 50.70 28.28 36.70 56.90 64.40 35.70 69.70 51.20 7.25 65.75 11.03

R, = total recovery; R~= froth zone recovery; R,, = collection zone recovery (R,, = Rt/Rf).

Figs. 9 and 10 show the recovery profiles along the froth for the 91-cm col-
umn (test G6) and the 45-cm column (test G2), respectively. Fig. 9 shows that
molybdenite was recovered preferentially. The average percentage solids was
around 2%, which is relatively low. For the 45-cm column, Fig. 10 shows the
preferential recovery of molybdenite is even more significant. Average solid
percent in the froth in this test was higher, around 15%, which may favour
froth stability and selectivity.
The froth reduces to some extent the yield of values. Experimentally it has
been shown that 5-20% of the values return from the froth to the collection
zone by disengagement from the bubbles in the froth. However, the froth ef-
fects have a much larger impact on concentrate grade, because of the selective
rejection of the less hydrophobic minerals. Increases in values grade by 10-
15% absolute are observed here in the column froth. Comparing feed grade to
the grade entering the froth shows upgrading from 15% to 25% absolute, which
can be attributed to upgrading in the collection zone. Thus, in the case de-
scribed here, 40% of the total upgrading was due to the froth zone.

Detachment mechanisms

The data show that particle disengagement occurs. One possible mechanism
is due to fluid drag. Drag caused by wash water increases towards the top of
the froth, because the bias rate is constant and the liquid holdup decreases.
Another mechanism for particle release is bubble coalescence. If the bubbles
are fully loaded (a condition more likely to be met in a column than in me-
chanical cells) any coalescence will lead to disengagement.
Detached particles could be entrained into the concentrate (especially if
released near the top of the froth) or returned to the collection zone. Alterna-
tively the detached particles could be selectively re-attached in the froth, de-
pending on particle hydrophobicity and size and bubble surface availability.
Selective detachment and re-attachment increases mineral grade along the
 291

TABLE III

Detachment rate constants, hi ( m i n - 1 )

Test No.: G2 G3 G4 G6

MoS2 0.04 0.08 0.04 0.28

FeS2 0.41 0.20 0.16 0.34
CuFeS2 0.30 0.28 0.25 0.93
SiO2 0.48 0.28 0.18 0.57

100

70
t"
60 f ~ TEST G6

50 I I I I I I
20 40 60 80 1OO 120 140
DISTANCE BELOW THE LIP, crn
Fig. 9. Recovery profiles along the froth. Column size 91 × 91 × 1200 cm (test G6).
100

~ so-

re 7O

SO
TEST G2

50 I I I I I I I
0 20 40 60 80 100 120 140 160
DISTANCE BELOW THE LIP, cm
Fig. 10. Recovery profiles along the froth. Column size 45 × 45 × 1200 cm (test G2 ).
292

f r o t h (at the e x p e n s e of r e c o v e r y }. R e g u l a t i o n of f r o t h b e h a v i o u r b e c o m e s a
p o t e n t i a l c o n t r o l a c t i o n ( K a w a t r a a n d Seitz, 1984), e.g. by using f r o t h depth.
T h e a n a l y s i s of individual m e c h a n i s m s of p a r t i c l e d e t a c h m e n t a n d r e a t t a c h -
m e n t is c o m p l e x a n d difficult to e v a l u a t e f r o m p l a n t o p e r a t i n g data, a n d no
c o n c l u s i o n c a n be d r a w n f r o m t h e p r e s e n t work. Closely c o n t r o l l e d l a b o r a t o r y
e x p e r i m e n t s are required.

CONCLUSIONS

(1) A p r o c e d u r e for q u a n t i f y i n g c o l u m n f r o t h p h e n o m e n a h a s b e e n devel-

o p e d a n d s h o w e d to fit well p l a n t data.
(2) In t h e e x a m p l e of M o cleaning, c o l u m n f l o t a t i o n f r o t h s larger t h a n I m
in d e p t h s h o w e d selective M o r e c o v e r y w h i c h i n c r e a s e d m o l y b d e n i t e grade b y
10% to 15% absolute.
(3) F r o t h s e l e c t i v i t y d e c r e a s e d w i t h shallow f r o t h d e p t h s , tess t h a n 1 m,
b e c a u s e of i n c r e a s e d m i x i n g due to t h e closeness of t h e w a s h w a t e r e n t r a n c e to
t h e p u l p / f r o t h interface.

REFERENCES

Coffin, V.L. and Miszczak, J., 1982. Column flotation at Mines Gaspd. CIM, 14th IMPC, Toronto,
Ont., pp. IV.21.1-19.
Cooper, H.R., 1966. Feedback process model of mineral flotation, I. Development of a model for
froth flotation. Trans. Soc. Min. Eng. AIME, 235: 439-446.
Cutting, G.W., Watson, D., Whitehead, A. and Barber, S.P., 1981. Froth structure in continuous
flotation cells: relation to the prediction of plant performance from laboratory data using pro-
cess models. Int. J. Miner. Process., 7: 347-369.
Dobby, G.S. and Finch, J.A., 1986. Flotation column scale-up and modelling. CIM Bull., 79 (889):
89-96.
Kawatra, S.K. and Seitz, R.A., 1984. Effect of froth structure on coal flotation and its use in
process control strategy development. APCOM'84, I.M.M., London, pp. 139-147.
Moys, M.H., 1978. A study of a plug-flow model for flotation froth behaviour. Int. J. Miner. Pro-
cess., 5: 21-38.
Moys, M.H., 1984. Residence time distributions and mass transport in the froth phase of the
flotation process. Int. J. Miner. Process., 13: 117-142.
Smith, H.W. and Ichiyen, N., 1973. Computer adjustment of metallurgical balances. CIM Bull.,
pp. 97-100.
Yianatos, J.B., Finch, J.A. and Laplante, A.R., 1986. Holdup profile and bubble size distribution
of flotation column froths. Can. Met. Q., 25 ( 1 ): 23-29.
Yianatos, J.B., Finch, J.A. and Laplante, A.R., 1987. The cleaning action in column flotation
froths. Trans. I.M.M. In press.