Taylor and Francis makes no representations or warranties as to the accuracy, completeness, or suitability for any purpose of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply of this article is strictly prohibited.
Original Description:
Original Title
The effect of bubble size on fine particle flotation.txt
Taylor and Francis makes no representations or warranties as to the accuracy, completeness, or suitability for any purpose of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply of this article is strictly prohibited.
Taylor and Francis makes no representations or warranties as to the accuracy, completeness, or suitability for any purpose of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply of this article is strictly prohibited.
This article was downloaded by: [McGill University Library]
On: 26 May 2014, At: 14:05
Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registere d office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Mineral Processing and Extractive Metallurgy Review: An International Journal Publication details, including instructions for authors and subscription informa tion: http://www.tandfonline.com/loi/gmpr20 The Effect of Bubble Size on Fine Particle Flotation R. H. YOON a & G. H. LUTTRELL a a Department of Mining and Minerals Engineering , Virginia Polytechnic Institute and State University , Blacksburg, Virginia, 24061, U.S.A Published online: 06 Apr 2007. To cite this article: R. H. YOON & G. H. LUTTRELL (1989) The Effect of Bubble Si ze on Fine Particle Flotation, Mineral Processing and Extractive Metallurgy Review: An International Journal, 5:1-4, 10 1-122, DOI: 10.1080/08827508908952646 To link to this article: http://dx.doi.org/10.1080/08827508908952646 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the informatio n (the Content) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or su itability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions a nd views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Conten t should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relati on to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions Mineral Processing and Extractive Metallurgy Review. 1989. Vol. 5. pp. 101-122 Reprints available directly from the publisher Photocopying permitted by license only 1989 Gordon and Breach. Science Publishers. Inc. Printed in Great Britain The Effect of Bubble Size on Fine Particle Flotation R. H. YOON and G. H. LUTIRELL Department of Mining and Minerals Engineering, Virginia Polytechnic Institute an d State University, Blacksburg, Virginia 24061 U.S.A. Expressions for the probability of collision (Pc) and adhesion (Pa) have been de rived for fine particle flotation by calculating the trajectory of particles as they flow past a bubble in streamline flow. Three different flow regimes have been considered in the present work. i.e .. Stokes. potential and intermediate. For the intermediate flow conditions in which most flotation opera tions are carried out, the particle trajectories have been determined using an empirical stream functio n derived in the present work. For the case of a very hydrophobic coal sample. the values of the probabil ity of collection (P) determined experimentally have been found to be in close agreement with the theo retically predicted P; values over a range of bubble and particle sizes. The expression for Pu has been derived by determining the time it takes for a pa rticle to slide along the surface of a bubble after collisioo. It has been assumed that the bubble-particl e adhesion occurs whcn the sliding time is equal to or exceeds the induction time, which varies with th e particle hydrophobicity. Pois shown to be a function of particle size, bubble size and induction time. Th e values of PQ predicted in the present work are in good agreement with the results of microfiotation tes ts conducted on a coal sample. INTRODUCTION As high-grade ore deposits are depleted. low-grade ores, which are usually fineg rained and complex, have to be mined and processed. Beneficiation of these lowgrade ores requires fine grinding to liberate the finely-disseminated valuable minerals from the waste rock, consuming a great deal of energy. Although the fro th flotation process developed at the turn of the century is still the best availab le technology for separating mineral fines, its efficiency deteriorates rapidly wit h decreasing particle size below approximately 10 ,um. Many investigators"? have delineated various reasons for this difficulty, including high reagent consumpti on, high rate of surface reactions. slime coating, morphological and surface chemica l changes during fine grinding, etc. One important factor that was noted was the unfavorable hydrodynamic conditions surrounding the bubbles generated in conventional flotation machines. It has been shown that when bubbles are too large, fine particles follow the streamline s around the bubbles due to their small inertial force and do not collide with the
lOt Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 102 R. H. YOON AND G. H. LUTIRELL bubbles. The consequence of this low probability of collision would be a low flotation rate and, hence, poor recovery. Earlier work on the hydrodynamics of bubble-particle adhesions'? suggested, however, that the probability of bubblepa rticle collision can be improved greatly when using smaller bubbles. It is the purpose of this chapter to address the importance of controlling the bubble size
during flotation. Bubble size is important in controlling not only the flotation
recovery but also the selectivity. Even the most hydrophobic particles require appropriate hydrodynamic conditions for bubble-particle adhesion to occur after the collision. BACKGROUND During the early stages of flotation research, most of the research efforts were
devoted to studies of reagent adsorption and surface wetting phenomena. Many of the pioneering concepts of flotation chemistry were developed during this period , particularly in the 1930's. It was not until the following decades that new theo ries and experimental evidence demonstrating the importance of the physical interacti ons in flotation began to emerge. The early experimental findings by Spedden and Hanna10, and later by Whelan and Brown1', provided direct photographic evidence for the important role of bubble-particle hydrodynamics in flotation. I t was the work done during this period that laid the foundation for our current understanding of the bubble-particle collision and adhesion mechanisms that occu r during flotation. Perhaps the first attempts to quantitatively describe the rate of flotation thro ugh physical interactions between particles and bubbles were those of Gaudin" and Schuhmann12, who considered the probability of bubble-particle collision to be a
fundamental parameter in determining the flotation rate. Their analyses were greatly hampered, however, by the lack of detailed information concerning the bubble-particle hydrodynamics. It was not until later that the first analytical relationships between the flotation rate and the hydrodynamics were derived by Sutherland" in his landmark article entitled "Kinetics of the Flotation Process. " In this article, Sutherland assumed that the probability of particle collection (P) by a bubble can be represented by, (1) where P; is the probability of collision between bubbles and particles, Pa is th e probability of adhesion after collision, and Pd is the probability that subseque nt detachment would not occur. Each of these probabilities was then functionalized in terms of bubble size and particle size by considering the case of a single parti cle interacting with an isolated bubble rising in the flotation pulp. He assumed tha t particles have no inertia and thus follow the streamlines around the bubbles, wh ich in turn are assumed to behave as non-deformable spheres. These assumptions' allowed him to determine those probabilities directly from hydrodynamic consider ations. It was unfortunate, however, that the predictions made using his Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 (2) THE EFFECT OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 103 theoretical analysis could not be supported by experimental data because he considered only the case of potential or inviscid flow. The assumption of potent ial flow is valid only for very high Reynolds numbers. Nonetheless, the basic concep ts outlined by Sutherland provided a framework upon which future investigators could build. Recognizing the shortfall of Sutherland's approach, Flint and Howarth 13 numeric ally solved the equations of motion for a particle approaching a bubble using the Navier-Stokes equations, and determined the probability of collision for mineral
flotation. This technique was similar to that employed by earlier investigators for the study of dust collection by raindrops'v". Flint and Howarth's':' analysis wa s later modified by Reay and Ratcliff 18, who derived their own expressions for effluent treatment. Using a numerical solution technique, they showed that proba bility of collision (Pc) could be given by, r, cc (~Y. in which Dp and Db, respectively, are the diameters of particles and bubbles. Th is relationship explains why the flotation recovery should decrease with decreasing
