You are on page 1of 6

A Model of Particle Sliding Time for Flotation Size Bubbles

G. S. DOBBY AND J. A. FINCH


Department of Mining & Metallurgical Engineering, McGill University, Montreal, Canada
Received March 6, 1985; accepted May 31, 1985

A model is developed to calculate the time a particle remains in contact with a flotation bubble after
collision (sliding time). This is a necessary component in calculating attachment efficiency. The model
is based on recent determinations of the liquid velocity distribution over the surface of rigid spheres for
Reynolds numbers less than 400. The model predictions fitted well the available data on sliding time.
The reported fit using a model based on potential flow is shown to result from an error which caused
sliding time to be overestimated. The model introduces the important result that sliding time is strongly
particle size dependent. © 1986AcademicPress,Inc.

1. INTRODUCTION of the bubble is required. Typically, flotation


To model particle collection by direct en- bubbles have an equivalent spherical diameter
counter with bubbles in flotation it is conve- between 0.05 and 0.15 cm and Reynolds
nient to divide the process into two subpro- number (Reb) from 30 to 300. They act as rigid
cesses, collision and attachment (1). Collision spheres due to the surfactants used for db less
has been analyzed several times, most recently than approximately 0.12 cm (3). Consequently
by Weber and Paddock (2). In contrast, at- recent experimental work on the velocity dis-
tachment has received relatively little attention tribution at the surface of a solid sphere (4) is
since the pioneering work of Suthedand in relevant to bubbles in flotation. Figure 1 shows
1948 (1). the tangential velocity distribution measured
The objective of an attachment model is to at Reb = 290 and 0 = 45 ° (along with the
predict the fraction of particles which having theoretical curves for potential and Stokes
collided with the bubble remain in contact flow).
with the bubble for a time greater than the Figure 1 shows there is a significant tangen-
induction time ti. This requires a knowledge tial velocity gradient on the upper surface of
of: (a) the proportion of particles colliding be- a flotation bubble. An important consequence
tween the front stagnation point and a given of this velocity gradient is best illustrated by
angle 0n, (b) the angle 0m at which fluid an example. Consider a bubble, db= 0.12 cm,
streamlines cause the particle to start to deviate Reb = 290, and two spherical particles of dp
from the bubble surface, and (c) the time to = 12 and 36 ~tm. Assume that the tangential
travel from Onto 0m (sliding time). It is the last velocity gradient is constant across the dimen-
two factors, the value of 0m and the time to sions of both particles and that each particle
travel from 0n to 0,n, which are addressed in travels with the velocity that the liquid would
this communication. have had at the particle center. The particle
center is located at the coordinate r* = 1
+ rp/rb (or, 1 + dp/db), where r* is the di-
2. MODEL OF SLIDING TIME, ts
mensionless radial coordinate r/rb. Then, at
2.1. Liquid velocity distribution over upper 0 = 45 ° the 12-/~m particle (r* = 1.01 on
c
surface of bubbles. In order to calculate ts the Fig. l) would be traveling with a velocity of
liquid velocity distribution over the upper part "~0.1Ub while the 36-#m particle (r* = 1.03)
493
0021-9797/86 $3.00
Copyright © 1986 by Academic Press, Inc.
Journal of Colloid and Interface Science, Vol. 109, No. 2, February 1986 All rights of reproduction in any form reserved.
494 DOBBY AND HNCH

u~ = u0 + ursin 0, [2]
1.0 - ~ Potential flow where ux is the terminal settling velocity of
J the particle. In potential flow 0m = 1I/2. Con-
sequently the time t~ to travel from 0n to 0m is
0.8
II/2 r'dO
.
ts =f
J On Upo
[3]
=m 0.6 Re = 2 9 0
which upon integration gives
II
"2 2r* ( ~ )
0.4
ts = 2(Ub + UT) + Ub(1/r*) 3 In cot • [41