particle size and, at the same time, suggests that theoretically this problem ca n be alleviated by decreasing the bubble size. More recently, a quantitative expression for particle capture by a nondeformable
spherical object has been derived by Weber l 9 and Weber and Paddock?". In their
analysis, the probability of collision (PC> was determined analytically for the cases of low Reynolds numbers. For larger Reynolds numbers, the expression was modified by applying a curve-fitting technique to the numerical solutions to the
Navier-Stokes equations obtained by Masliyalr" and W0022 . In this manner, Weber and Paddock derived an expression for P; as follows: _ ~(P..e.)2[ (3/16)Re] P, -2 Db 1 + 1 + 0.249 Re0 .56 (3) in which Re is the Reynolds number of the bubble. Unlike previous studies, this work was the first of its kind in which a single expression can predict P; for a wide range of bubble and particle sizes. However, Weber and Paddock made no experimen tal measurements to support their analysis. A reduction in particle size not only leads to a decrease in flotation recovery, but it can also have an adverse impact on flotation selectivity. The poor selectivit y of fine particle flotation has long been recognized and has generally been attribut ed to many factors such as the high surface energy of fine particles, fine particle s stabilizing froth and nonselective entrainmcnrv". A more quantitative account of
this problem has been given by Sutherland". He considered that upon collision, a
particle slides over the bubble surface for a finite period of time, and this "s liding time" is determined by the velocity of the liquid flowing past the rising bubble
and the size of the bubble. In cases where bubble-particle attachment is thermod ynamically possible, adhesion occurs only when the sliding time is longer than the time required to thin and rupture the intervening film, known as the "disjoining
Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 104 R. H. YOON AND G. H. LUTTRELL film", between the bubble and the particle. The minimum time that is required to
thin and rupture the film is known as "induction time". The induction time can b e subdivided into i) the time to thin to a critical thickness (HJ, and ii) the tim e required for the thin film to spontaneously rupture to establish a three-phase contact forming a contact angle. It has been shown, however, that the time it takes for the thin film to rupture at the thickness, He' is on the order of 10-9 seconds, which is negligible compared to the time required for the film thinning process. The experimentally determined induction times reported in the literature are on the order of 10-2 seconds or greater. Thus, induction time refers essentially to the time required for the fi lm thinning and drainage process, which is determined by the hydrodynamics of the system. The value of He, on the other hand, is determined by the thermodynamics of the system. The technique for measuring induction time was first conceived by
Sven-Nillsorr'", and implemented by Eigeles and Volova?", Glcmbotsky'? and Evans
and Ewers" in mineral flotation systems. An improved version of Glernbotsky's'" apparatus for measuring induction time has been reported more recently by Yordan and Yoorr". The sliding/induction time concept provides a convenient framework for quantifyi ng the probability of bubble-particle adhesion (Pa ) . This type of analysis has recently been utilized by Dobby and Finch:" for modeling fine particle flotation . Although their model is identical conceptually to that proposed by Sutherland", the assumption of potential flow has now been eliminated and, as a result, the predictions are considerably more realistic. Two major findings have been made from the work of Dobby and Finch3o . First, their model suggests that the decrease in floatability with increasing particle size beyond the upper flotation limit is due to a decrease in Pa Previously, it was explained by an increase in the inertial force with increasing particle size, wh ich in turn increases the detachment force and, hence, decreases Pd. Secondly, the mode l suggests that P; for very small particles, e.g., less than 10.um, is very high a nd somewhat independent of the hydrophobic/hydrophilic nature of particle surfaces.
This finding implies that selectivity will be very poor when floating fine parti cles. However, many investigators/"?" have shown in experiments that ultrafine particles can be selectively recovered by flotation. MODEL DEVELOPMENT The process of collecting particles by bubbles generally takes place in a comple x and highly turbulent environment inside a flotation cell. It may be assumed, however, that as the distance separating the particle and the bubble decreases, conditions will develop which may to some extent be simulated by an unperturbed flow field. This would particularly be the case of pneumatic or column flotation
where no external agitation is employed. It is then possible to quantify some of the fundamental sub-processes occurring in flotation for a number of ideal and simplified cases. Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 (4) THE EFFECr OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 105 Bubble-Particle Collision Let us consider the case of an isolated bubble rising through a suspension of particles in a quiescent environment. As the liquid sweeps past the bubble, a fl ow pattern represented by an infinite series of streamlines develops. Streamlines a re defined so that a velocity vector of the fluid at every point on the line is tan gent to the streamline at a given instant. Arrangement of the streamlines depends on the
properties of the fluids involved and the geometry of the obstacle placed in the flow field. It is conceivable that small particles having negligible inertial force w ill follow the streamline. while large particles deviate from the streamline. Stokes number (St), which represents a ratio of inertial to drag forces, is usef ul for determining the tendency of a particle to adjust to changes in fluid flow. Particles having small values of 51 will follow the streamlines closely, while t hose having larger values of St will deviate from the streamlines. The following equa tion gives the Stokes number, 51 = !~(PE)2Re 9 (21 Db as a function of the Reynolds number (Re) of the bubble, particle diameter (Dp ) , bubble diameter (Db), particle density (Op) and liquid density (0,). As shown, the tendency of a particle to follow a streamline around the bubble depends on b oth bubble and particle sizes. It can be readily shown that for Db of 1 mm and Op of
2.6 gm/crrr', the inertial forces need not be considered for particles smaller t han approximately 50.um in diameter. If particle inertia can be neglected, one can then consider that particles follo w streamlines and calculate the probability of collision from the streamline patte rn. For mathematical convenience, a bubble can be considered stationary while the fluid is moving past the bubble at a velocity equal to the terminal rise velocit y of the bubble but in the opposite direction (Figure 1). Polar coordinates are used to l ocate a point in the fluid. The bubbles and particles under consideration are assumed to behave as rigid spheres, which is reasonable for bubbles less than approximately
1200.um in diameter" and for finely ground particles". The trajectory of a particle is considered to be determined by the streamline which passes through its center':', For obvious reasons, this assumption becomes
less valid as the ratio of particle to bubble diameter increases. Assuming that the streamlines come closest to the bubble at its equator, a grazing streamline is defined as the one passing through the distance of particle radius (Rp ) at the equator and the distance of Ro at an infinite distance from the bubble. It is cl ear, then, that of the particles located in the path of the bubble, only those within the limiting radius, Ro , will collide with the bubble. Those which lie outside this area will sweep past the bubble without having an opportunity for contact. The probab ility of collision, which is defined as the fraction of particles in the path of the bubble that actually collide with it, is therefore determined by the ratio of th e area (A o ) inscribed by the limiting radius (Ro ) to the area (A b ) inscribed by th e bubble radius (R b ) . Thus, the probability of collision (Pc) is given as: Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 106 R. H. YOON AND G. H. LUITRELL FIGURE 1 Polar coordinate system describing the trajectory of a particle moving past a bubble in streamline flow. (5) The value of Ro is unknown and must be determined from the mathematical description of the grazing streamline. The denominator of equation (5) should actually be (R b + Rp ) , but equation (5) holds when Rb ~Rp . Because of its importance in a wide range of disciplines, the study of fluid flo w around a spherical obstruction (such as a raindrop or a bubble) is a classical problem with extensive literature. Particularly detailed in the literature is th e case of low Reynolds numbers for which analytical solutions to the Navier-Stokes and continuity equations are possible. For the case of higher Reynolds numbers, seve ral approximate numerical solutions have also been presented40-45 . Flow conditions 'represeritative of very high Reynolds numbers (inviscid flow) can be approximat ed by the potential flow solution to the Navier-Stokes equations. Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 THE EFFECf OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 107 Stokes flow conditions For cases where Reynolds number is much less than unity, the streamlines are described by the well-known Stokes stream function (l/J), wh ich may be expressed mathematically as, (6) where Ub is the bubble rise velocity, () and r are the angular and radial coordi nates, respectively, as defined in Figure 1, and x = rlR b . At a distance far ahead of the bubble, sin /I = Rolr, which can be substituted into equation (6) to yield, R~= 21/1lub' (7) (8) (10) (11) (12) (14) as R approaches infinity. By combining equations (5), (6), and (8), the probabil ity of particle collision (Pc) for Stokes flow is obtained as follows: p =.J::L = sin2 /1 (x2 -~X + -l) (9) c ubRl 2 2x' For the limiting case of the particle just grazing off the bubble equator, r = Rp + Rb X = 1 + RplRb, and /I = 90. Substituting equations (11) and (12) into equation (9), one can obtain, P, = H~r(13) which is in identical form as the one obtained by Gaudin'". Equation (13) shows that the probability of collision for Stokes flow conditions varies as the squar e of the RplRb ratio. Potential flow conditions For the potential flow conditions (very large Reynolds
numbers), the stream function is given by the following equation: 1/1 = ubRlsin2/1 [~X2-~J. Using the same procedure employed for the Stokes flow conditions, one can obtain
P; for potential flow conditions as follows, _ (!!e.)P, -3 R b ' (15) which is of the same form as derived by Sutherland". As shown, the probability o f collision varies linearly with the RplR b ratio as opposed to a squared relation ship Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 108 R. H. YOON AND G. H. LUTIRELL (16) (18) for Stokes flow. This means that the particle-to-bubble size ratio has a much mo re significant effect on flotation in Stokes flow regions than in potential flow re gions. Intermediate flow conditions The preceeding analyses are useful for determining collision probabilities at extremely small and large Reynolds numbers. However, the range of Reynolds numbers encountered for the sizes of bubbles typically employed during flotation fall into a range between 0.2 to 100. As a result, nei ther of the previous solutions is appropriate for describing the flotation process. Therefore, a new stream function for intermediate Reynolds numbers (l/J) has been developed in the present investigation by combining the Stokes and potentia l flow equations, i.e., equations (6) and (14), respectively. These two equations are special cases of the following expression, 2 . 2 [I 2 3 1 3aJ1jJ = UhRhSIn e -x --ax --+2 4 2x 4x where a is a dimensionless parameter which varies between 0 and 1. If the value of a is set equal to 0, equation (16) is reduced to equation (14), while if a is se t equal to I. it becomes equation (6). Thus, the values of a between 0 and 1 represent c ases of intermediate Reynolds numbers. I n order to determine the values of a as a function of Re for intermediate flow
conditions, streamline patterns around a spherical obstruction have been collect ed from literature for a wide range of Reynolds numbers/! .40.47-51. Far ahead of t he spherical obstruction, a given streamline can be uniquely identified by a fixed value of Ro . Therefore, by directly measuring the values of Ro from the streamline patterns obtained at a given Reynolds number and knowing the values of bubble rise velocity (Uh). one can determine l/J for each streamline using equation (8) . Once the values of l/J are known, the values of a can then be determined from equation (16) for any combination of rand O. The a values obtained as such are plotted as a function of (I -IIx) for Reynolds numbers of 1. 10, 50 and 100, as Shown in Figure 2. It was found that the a values can be normalized with respect to Reynolds number using the relationship: a = 1 -4Re 72 [I -.!.] (17) 45 x for Reynolds numbers of 0 to 100. Equation (17) may also be applicable for Re > 100, although no experimental (streamline) data have been available in the present work. Substitution of equation (17) into equation (16) yields a generalized stream function for intermediate Reynolds numbers as follows: ," = UR 2sin2e [.!.x2~ x+ 1-+ Re 72 ( "; _.!. + X-I)] 'r h h 2 4 4x 15 xx Note here that as the Reynolds number approaches zero, equation (18) is reduced to equation (6), which is for Stokes flow conditions. In Figure 3, each of the stream functions, i.e., equations (6), (14) and (16), h as Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 THE EFFECT OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 109 0.60.5 o 0.40.30.20.1 0.0 L-_---J__-----'__-----'-__-----"__---- L__---J 0.0 oRe = 1 eRe = 10 .. Re = 50 Re = 100
I 0.. <i 1-1/X FIGURE 2 a versus I -I1X plot at various Reynolds numbers. 8=60 z o I-0.5 r---.----....-----,-----r.,------, U Z ::> u, 0.4 ::E <r w 0:: 0.3 I-en ~0.2 W ...J Z 00.1 en Z w ::E zs FIGURE 3 Dimensionless stream function versus dimensionless radial distance for a range of Reynolds numbers at () = 600. Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECT OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 109 0.60.5 o 0.40.30.20.1 0.0 L-_---J__-----'__-----'-__-----"__---- L__---J 0.0 oRe = 1 eRe = 10 .. Re = 50 Re = 100
I 0.. <i 1-1/X FIGURE 2 a versus I -I1X plot at various Reynolds numbers. 8=60 z o I-0.5 r---.----....-----,-----r.,------, U Z ::> u, 0.4 ::E <r w 0:: 0.3 I-en ~0.2 W ...J Z 00.1 en Z w ::E zs FIGURE 3 Dimensionless stream function versus dimensionless radial distance for a range of Reynolds numbers at () = 600. Downloaded by [McGill University Library] at 14:05 26 May 2014 110 R. H. YOON AND G. H. LUlTRELL been made dimensionless by dividing it with ubR~and plotted as a function of dimensionless radial distance from the bubble surface, i.e., R/Rb -1, for compar ison. It shows that as the Reynolds number increases there is an upward shift of the stream function toward that of potential flow. From equation (18), an expression for P; applicable for intermediate Reynolds numbers can now be derived. Using the same procedure as employed for Stokes and potential flow conditions, it can be readily shown that, = [l 4Re. 72 ](I!)2r. 2 + 15 R b ' (19) Equation (19) shows that P; varies with R~and inversely with D~,as is the case with Stokes flow conditions (equation (13)). However, equation (19) is also a function of the Reynolds number of the bubble, so that P; does not vary exactly as the inverse square of the bubble size. It can be shown that for bubbles of very large Reynolds numbers, r, oc un, and for bubbles of small Reynolds numbers, r, oc I/R~. (20) The three expressions for P; applicable in different Reynolds number regions, i.e., equations (13), (15) and (19), can be summarily represented by the followi ng single expression: 800600400200 -4 L-_......L.-_........__........__.......__....... o 1000 0 Dp =10jJm -I -~ -2--t9 s -3 BUBBLE DIAMETER (J.lm) FIGURE 4 Relationship between P< and Do for potential, intermediate and Stokes f low conditions for Dp = 10I'm. Downloaded by [McGill University Library] at 14:05 26 May 2014 110 R. H. YOON AND G. H. LUlTRELL been made dimensionless by dividing it with ubR~and plotted as a function of dimensionless radial distance from the bubble surface, i.e., R/Rb -1, for compar ison. It shows that as the Reynolds number increases there is an upward shift of the stream function toward that of potential flow. From equation (18), an expression for P; applicable for intermediate Reynolds numbers can now be derived. Using the same procedure as employed for Stokes and potential flow conditions, it can be readily shown that, = [l 4Re. 72 ](I!)2r. 2 + 15 R b ' (19) Equation (19) shows that P; varies with R~and inversely with D~,as is the case with Stokes flow conditions (equation (13)). However, equation (19) is also a function of the Reynolds number of the bubble, so that P; does not vary exactly as the inverse square of the bubble size. It can be shown that for bubbles of very large Reynolds numbers, r, oc un, and for bubbles of small Reynolds numbers, r, oc I/R~. (20) The three expressions for P; applicable in different Reynolds number regions, i.e., equations (13), (15) and (19), can be summarily represented by the followi ng single expression: 800600400200 -4 L-_......L.-_........__........__.......__....... o 1000 0 Dp =10jJm -I -~ -2--t9 s -3 BUBBLE DIAMETER (J.lm) FIGURE 4 Relationship between P< and Do for potential, intermediate and Stokes f low conditions for Dp = 10I'm. Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECT OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 111 TABLE I Values of n and A determined for Stokes, potential and intermediate flow conditi ons. Flow Conditions Stokes Intermediate (Present Work) Intermediate (Weber and Paddock. 1983) Potential A 3/2 3 4Reo." -+-2 15 ~[1 + _o.::(3::..;/1:.::6t..:.)R.:.:ec:-::::] 2 1 + O.249Retl 5 6 3 n 2 2 2 (22)r, = A(~rin which A and n are the parameters which vary depending on the Rey nolds number. Table I gives the values of A and n for the three different flow regimes