The important difference from Sutherland's


0.2 ~ o w solution (Ref. 1, Eq. (13)) is the 2 in the nu-
merator, rather than 4; Sutherland's solution
0 I I I overestimates ts by a factor of 2. (As well, his
1.1 1.2 1.3 1.4
solution does not account for UT.)
r*
2.3. Intermediate Reb solution for ts. The
FIG. 1. Liquid tangential velocity on the surface of a tangential velocity on the bubble surface can
solid sphere versus distance at 0 = 45° (8). Velocityand be computed from the surface vorticity ~s.
distance are dimensionless.
Surface vorticity is the dimensionless rate of
change of tangential velocity with dimension-
less radial distance at the surface. Woo (8) pre-
would be traveling at ~0.25ub. If this differ- sented data on ~ derived from numerical so-
ence is maintained at all angles then, if the lutions to the Navier-Stokes equations for Re
two particles start at the same angle on the = 0.2 to 400 and 0 = 0 ° to 180 °. Results of a
bubble, the 12-#m particle would have a slid- regression analysis on these data are presented
ing time about 2.5 times that of the 36-~tm in Appendix 2.
particle. Figure 2 shows again the experimental data
Potential flow was assumed by Sutherland at Reb = 290 and 0 = 45 ° and the computed
(l) to derive an analytical solution for t~. At ~s (~s = 15.7). Also shown are two linear mod-
the time, there was little alternative but Fig. 1 els which are used to approximate the exper-
clearly shows that potential flow is not relevant imental data. These models are
to flotation size bubbles; potential flow gives
velocities which are too high and fails to pre- u* = 0.7~s(r* - 1) for r* ~< 1.06 [5a]
dict the important particle size effect. How- and
ever, before developing a more appropriate
model, the solution based on potential flow u* = (0.06)0.7~s for r* > 1.06. [5b]
will be rederived. This is done to correct The factor 0.7 fits the velocity gradient at Reb
the result of Sutherland which is still being = 290 and 0 = 45 ° and it is assumed that this
used (5-7). factor also applies at all other Reb and 0.
2.2. Potential flow solution for ts. In poten- For dp/db <<.0.03 (i.e., r* - 1 ~< 0.03) Eq.
tial flow the tangential liquid velocity over the [5a] alone is an adequate fit. Therefore particle
top half of the bubble at the particle center is tangential velocity is
given by
uo = UbSin 0(1 + 0.5(1/r*)3), [1] Upo = 0.7~sUb(d~) + uTsin 0
where the symbols are given in Appendix 1.
Therefore, the tangential particle velocity is for dp/db < 0.03. [6a]
Journal of Colloidand InterfaceScience, Vol.109,No.2, February1986
PARTICLE SLIDING TIME 495

i ! ! i
ponent of the liquid velocity Ur calculated at
1.0 /,~ ,~s= 15.7 r = rb + rp (directed away from the bubble
surface). An estimate of Ur close to the sphere
0.8 is obtained as follows.
/ ~ re1 For axisymmetric flow around a sphere

II
0.6
%= u~ = r*2sin 0 '
~ E quat~n 5a
0.4 where ~ is the dimensionless stream function.
Near the surface of the sphere, the stream
function can be approximated by (10)
0.2
= 1 r,2 ~s sin 0. [9]

1.06 1.1
I i
1.2
I
1.3 Since ~s is known as a function of 0, ff can be
I
1.4
calculated and 0~/00 can be estimated at any
r* value of r* and 0. Thus, u~ can be calculated
and compared to UTCOS0. Some values of 0m
FIG. 2. Liquid tangential velocitygradient at surface of
determined in this m a n n e r are reported in Ta-
a sphere at Re = 290 and 0 = 45 °. From Eq. [A-l], ~s
= 15.7 under these conditions. ble I for Reb = 100, dp = 20 pm, and pp = 1.3
to 7.0 g/era 3. 0m increases as the particle den-
sity increases. It can also be seen from Table
For dp/db > 0.03 the particle velocity is cal- I that 0m is insensitive to dp.
culated by dividing the particle into two zones,
one that sees a velocity gradient described by
TABLE I
Eq. [5a], the other that sees a constant velocity
(Eq. [5b]). Then up0 is given by Maximum Angle of Contact O~

Upo = O.7~sUb[(dp -- O'O3db]o 06 Conditions


4, /" db = 0.1 cm
Ub= 10 cm/s
+\
(°-°3%o31
dp ] " J + u x s i n 0
p =0.01 P
Reb = lOO
A. dp = 20 pm
for dp/db > 0.03. [6bl
Pp
(g/cm3)
Om
(degrees)
The particle sliding time can now be cal-
culated. With 0 in degrees, this is 1.3 67
2.0 71
3.2 75
ts = ( ~ ) I I ( d b + dp)/Upo, [7] 4.0 77
5.5 79
where zT~ is the average Upe determined from 7.0 81
Eqs. [6a] or [6b] using average ~s and sin 0. B. pp = 4.0g/em3
2.4. Value of Ore. The end of particle-bub-
ble contact (assuming that attachment does (am) (degrees)
not occur) is calculated by determining the
angle at which the radial c o m p o n e n t of the 5 77
particle settling velocity Ux (directed toward 20 77
40 77
the bubble surface) is equal to the radial com-
Journal of Colloid and Interface Science, Vol. 109,No. 2, February 1986
496 DOBBY AND FINCH

6O
° o •
Equation (7) 7 •
/
Experimental Values
uJ • /~,

20 ~ • • •

I i i
lo 2'0 3'o 4'o 2'0 go 7o 8o 9o
ANGLE OF CONTACT, On(degrees)

FIG. 3. Particle sliding times; measurements by Schulze and Gottschalk (5, 9) and predictions from Eq.
[7] and potential flow assumption (Eq. [4]).

3. TEST OF MODEL = 90 °.) Also shown in Fig. 3 is the sliding t i m e


A n e x p e r i m e n t a l test o f the m o d e l is p r o - p r e d i c t e d a s s u m i n g p o t e n t i a l flow (Eq. [4]).
v i d e d b y the p h o t o g r a p h i c d a t a o f Schulze a n d This gives p o o r agreement because the velocity
G o t t s c h a l k (5, 9). T h e y r e p o r t e d sliding t i m e up0 e s t i m a t e d b y p o t e n t i a l flow is t o o high (see
as a f u n c t i o n o f angle o f c o n t a c t (see Fig. 3). Fig. 1).
T h e c o n d i t i o n s ( a n d d e r i v e d data, e.g., up) are
given in T a b l e IIa. T h e sliding t i m e as pre- 4. DISCUSSION
d i c t e d f r o m Eq. [7] c o m p a r e s well with the
e x p e r i m e n t a l results. A s p e c i m e n c a l c u l a t i o n It is clear that potential flow is an i n a d e q u a t e
is s h o w n in T a b l e IIb; e.g., for 0n = 40 ° a n d a s s u m p t i o n for flow a r o u n d flotation size
0m = 90 ° ( a s s u m p t i o n ) , ts "~ 31 ms. ( U n d e r bubbles. In their original w o r k Schulze a n d
these c o n d i t i o n s 0m = 82 °. H o w e v e r , since the G o t t s c h a l k s h o w e d a r e a s o n a b l e fit with Suth-
e x p e r i m e n t a l m e a s u r e m e n t s were m a d e with e r l a n d ' s p o t e n t i a l flow s o l u t i o n (approx. Eq.
the a s s u m p t i o n o f c o n t a c t to 90 ° , the m o d e l [4]). H o w e v e r , this was a c o n s e q u e n c e o f the
calculations (Eq. [7]) were p e r f o r m e d at 0m o v e r e s t i m a t e in ts, resulting in a fortuitous fit

TABLE IIa TABLE IIb


Conditions Employed by Schulze and Gottschalk (5, 9) Specimen Calculation of t,

db= 0.306 cm Reb = 258 a= --0.1181


dp= 160 #m 0m = 90 ° b= +0.5125
op = 2.5 g]cm 3 0. = 40 ° c= -0.0032
tz =0.01 P d= -2.072 × 10-5
up + ub = 10 cm/s u~ t=
up = 1.58 cm/s (11) ~, sin 0 (cm/s) (ms)
.'. ub = 8.42 cm/s
Reb = 258 12.5 0.88 4.5 31.0