considered in the present work, i.e., Stokes, potential and intermediate flow. A lso shown in this table are the values of A and n obtained by Weber and Paddock". For the flow conditions pertinent to most flotation operations, n is equal to 2. The values of A are 3/2 and 3 for Stokes and potential flow conditions, respectively , while they vary with the Reynolds number of the bubble for the intermediate flow
conditions. Figure 4 compares the P; values for lO-,um particles calculated as a function of
bubble size using the expressions given in Table I. The Stokes flow solution agr ees with the solutions for the intermediate flow range only for the case of very sma ll bubbles. The potential flow solution obtained by Sutherland" is far from those o f the intermediate flow cases considered in the present work. The dashed line repr esents equation (3) derived by Weber and Paddock?". As shown, there is agreement between Weber and Paddock's predictions and those of the present work despite the differences in their functional forms. Bubble-Particle Adhesion Clearly, not all particles which collide with bubbles become attached. Otherwise , no separation would be possible by flotation. Qualitatively, only those particle s that are hydrophobic enough attach themselves to the bubble by forming a three-phase contact with a finite contact angle. To be able to accurately predict flotation response, it is therefore necessary to evaluate the probability of adhesion in addition to the P; term that has been considered in the foregoing section. For t he small particles considered in the present work, detachment need not be considere d', i.e., P" = 1. As soon as a particle collides with a bubble, it begins to slide over the surfac e of the bubble and resides on it for a finite period of time, which is generally ref erred to as the "sliding" time. The magnitude of the sliding time is determined by the
velocity at which the liquid sweeps the particle along the bubble surface. Durin g Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECT OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 111 TABLE I Values of n and A determined for Stokes, potential and intermediate flow conditi ons. Flow Conditions Stokes Intermediate (Present Work) Intermediate (Weber and Paddock. 1983) Potential A 3/2 3 4Reo." -+-2 15 ~[1 + _o.::(3::..;/1:.::6t..:.)R.:.:ec:-::::] 2 1 + O.249Retl 5 6 3 n 2 2 2 (22)r, = A(~rin which A and n are the parameters which vary depending on the Rey nolds number. Table I gives the values of A and n for the three different flow regimes
considered in the present work, i.e., Stokes, potential and intermediate flow. A lso shown in this table are the values of A and n obtained by Weber and Paddock". For the flow conditions pertinent to most flotation operations, n is equal to 2. The values of A are 3/2 and 3 for Stokes and potential flow conditions, respectively , while they vary with the Reynolds number of the bubble for the intermediate flow
conditions. Figure 4 compares the P; values for lO-,um particles calculated as a function of
bubble size using the expressions given in Table I. The Stokes flow solution agr ees with the solutions for the intermediate flow range only for the case of very sma ll bubbles. The potential flow solution obtained by Sutherland" is far from those o f the intermediate flow cases considered in the present work. The dashed line repr esents equation (3) derived by Weber and Paddock?". As shown, there is agreement between Weber and Paddock's predictions and those of the present work despite the differences in their functional forms. Bubble-Particle Adhesion Clearly, not all particles which collide with bubbles become attached. Otherwise , no separation would be possible by flotation. Qualitatively, only those particle s that are hydrophobic enough attach themselves to the bubble by forming a three-phase contact with a finite contact angle. To be able to accurately predict flotation response, it is therefore necessary to evaluate the probability of adhesion in addition to the P; term that has been considered in the foregoing section. For t he small particles considered in the present work, detachment need not be considere d', i.e., P" = 1. As soon as a particle collides with a bubble, it begins to slide over the surfac e of the bubble and resides on it for a finite period of time, which is generally ref erred to as the "sliding" time. The magnitude of the sliding time is determined by the
velocity at which the liquid sweeps the particle along the bubble surface. Durin g Downloaded by [McGill University Library] at 14:05 26 May 2014 112 R. H. YOON AND G. H. LUTIRELL \ SLIDING 'rDISTANCE FIGURE 5 Polar coordinate system used in determining the critical angle of incid ence and probability of adhesion. this time period, the intervening film of water must be thinned and ruptured if the particle is to be attached to the surface of the bubble and eventually collected
in the froth phase. The sweeping action of the particle over the bubble surface has been verified experimentally and is well documented". A limited number of experimentally measured sliding times have also been reported by Schulze and Gottschalk". Bubble-particle attachment should occur when the sliding time is longer than the
induction time", which is defined as the time required for the film thinning pro cess. Therefore, the particle must slide a finite distance over the bubble surface bef ore the attachment occurs. For a given bubble and particle size, the distance travel led by a particle along the surface of a bubble is a function of the angle of incide nce (6;) at which the particle strikes the bubble. Only when 6; is smaller than the limit ing angle (60 ) will the particle have a sliding time longer than the induction time and become attached. The probability of bubble-particle adhesion (Pa) can now be defined as the fraction of particles in the path of the bubble that actually adhere to the bubb le. In reference to Figure 5, Po should be the ratio of the area inscribed by the limit ing radius (Rol to the area inscribed by the sum of the bubble and particle radii (R; + R,,). Using this concept, P~Ican be mathematically expressed as, r, = (Rh :~R p ) 2 (23) Downloaded by [McGill University Library] at 14:05 26 May 2014 112 R. H. YOON AND G. H. LUTIRELL \ SLIDING 'rDISTANCE FIGURE 5 Polar coordinate system used in determining the critical angle of incid ence and probability of adhesion. this time period, the intervening film of water must be thinned and ruptured if the particle is to be attached to the surface of the bubble and eventually collected
in the froth phase. The sweeping action of the particle over the bubble surface has been verified experimentally and is well documented". A limited number of experimentally measured sliding times have also been reported by Schulze and Gottschalk". Bubble-particle attachment should occur when the sliding time is longer than the
induction time", which is defined as the time required for the film thinning pro cess. Therefore, the particle must slide a finite distance over the bubble surface bef ore the attachment occurs. For a given bubble and particle size, the distance travel led by a particle along the surface of a bubble is a function of the angle of incide nce (6;) at which the particle strikes the bubble. Only when 6; is smaller than the limit ing angle (60 ) will the particle have a sliding time longer than the induction time and become attached. The probability of bubble-particle adhesion (Pa) can now be defined as the fraction of particles in the path of the bubble that actually adhere to the bubb le. In reference to Figure 5, Po should be the ratio of the area inscribed by the limit ing radius (Rol to the area inscribed by the sum of the bubble and particle radii (R; + R,,). Using this concept, P~Ican be mathematically expressed as, r, = (Rh :~R p ) 2 (23) Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFEcr OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 113 or, (24) (25) (26) (27) The value of the limiting angle (80 ) is determined by the magnitude of the tangential velocity of the particle as it slides over the bubble surface. The ve locity of a small particle can be assumed to be the same as that of a streamline passin g through its center!". Therefore, the sliding velocity of a particle is determine d by the tangential velocity of the streamline (u ,) at a distance of one particle ra dius from the bubble surface. The particle sliding time (t s ) can then be calculated by dividing the arc length travelled by the particle by the velocity of the particl e, i.e., J "12 R b + R t, = PdO. 0 0 Ur An upper integration limit of 7[/2 is used since the streamline flow will move a