Journal of Colloid and Interface Science, Vol. 109, No. 2, February 1986
PARTICLE SLIDING TIME 497

to the experimental results. It is evident from The model introduces the important fact
Fig. 1 that for smaller values of dp/db than that sliding time decreases with increasing
that of the example presented here, the differ- particle size.
ence between the potential flow tangential ve-
locity and the true particle velocity becomes
APPENDIX 1: NOMENCLATURE
extremely pronounced.
The proposed model (Eq. [7]) fits well. This
model i~ firmly based on a knowledge of the db, rb diameter and radius of bubble
liquid velocity distribution. Some approxi- dp, rp diameter and radius of particle
mations are necessary to translate this to par- r radial coordinate
title velocity. The fluid velocity at the particle r* dimensionless radial distance from
centre is assumed while the particle actually bubble center, r/rb
experiences a velocity gradient across its di- Reb bubble Reynolds number
mension. This gradient will impart spin to the t~ time of contact from 0n to Om
particle, the consequences of which are ignored ti induction time
here. Also ignored are possible particle bounce u* Uo/Ub
and inertia effects; detailed analysis through Ub rise velocity of bubble
trajectory calculations indicate that these fac- ur radial component of liquid velocity
tors are not very important (12). ux terminal settling velocity of particle
These model simplifications not withstand- u0 tangential component of liquid veloc-
ing, an important effect of particle size on t~ ity
is revealed. Large particles have a higher ve- up0 particle tangential velocity
locity on the bubble surface compared to small u0 + ursin O
particles, giving them a shorter sliding time. 0 radial coordinate, measured from
The shorter sliding time will mean reduced front stagnation point
attachment efficiency (for a fixed induction 0n angle of collision (beginning of con-
time). This decrease in attachment efficiency tact)
with particle size, the authors have shown, 0m angle at which particle starts to deviate
gives rise to the collection efficiency (and, from bubble (end of contact if at-
therefore, the recovery) passing through a tachment has not occurred during
maximum with increasing particle size. This the time t~)
is the usual observation, but is generally ex- ~ surface vorticity, u*/r* at r* = 1
plained by invoking particle/bubble detach- liquid viscosity
ment forces. The proposed model suggests an op particle density
alternative explanation. This will be explored
more fully in a subsequent paper. APPENDIX 2: DATA CORRELATION FOR
SURFACE VORTICITY ~s
5. CONCLUSIONS
The surface vorticity data of Woo (8) has
A model of particle sliding time for flotation been correlated to bubble Reynolds number
size bubbles has been developed based on the Reb and angle 0 measured from the front stag-
liquid velocity distribution on the upper sur- nation point. Woo reported ~s for the following
face of the bubble. This model fitted the ex- Reynolds numbers: 0.2, 0.5, 0.75, 1, 2, 3, 5,
perimental results of Schulze and Gottschalk. 10, 20, 30, 40, 100, 200, 300, and 400 and at
Their reported fit using the potential flow so- every 12 ° for Re < 40 and every 3 ° for ~s
lution (due to Sutherland) was traced to an > 100. The following correlations are for 0 °
error which overestimated sliding time by a < 0 < 90 °.
factor of 2. ~s = a + bO + cO2 + dO 3, [A-1]
Journal of Colloid and Interface Science, Vol. 109, No. 2, February 1986
498 DOBBY AND FINCH

where, for 20 < Reb < 400: 100 12 3.207 3.362 4.7
48 9.241 9.229 0.1
a = --0.01082 -- 7.273 X 10 -4 R e + 1.735 84 6.240 6.248 0.1