particle away from the bubble surface after it passes the bubble equator. For axisymmetric flow around a rigid sphere, the instantaneous tangential velocity of the fluid (u ,) can be determined from the stream function (1/J) as follows, UI=_I_~ Rsin 0 dr in which r is the radial distance between the center of the bubble and the strea mline in question and 8 is the angular coordinate (see Figure 1). As has been discussed in the foregoing paragraph, the choice of an appropriate stream function depends on the flow conditions characterized by the Reynolds number. The stream functions for the Stokes, intermediate and potential flow conditions are given by equations (6), (18) and (14), respectively. Substituting
these stream functions into equation (26), one can obtain u, of the fluid at any point in the vicinity of the bubble. Substituting equation (26) into equation (25), followed by integration, one can obtain an expression for t s which will become a function of 80 , Solving it for 00 , and substituting it into equation (24), an expression for P" can be obtained. Th e ts term in this expression can be substituted directly by induction time (t;) since Is = ti when 8j = 00The expressions for Pa have been derived in this manner for the three different flow conditions and are given as follows: P _ . 2[2 {-3Ubli } ] u -SIn arctan exp 2R b(Rb/Rp + 1) for the Stokes flow conditions, _ 2 [ r-(45 + 8Re.72) Uh1i}] P; -SIn 2 arctan exp l 30R b(Rb/Rp + 1) for the intermediate flow conditions, and P _ . 2[2 { -3Ubl; }] Q -SIn arctan exp 2( R p + R b ) (28) (29) Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 114 R. H. YOON AND G. H. LUTTRELL for the potential flow conditions. Since each of these expressions is in the for m of a sine function, the value of Pa falls between zero and unity. This is an improvem ent over the numerical solutions obtained by Dobby and Finch30 whose P; values exceed 1 when the bubble size is very small or the induction time is short. A sa mple calculation for using equation (28) may be given here: a weakly floating particl e whose t, = 100 msec and D" = lO.um will have its Pa = 0.08. If t, is reduced to 10 msec by increasing the collector addition, for example, Pa will increase to 0 .98. To demonstrate the utility of equation (28), which is more applicable than the other two to most flotation practice, P; is plaited in Figure 6 as a function of
particle size (D,,) for the case of 1000-.um diameter bubbles and particles with
induction time (I;) values of 10, 20, 30 and 50 msec. As expected, the smaller t he induction time, the higher the value of Pa for a given particle size. More importantly, P" is shown to increase with decreasing particle size, which is a r esult of the lower tangential velocity (ut ) and hence the longer sliding time for sma ller particles. Thus, equation (28) suggests that small particles become attached mor e readily than large particles once the collision has taken place. Db' 1000 micronsc 0 (f) Q) -C "D ~ '+0 P 0.4 ...0 0 ...0 0 0.2 L 0.... 0.0 a 20 40 60 80 100 Particle Diameter (Microns) FIGURE 6 Effect of particle size on the probability of adhesion for induction ti mes of 10. 20, 30 and 50 msec. Downloaded by [McGill University Library] at 14:05 26 May 2014 114 R. H. YOON AND G. H. LUTTRELL for the potential flow conditions. Since each of these expressions is in the for m of a sine function, the value of Pa falls between zero and unity. This is an improvem ent over the numerical solutions obtained by Dobby and Finch30 whose P; values exceed 1 when the bubble size is very small or the induction time is short. A sa mple calculation for using equation (28) may be given here: a weakly floating particl e whose t, = 100 msec and D" = lO.um will have its Pa = 0.08. If t, is reduced to 10 msec by increasing the collector addition, for example, Pa will increase to 0 .98. To demonstrate the utility of equation (28), which is more applicable than the other two to most flotation practice, P; is plaited in Figure 6 as a function of
particle size (D,,) for the case of 1000-.um diameter bubbles and particles with
induction time (I;) values of 10, 20, 30 and 50 msec. As expected, the smaller t he induction time, the higher the value of Pa for a given particle size. More importantly, P" is shown to increase with decreasing particle size, which is a r esult of the lower tangential velocity (ut ) and hence the longer sliding time for sma ller particles. Thus, equation (28) suggests that small particles become attached mor e readily than large particles once the collision has taken place. Db' 1000 micronsc 0 (f) Q) -C "D ~ '+0 P 0.4 ...0 0 ...0 0 0.2 L 0.... 0.0 a 20 40 60 80 100 Particle Diameter (Microns) FIGURE 6 Effect of particle size on the probability of adhesion for induction ti mes of 10. 20, 30 and 50 msec. Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECf OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 115 1.0 C 0 (f) ill ...c -0 <t: ~ 0 z;-. - ...0 0 ...0 0 L Q.. 400 800 1200 1600 2000 Bubble Diameter (Microns) FIGURE 7 Effect of bubble size on the probability of adhesion for induction time s of 10, 20, 30 and 50 msec. In Figure 7, equation (28) is plotted as a function of bubble size (Db) for the case of floating 20-,um particles having I; values of 10, 20, 30 and 50 msec. As has been shown to be the case in Figure 6, a decrease in Ii results in an increase in Pa , which requires no further explanation. At a given Ii, P; is shown to increase with decreasing Db until a bubble size of approximately 350,um is reached. This is du e to the fact that as the bubble size is reduced, the bubble rise velocity (Ub) is also reduced, which in turn results in' an increased sliding time (Is)' A further dec rease in Di; however, causes a decrease in the sliding distance, which results in a de crease in t, and hence a decreased Pa . The changes in Pa with particle and bubble size have important implications regarding flotation selectivity. Let us consider strongly and weakly floating (i .e., middling) particles having induction times of 10 msec and 100 msec, respectively , and calculate the P; values using equation (28) as a function of particle size u sing two different bubble sizes, i.e., 100 and 1000 ,urn in diameter. As a means of quantifying selectivity by means of Pa , a "selectivity index" is defined as fol lows: (30) Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECf OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 115 1.0 C 0 (f) ill ...c -0 <t: ~ 0 z;-. - ...0 0 ...0 0 L Q.. 400 800 1200 1600 2000 Bubble Diameter (Microns) FIGURE 7 Effect of bubble size on the probability of adhesion for induction time s of 10, 20, 30 and 50 msec. In Figure 7, equation (28) is plotted as a function of bubble size (Db) for the case of floating 20-,um particles having I; values of 10, 20, 30 and 50 msec. As has been shown to be the case in Figure 6, a decrease in Ii results in an increase in Pa , which requires no further explanation. At a given Ii, P; is shown to increase with decreasing Db until a bubble size of approximately 350,um is reached. This is du e to the fact that as the bubble size is reduced, the bubble rise velocity (Ub) is also reduced, which in turn results in' an increased sliding time (Is)' A further dec rease in Di; however, causes a decrease in the sliding distance, which results in a de crease in t, and hence a decreased Pa . The changes in Pa with particle and bubble size have important implications regarding flotation selectivity. Let us consider strongly and weakly floating (i .e., middling) particles having induction times of 10 msec and 100 msec, respectively , and calculate the P; values using equation (28) as a function of particle size u sing two different bubble sizes, i.e., 100 and 1000 ,urn in diameter. As a means of quantifying selectivity by means of Pa , a "selectivity index" is defined as fol lows: (30) Downloaded by [McGill University Library] at 14:05 26 May 2014 116 R. H. YOON AND G. H. LUlTRELL 1.0 I I I /' I I I O.B r-I - X I Q) I U I C 0.6 f-I - 1 >, I......., I > I......., 0.4 l-I - U Q) I -IQ) --1000 microns(f) I 0.2 1-1 -100 microns - 1 I 0.0 II' I I I I 0 20 40 60 BO 100 Particle Diameter (Microns) FIGURE 8 Effect of particle size on the selectivity index for bubbles of 100 ,ur n and 1000,urn in diameter. in which the superscripts sand w refer to strongly and weakly floating particles . The larger the value of SI, the higher the efficiency of the flotation separation wi ll be. The results given in Figure 8 show that SI remains high until the particle size is reduced to approximately 40,um. Below this, the SI decreases very rapidly with decreasing particle size and eventually passes through zero for an infinitely sm all particle. This finding is similar to, those of Dobby and Finch 30 who also showe d that selectivity diminishes with decreasing particle size. Thus, to counter the effec t of deteriorating selectivity with decreasing