N 10 -6 R e 2 - 2.046 × 10 -9 R e 3 [A-2]

b = 0.0745 + 3.013 N 10 -3 R e - 7.402 T h ese are values o f surface vorticity for sin-
gle spheres, or single bubbles. T h e effect o f
N 10 -6 R e 2 + 8.931 × 10 -9 R e 3 [A-3]
b u b b l e - b u b b l e i n t e r a c t i o n u p o n ~s has b e e n
c = - 4 . 2 7 6 X 10 - 4 - 1.977 × 10 -5 R e calculated by LeCl ai r (13) an d LeClair's data
has b e e n s u m m a r i z e d t h r o u g h regression
+ 5.194 X 10 -8 R e 2 - 6.520 × 10 -11 R e 3
analysis for gas h o l d u p s ~ 2 6 % (12).
[A-4]

d = - 1 . 1 0 3 × 10 -6 - 1.032 × 10 -7 R e ACKNOWLEDGMENTS
The authors wish to thank Noranda Mines Ltd. for a
+ 1.397 X 10 -1° R e z - 1.334 × 10 -13 R e 3
scholarship for G. Dobby and the Natural Sciences and
[A-5] Engineering Council, Canada (operating and strategic
grants), Department of Energy, Mines and Resources,
and, for 0.2 < Reb < 20: Canada and Quebec Department of Education (FCAC
a = --1.217 × 10 -3 -- 1.745 × 10 -3 R e program), Canada for funds for research into column flo-
tation.
+ 5.143 × 10 -5 R e 2 - 1.165 × 10 -6 R e 3
[A-6] REFERENCES

b = 0.02859 + 9.229 × 10 -3 R e - 3.85 1. Sutherland, K. L., J. Phys. ColloidSci. 52, 394 (1948).
2. Weber, M. E., and Paddock, D., J. Colloid Interface
× 10 -4 R e 2 + 9.190 × 10 -6 R e 3 [A-7] Sci. 94, 328 (1983).
3. Cliff, R., Grace, J. R., and Weber, M. E., "Bubbles,
c = - 4 . 0 6 0 × 10 -5 - 5.857 × 10 -5 R e Drops, and Particles," Chaps. 5 and 7. Academic
+ 1.620 × 10 -6 R e 2 - 2.992 × 10 -8 R e 3 Press, New York, 1978.
4. Seeley, L. E., Hummel, R. L., and Smith, J. W., J.
[A-8] Fluid Mech. 68, 591 (1975).
5. Schulze, H. J., and Gottschalk, G., in "Proceedings,
d = - 9 . 6 1 0 × 10 -7 - 2.54 × 10 -7 R e 13th Int. Min. Proc. Cong. Warsaw, June 1979"
(J. Laskowski, Ed.), Part A, pp. 63-84. Elsevier,
+ 1.74 × 10 -8 R e 2 - 5.10 × 10 -1° R e 3
New York, 1981.
[A-9] 6. Jowett, A., in "Fine Particles Processing," Proc. Int.
Symp., Las Vegas, Nevada, Feb. 1980, Chap. 37.
In each o f the regressions, Eqs. [A-2] to (P. Somasundaran, Ed.), AIME, New York, 1980.
[A-9], the c o r r e l a t i o n coefficient is >0.99. 7. Kelly, E. G., and Spottiswood, D. J., "Introduction
S o m e c o m p a r i s o n s b e t w e e n W o o ' s values to Mineral Processing," p. 108. Wiley-Interscience,
New York, 1982.
a n d those calculated f r o m Eq. [A-1] are
8. Woo, S. W., Ph.D. thesis. McMaster University,
Hamilton, Ontario, 1971.
0 ~s by ~s by Eq. Difference
9. Schulze, H. J., "Physico-Chemical Elementary Pro-
Reb (degrees) Woo [A- 1] (%)
cesses in Flotation." Elsevier, Amsterdam, 1984.
10. Weber, M. E., J. Sep. Process Technol. 2, 29 (1981).
1.0 12 0.430 0.430 0
11. Concha, F., and Almendra, E. R., Int. J. Min. Proc.
48 1.452 1.437 1.0
5, 349 (1979).
84 1.754 1.743 0.6
12. Dobby, G. S., Ph.D. thesis. McGill University, Mon-
20 12 1.401 1.436 2.5 treal, Quebec, 1984.
48 4.176 4.111 1.5 13. LeClair, B. P., Ph.D. thesis. McMaster University,
84 3.596 3.544 1.5 Hamilton, Ontario, 1970.

Journal of Colloid and Interface Science, Vol. 109, No. 2, February 1986

You might also like