particle size, it would be necessary f or the induction time difference between the hydrophobic and the hydrophilic particles to be increased by chemical additives. It should be noted, however, that as the par ticle size is reduced, the population of composite (or middling) particles will be red uced, which in turn will increase the induction time difference between the floatable and the non-floatable particles, thereby increasing the separation efficiency. Figur e 7 also shows that the difference between using 100-or lOOO-pm bubbles is minimal, although 1DO-pm bubbles tend to show a slightly higher selectivity. Downloaded by [McGill University Library] at 14:05 26 May 2014 116 R. H. YOON AND G. H. LUlTRELL 1.0 I I I /' I I I O.B r-I - X I Q) I U I C 0.6 f-I - 1 >, I......., I > I......., 0.4 l-I - U Q) I -IQ) --1000 microns(f) I 0.2 1-1 -100 microns - 1 I 0.0 II' I I I I 0 20 40 60 BO 100 Particle Diameter (Microns) FIGURE 8 Effect of particle size on the selectivity index for bubbles of 100 ,ur n and 1000,urn in diameter. in which the superscripts sand w refer to strongly and weakly floating particles . The larger the value of SI, the higher the efficiency of the flotation separation wi ll be. The results given in Figure 8 show that SI remains high until the particle size is reduced to approximately 40,um. Below this, the SI decreases very rapidly with decreasing particle size and eventually passes through zero for an infinitely sm all particle. This finding is similar to, those of Dobby and Finch 30 who also showe d that selectivity diminishes with decreasing particle size. Thus, to counter the effec t of deteriorating selectivity with decreasing particle size, it would be necessary f or the induction time difference between the hydrophobic and the hydrophilic particles to be increased by chemical additives. It should be noted, however, that as the par ticle size is reduced, the population of composite (or middling) particles will be red uced, which in turn will increase the induction time difference between the floatable and the non-floatable particles, thereby increasing the separation efficiency. Figur e 7 also shows that the difference between using 100-or lOOO-pm bubbles is minimal, although 1DO-pm bubbles tend to show a slightly higher selectivity. Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECf OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 117 MODEL VALIDATION Two series of experiments were undertaken to establish the validity of the preceding theoretical analyses. In the first set of tests, the probability of co llection (P) was determined for a single bubble rising through a suspension of very hydrophobic coal particles. The value of P determined as such may closely approx imate the probability of collision (Pc) since the probability of adhesion (Pa) for very hydrophobic particles should approach unity. In the second set of tests, Pa
values were determined from experimentally measured values of induction time (li ) for a run-of-mine coal sample, and compared with microflotation results. The particles used in the flotation experiments were so large that the probability o f collision (Pc) may be considered unity. This will allow a direct comparison betw een Pa and floatability. Validation of the Probability of Collision Model Figure 9 shows the results of probability of collection (P) measurements conduct ed on the Buller seam coal, New Zealand, as a function of bubble size. This coal z 0 l- t> W ....J ....J 0 0.1 t> u, 0 >I0.01 ....J CD <t CD 6 0a:: a.. o 500 600 BUBBLE DIAMETER (jJm) FIGURE 9 Comparison of experimental probability of collection (points) versus th e probability of collision determined using Equation (19) for different bubble and particle sizes . Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECf OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 117 MODEL VALIDATION Two series of experiments were undertaken to establish the validity of the preceding theoretical analyses. In the first set of tests, the probability of co llection (P) was determined for a single bubble rising through a suspension of very hydrophobic coal particles. The value of P determined as such may closely approx imate the probability of collision (Pc) since the probability of adhesion (Pa) for very hydrophobic particles should approach unity. In the second set of tests, Pa
values were determined from experimentally measured values of induction time (li ) for a run-of-mine coal sample, and compared with microflotation results. The particles used in the flotation experiments were so large that the probability o f collision (Pc) may be considered unity. This will allow a direct comparison betw een Pa and floatability. Validation of the Probability of Collision Model Figure 9 shows the results of probability of collection (P) measurements conduct ed on the Buller seam coal, New Zealand, as a function of bubble size. This coal z 0 l- t> W ....J ....J 0 0.1 t> u, 0 >I0.01 ....J CD <t CD 6 0a:: a.. o 500 600 BUBBLE DIAMETER (jJm) FIGURE 9 Comparison of experimental probability of collection (points) versus th e probability of collision determined using Equation (19) for different bubble and particle sizes . Downloaded by [McGill University Library] at 14:05 26 May 2014 118 R. H. YOON AND G. H. LUTIRELL contained only 0.13% ash and was very hydrophobic as mined. The apparatus and experimental procedure used in the determination of P have been described elsewherc'". Experiments were carried out using samples having 40.1-, 31.0-and 11.4-pm mean diameters. The solid lines in this figure represent the theoretical
probability of collision (Pc) calculated using equation (19). As has already bee n noted, the probability of collection (P) measured in experiment should be equal to Pc, if P; is assumed to be unity for this particular coal sample. Both the experimental P and the calculated P; values show a drastic increase as the bubbl e diameter is reduced. Considering the log-scale used for plotting P, the benefits of using small bubbles for fine particle flotation are quite substantial. Ingeneral , there is excellent agreement between the theory and the experiment. Note, however, that the experimental P values are considerably higher than predicted when collecting large particles with large bubbles. This discrepancy may be attribute d to the likelihood that the large particles deviate from the streamlines due to iner tial force. The lack of fit observed when using small bubbles to collect the 11.4-pm particles may be due to some sort of surface force which becomes more predominan t as particle size becomes very small. 100 1.0 c Recovery o Po C() Induction Time 0 8 ,.-.... 80 (fJ (fJ EQ) ,.-.... J:: <:: ~-0 Q)'-../ 6 >, '+E L 0 i-=Q) > J:;C0 4 0U :;::; Q) .0 Uo: 0 :J .0 2 -00 CL 0.... a 0.0 a a 5 10 15 20 25 Oxidation Time (Days) FIGURE 10 Effect of oxidation time on recovery induction time and probability of adhesion predicted using Equation (28). Downloaded by [McGill University Library] at 14:05 26 May 2014 118 R. H. YOON AND G. H. LUTIRELL contained only 0.13% ash and was very hydrophobic as mined. The apparatus and experimental procedure used in the determination of P have been described elsewherc'". Experiments were carried out using samples having 40.1-, 31.0-and 11.4-pm mean diameters. The solid lines in this figure represent the theoretical
probability of collision (Pc) calculated using equation (19). As has already bee n noted, the probability of collection (P) measured in experiment should be equal to Pc, if P; is assumed to be unity for this particular coal sample. Both the experimental P and the calculated P; values show a drastic increase as the bubbl e diameter is reduced. Considering the log-scale used for plotting P, the benefits of using small bubbles for fine particle flotation are quite substantial. Ingeneral , there is excellent agreement between the theory and the experiment. Note, however, that the experimental P values are considerably higher than predicted when collecting large particles with large bubbles. This discrepancy may be attribute d to the likelihood that the large particles deviate from the streamlines due to iner tial force. The lack of fit observed when using small bubbles to collect the 11.4-pm particles may be due to some sort of surface force which becomes more predominan t as particle size becomes very small. 100 1.0 c Recovery o Po C() Induction Time 0 8 ,.-.... 80 (fJ (fJ EQ) ,.-.... J:: <:: ~-0 Q)'-../ 6 >, '+E L 0 i-=Q) > J:;C0 4 0U :;::; Q) .0 Uo: 0 :J .0 2 -00 CL 0.... a 0.0 a a 5 10 15 20 25 Oxidation Time (Days) FIGURE 10 Effect of oxidation time on recovery induction time and probability of adhesion predicted using Equation (28). Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECT OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 119 Validation of the Probability of Adhesion Model A run-ot-rruue Elkhorn seam coal, Virginia, having a feed ash of IS.S'Y" was use d for both the microflotation and induction time measurements. Flotation tests wer e conducted using a 22-mm diameter Partridge and Smith'" type microflotation cell.
The induction time apparatus employed in these experiments has been described elsewhere by Yordan and Yoorr", and is similar to that used by Eigeles and Votova"'. Figure 10 shows the results of the microflotation tests and the induction time measurements carried out on the coal sample aged (or oxidized) in water at an ambient temperature for varied lengths of time. In general, a prolonged oxidatio n resulted in an increase in induction time and a corresponding decrease in floata bility. When the coal sample was fresh, the recovery was approximately 70%, while after 20 days of oxidation, the floatability was reduced to approximately 20%. Also shown in Figure 10 is the probability of adhesion (Po) predicted using
equation (28). The predictions were made by substituting the measured induction times into the equation along with the experimental variables, such as particle size and bubble size, employed in the flotation experiments. As shown, the Po values calculated as such decrease rapidly as the induction time increases and s how a reasonable agreement with the experimental floatabilities. The calculated Po values are considerably lower than the fractional floatabilities (% floatability
divided by 100), but the trend is the same. If the flotation time had been short er than actually employed (1 minute), the calculated Po values would have been more
closely matched by the flotation results. As another example, a series of microflotation experiments were conducted on the Elkhorn seam coal at various pH values, and the results are compared in Figu re 11 with the P; values calculated from the induction times measured concurrently and the particle and bubble sizes employed in the flotation experiments. The floatability is shown to reach a maximum at approximately pH 8, where the induction time is minimum. Although the predicted Po values are significantly higher than the fractional floatabilities, the Po values calculated using equati on (28) can predict the trend and the maximum flotation pH reasonably well. In this case , the predicted Po values would have been closer to the flotation results if the flotation time had been longer than actually employed. The results given in Figu res 10 and 11 demonstrate that results of microflotation experiments can be predicte d from induction time, particle size and bubble size, and by considering the flota tion time as an adjustable parameter. SUMMARY AND CONCLUSIONS Particle collection during flotation has been divided into two elementary subpro cesses, i.e., bubble-particle collision and adhesion. Detailed hydrodynamic analysis of the bubble-particle interaction resulted in the development of funct ional forms for probability of collision (Pc) and probability of adhesion (Po)' This w as Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 120 R. H. YOON AND G. H. LUITRELL Q) E i-= c o :..;::; U ::J -0 C 5 4 3 2 o c o (f) Q) ...c -0 <{ ...0 o ...0 o l.... 0.. 10a642 6. Induction TIme c Recovery o Po o L-_---l__--'-__----'-__-"-__--L__0.0 o 12 ao 1OOr---r---r---,..---,..---,..-----, 1.0 20 c Q) > o 40 U Q) 0:::: pH FIGURE II Effect of pH on recovery, induction time and probability of adhesion p redicted using Equation (28). made possible by deriving a stream function that can describe the streamlines around a bubble in the intermediate Reynolds number range, e.g., 0-100. The stream function has been derived using a curve-fitting technique for the streaml ine patterns available in literature. In deriving an expression for Po assumptions h ave been made that particles are much smaller than bubbles, have no inertia and, therefore, are subjected to streamline flow. . The results of the present work show that Pc varies as the square of the particl e size (Dp ) and inversely as the square of the bubble size (Db)' This finding is the same as that obtained by Gaudin?" for Stokes flow conditions except that the proportionality constant is shown to vary with the Reynolds number. The P; value s predicted at various particle sizes and bubble sizes are in good agreement with the probability of collection (P) determined experimentally using a hydrophobic coal
sample whose Pa values approximate unity. The availability of the stream function allowed calculation of tangential veloci ty profiles of the fluids around the bubble, which in turn made it possible to dete rmine the particle sliding time. By relating the sliding time with the induction time (r.), an expression for P; has been derived which is shown to be a function of induction time, particle size and bubble size. It shows that Pa increases with decreasing Downloaded by [McGill University Library] at 14:05 26 May 2014 120 R. H. YOON AND G. H. LUITRELL Q) E i-= c o :..;::; U ::J -0 C 5 4 3 2 o c o (f) Q) ...c -0 <{ ...0 o ...0 o l.... 0.. 10a642 6. Induction TIme c Recovery o Po o L-_---l__--'-__----'-__-"-__--L__0.0 o 12 ao 1OOr---r---r---,..---,..---,..-----, 1.0 20 c Q) > o 40 U Q) 0:::: pH FIGURE II Effect of pH on recovery, induction time and probability of adhesion p redicted using Equation (28). made possible by deriving a stream function that can describe the streamlines around a bubble in the intermediate Reynolds number range, e.g., 0-100. The stream function has been derived using a curve-fitting technique for the streaml ine patterns available in literature. In deriving an expression for Po assumptions h ave been made that particles are much smaller than bubbles, have no inertia and, therefore, are subjected to streamline flow. . The results of the present work show that Pc varies as the square of the particl e size (Dp ) and inversely as the square of the bubble size (Db)' This finding is the same as that obtained by Gaudin?" for Stokes flow conditions except that the proportionality constant is shown to vary with the Reynolds number. The P; value s predicted at various particle sizes and bubble sizes are in good agreement with the probability of collection (P) determined experimentally using a hydrophobic coal
sample whose Pa values approximate unity. The availability of the stream function allowed calculation of tangential veloci ty profiles of the fluids around the bubble, which in turn made it possible to dete rmine the particle sliding time. By relating the sliding time with the induction time (r.), an expression for P; has been derived which is shown to be a function of induction time, particle size and bubble size. It shows that Pa increases with decreasing Downloaded by [McGill University Library] at 14:05 26 May 2014 THE EFFECT OF BUBBLE SIZE ON FINE PARTICLE FLOTATION 121 induction time and also with decreasing particle size. The latter finding sugges ts that it becomes increasingly difficult to achieve selectivity with decreasing pa rticle size. With respect to bubble size, P; reaches a maximum at approximately 350,um for the case of floating 20-,um particles of different induction times. The Pa v alues predicted from the experimentally determined induction times and the bubble size
and particle size used in the microflotation experiments are in reasonable agree ment with the flotation results. Acknowledgments The authors which to acknowledge the financial support for this research which h as been provided by the U.S. Department of Energy under a University Coal Research Grant (No. DE-FG22-83 PC60806) and Sponsored Contract (No. DE-AC22-86PC91221). They are also grateful to Beth Dilli nger Howell for her patient assistance with the manuscript. References 1. V. J. Klassen and V. A. Mokrousov, An Introduction to the Theory of Flotation , Butterworths, London (1963) pp. 402-415. 2. W. J. Trahar, Int. 1. Mineral Processing, 8, 289 (1981). 3. P. Somasundaran, Engineering and Mining Journal, 64-68 December, (1979). 4. P. Somasundaran, Mining Engineering, 1177-1186 August, (1984). 5. D. W. Fuerstenau, "Fine Particle Flotation", in Fine Particle Processing, vol . 1 (P. Somasundaran, ed.), AIME, New York (1980) p. 669. 6. J. A. Kitchener and R. J. Gochin, Water Resources, 15, 585 (1980). 7. G. J. Jameson, S. Nam and M. M. Young, Minerals Sci. & Engrg., 9(3),103 (1977 ). 8. A. M. Gaudin, Flotation, McGraw Hill, New York (1932). 9. K. L. Sutherland, J. Phys. Chem., 52. 394 (1948). 10. H. R. Spedden and W. S. Hanna, Tech. Publ. No. 2354, AIME, March, (1948). 11. P. F. Whelan and D. J. Brown, Bulletin lnst. Mining & Metall., 65(591),181 ( 1956). 12. R. Schuhmann, J. Phys. Chem., 46, 891 (1942). 13. L. R. Flint and W. J. Howarth, Chem. Eng. Sci. 26,1155 (1971). 14. T. Pearcy and G. W. Hill. J. Royal Metall. Society, 83, 77 (1957). 15. A. Fonda and H. Herne, U.K. National Coal Board, Mining Research Establishme nt, Report No. 2068(1957). 16. L. M. Hocking, Int. J. Air Pollution, 13, 154 (1960). 17. U. Shafrir and N. Neiburger, J. Geophys. Res.. 68, 4141 (1964). 18. D. Reay and G. A. Ratcliff, Canadian 1. Chem. Engrg., 51, 178 (1973). 19. M. E. Weber, J. Separation Process Technol., 2. 29 (1981). 20. M. E. Weber and D. Paddock, J. Colloid and Interface Sci., 94(2). 328 (1983) . 21. J. H. Masliyah, Ph.D. Dissertation, University of British Columbia, Vancouve r, British Columbia, Canada (1970). 22. S. W. Woo, Ph.D. Dissertation. McMaster Univesity, Hamilton, Ontario, Canada (1971). 23. L. J. Warren, "Ultrafine Particles in Flotaton", in Principles of Mineral Fl otation, The Wark Symposium. Symp. Series No. 40 (M. H. Jones and J. T. Woodcock, eds.), Austr. In stit. Mining & Metall., Victoria, Australia (1984), p. 185. 24. H. J. Schulze, Physico-Chemical Elementary Processes in Flotation, Elsevier, Amsterdam (1983). 25. J. Sven-Nillson, Kolloid Z., 69, 230 (1934). 26. M. A. Eigeles and M. L. Volova, "Kinetic Investigation of Effect of Contact Time, Temperature and Surface Condition on the Adhesion of Bubbles to Mineral Surfaces", Proceedin gs, 5th Int. Mineral Processing Congress, IMM, London (1960), p. 271. 27. V. A. Glembotsky, The Time of Attachment of Bubbles to Solid Particles in Fl otation and Its Measurement, Lzv. Akad. Nauk SSSR, Otd. Tckhn. Nauk, pp. 1524-1531 (1953). 28. L. F. Evans and W. E. Ewers, Ind. & Engng. Chem .. 46(11), 2420 (1954). 29. J. L. Yordan and R. H. Yoon, "Induction Time Measurements for the Quartz-Ami ne Flotation Downloaded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [Mc Gill University Library] at 14:05 26 May 2014 122 R. H. YOON AND G. H. LUITRELL System", Presentation, 115th SME-AIME Annual Meeting, New Orleans, Louisiana, Ma rch 2-6, 1986; J. L. Yordan and R.H. Yoon, "Induction Time Measurements for a coal flotat ion system", in Interfacial Phenomena in Biotechnology and Materials Processing (Y. A. Attia, B. M. Moudgil and S. Chander, eds.), Elsevier, Amsterdam (1988) p. 333. 30. G. S. Dobby and J. A. Finch "Particle Size Dependence in Flotation Derived f rom a Fundamental Model of the Capture Process", 114th Annual Meeting of SME-AIME, New York, Prepr int No. 85-124 (1985), 10 pp. 3\. A. M. Gaudin, R. Schuhmann and A. W. Schlechten, J. Phys. Chem., 46, 902 (19 42). 32. R. M. Lewis and T. M. Morris, "Operating Data from a Sulfide Flotation Plant ", in Froth Flotation, 50th Anniversary Volume (D. W. Fuerstenau, ed.), AIME, New York (1962 ). 33. D. W. Fuerstenau, R. W. Harper and J. W. Miller, Trans. AIME, 247, 69 (1970) . 34. D. F. Kelsall, P. S. B. Stewart and W. J. Trahar, "Diagnostic Metallurgy. A Systematic Method of Plant Optimization", Proceedings, Symposium on Optimization and Control of Mi neral Processing Plants, Brisbane, July 1974. Australian Mineral Ind. Res. Assoc., Par kville, Victoria (1974) p. 53. 35. P. Sennett and R. H. Young, "Current Problems in Beneficiation of Kaolin Cla y", in Beneficiation of Mineral Fines-Problems and Research Needs (P. Somasundaran and N. Arbiter, ed s.), Report of Workshop by Columbia University, Sterling Forest, New York, August 27-29, 197 8 (1979). 36. D. C. Yang, "Static Tube Flotation for Fine Coal Cleaning", Proceedings, 6th Int. Symp. on Coal Slurry Combustion and Technology, Orlando, FL, June 25-27,1984 (1984) p. 582. 37. R. H. Yoon, Mining Congress Journal, 68(12), 76 (1982). 38. R. C. Cliff, J. R. Grace and M. E. Weber, Bubbles, Drops and Particles, Acad emic Press, New York (1978) p. 27. 39. E. G. Davis, J. P. Hansen and G. V. Sullivan, "Attrition Microgrinding", in Fine Particles Processing, vol. I, (P. Somasundaran, ed.), AIME, New York (1980) p. 74. 40. V. G. Jenson, Proceedings, Royal Society, Section A, 249, 346 (1959). 4\. A. E. Hamielic, T. W. Hoffman and L. L. Ross, J. Amer. Inst. Chern. Engrg.; 13,212 (1967). 42. Y. Rimon and S. I. Cheng, Physics Fluids, 12,949 (1969). 43. D. H. Michael and P. W. Norey, J. Fluid Mechanics, 37(Part 3). 565 (1969). 44. B. P. LeClair, Ph.D. Thesis, McMaster University, Hamilton, Ontario, Canada (1970). 45. H. R. Pruppacher, B. P. LeClair and A. E. Hamielec, J. Fluid Mech., 44, 781 (1970). 46. A. M. Gaudin, Flotation, 2nd edition, McGraw-Hili, New York (1957). 47. M. Coutanceau, J. de Mecanique, 7(1), 4 (1968). 48. J. H. Masliyah and N. Epstein, Prog. Heat Mass Transfer, 6, 613 (1972). 49. M. Payard and M. Coutanccau, C. R. A cad. Sci., Series B, 278, 369 (1974). 50. L. E. Seeley, R. L. Hummel and J. W. Smith, J. Fluid Mechanics, 68(Part 3). 591 (1975). 5\. M. Van Dyke, An Album of Fluid Motion, Parabolic Press, Stanford, California (1982) p. 176. 52. B. V. Derjaguin and S. S. Dukhin, "Kinetic Theory of the Flotation of Fine P articles", Proceedings, 13th International Mineral Processing Congress, (J. Laskowski, ed.) , Warsaw, June 4-9, 1979 (1979) p. 2\. 53. R. P. King, T. A. Hatton and D. G. Hulbert, Trans. IMM, Section C, March, C9 (1974). 54. H. J. Schulze and G. Gottschalk, in Proceedings, 13th International Mineral Processing Congress, Warsaw, Poland, June, 1979 (J. Laskowski, ed.), Elsevier, New York, Part A (1981 ) p. 63. 55. R. H. Yoon and G. H. Luttrell, Coal Preparation, An International Journal, 2 , 179 (1986). 56. A. C. Partridge and G. W. Smith, Trans. IMM, Sec. C, 80, C199 (1971). Downlo aded by [McGill University Library] at 14:05 26 May 2014 Downloaded by [McGill U niversity Library] at 14:05 26 May 2014