Search

Froth models proposed in the literature are reviewed to identify their significance and usefulness in modelling and scale-up of the froth phase. Froth-phase performance is better understood in terms of froth recovery, the fraction of material from the froth-pulp interface that reports to the concentrate. Froth recovery may be a strong function of the rate of drainage of particles from the froth to the slurry phase. Drainage rate is determined by physical factors such as froth removal technique, flotation cell geometry, air flux and distribution at the pulp-froth interface, water content, particle size and solids content, and chemical factors such as froth stability and loading. These factors influence froth residence time, a key parameter. Future work should focus on developing a methodology to investigate froth performance on the basis of froth recovery in different systems., Froth models proposed in the literature are reviewed to identify their significance and usefulness in modelling and scale-up of the froth phase. Froth-phase performance is better understood in terms of froth recovery, the fraction of material from the froth-pulp interface that reports to the concentrate. Froth recovery may be a strong function of the rate of drainage of particles from the froth to the slurry phase. Drainage rate is determined by physical factors such as froth removal technique, flotation cell geometry, air flux and distribution at the pulp-froth interface, water content, particle size and solids content, and chemical factors such as froth stability and loading. These factors influence froth residence time, a key parameter. Future work should focus on developing a methodology to investigate froth performance on the basis of froth recovery in different systems.

A method with excellent reproducibility has been developed for measuring the loading of bubbles by particles in a microflotation cell. Bubble size and distribution are determined with a UCT Bubble Sizer. Air is introduced through a syringe at the base of the cell, at a set flow rate and pressure. Loaded bubbles rise through the cell, are deflected off an inverted cone and report to the launder. After a set time of 30 s the syringe is removed and the flotation product collected and weighed. Air flow rate and bubble size readings are re-measured and the average bubble loading calculated from the mass of mineral recovered and the measured bubble sizes. The number of particles loaded per bubble was found to decrease with increasing particle size. A good correlation was observed between bubble loading and the results of batch flotation tests. The method presents an opportunity for screening reagents and possibly for dynamic measurement of hydrophobicity., A method with excellent reproducibility has been developed for measuring the loading of bubbles by particles in a microflotation cell. Bubble size and distribution are determined with a UCT Bubble Sizer. Air is introduced through a syringe at the base of the cell, at a set flow rate and pressure. Loaded bubbles rise through the cell, are deflected off an inverted cone and report to the launder. After a set time of 30 s the syringe is removed and the flotation product collected and weighed. Air flow rate and bubble size readings are re-measured and the average bubble loading calculated from the mass of mineral recovered and the measured bubble sizes. The number of particles loaded per bubble was found to decrease with increasing particle size. A good correlation was observed between bubble loading and the results of batch flotation tests. The method presents an opportunity for screening reagents and possibly for dynamic measurement of hydrophobicity.

It is shown that classical bubble theory, incorporating the use of the axial dispersion model and assuming flotation to be a first order process, is adequate for predicting the performance of a large column using a rate constant determined in a small column as long as due consideration is given to any difference in bubble size between the two cells. Equations are given for the scale-up procedure, which is summarised as follows. Ore type and pulp chemistry must be the same in both columns. The vessel dispersion coefficient is predicted for the large vessel and the first order rate constant for the small column determined experimentally by operating in a frothless mode. Average bubble size in the small column is measured and manufacturers' specifications or previous data used to predict the bubble size in the large column. The rate constant for the large column is then predicted and the collection zone recovery determined. Finally, overall recovery in the large column is predicted assuming a froth zone recovery of about 25%., It is shown that classical bubble theory, incorporating the use of the axial dispersion model and assuming flotation to be a first order process, is adequate for predicting the performance of a large column using a rate constant determined in a small column as long as due consideration is given to any difference in bubble size between the two cells. Equations are given for the scale-up procedure, which is summarised as follows. Ore type and pulp chemistry must be the same in both columns. The vessel dispersion coefficient is predicted for the large vessel and the first order rate constant for the small column determined experimentally by operating in a frothless mode. Average bubble size in the small column is measured and manufacturers' specifications or previous data used to predict the bubble size in the large column. The rate constant for the large column is then predicted and the collection zone recovery determined. Finally, overall recovery in the large column is predicted assuming a froth zone recovery of about 25%.

The mechanisms by which pure thiol collectors produce a hydrophobic mineral surface need to be understood in order to establish the reasons for their synergism when combined. It was found that sodium cyclohexyl dithiocarbamate and potassium butyl xanthate undergo the same initial reactions with pyrite in acidic conditions, releasing heat at about -67 kJ/mole. This reaction corresponds to the formation of the metal thiolate which is chemisorbed. The reaction of sodium cyclohexyl dithiocarbamate stops at pH above about 7 despite continued reagent addition. Potassium butyl xanthate undergoes a further reaction which is interpreted as oxidation of the adsorbed collector to dixanthogen; ferrous dithiocarbamate, on the other hand, is strongly stable and does not react further. The increased heat generated when mixtures are used indicates that the reactions are stronger, possibly because the presence of dithiocarbamate facilitates the increased oxidation of xanthate to dixanthogen. There may also be a stronger bond formed between the mixed collector and the mineral surface due to collector interaction., The mechanisms by which pure thiol collectors produce a hydrophobic mineral surface need to be understood in order to establish the reasons for their synergism when combined. It was found that sodium cyclohexyl dithiocarbamate and potassium butyl xanthate undergo the same initial reactions with pyrite in acidic conditions, releasing heat at about -67 kJ/mole. This reaction corresponds to the formation of the metal thiolate which is chemisorbed. The reaction of sodium cyclohexyl dithiocarbamate stops at pH above about 7 despite continued reagent addition. Potassium butyl xanthate undergoes a further reaction which is interpreted as oxidation of the adsorbed collector to dixanthogen; ferrous dithiocarbamate, on the other hand, is strongly stable and does not react further. The increased heat generated when mixtures are used indicates that the reactions are stronger, possibly because the presence of dithiocarbamate facilitates the increased oxidation of xanthate to dixanthogen. There may also be a stronger bond formed between the mixed collector and the mineral surface due to collector interaction.

A novel bubble sampler is described which allows the measurement of bubble size distributions in flotation cells containing slurries with high solids concentration. The device separates bubbles from a slurry into a water solution where they can be measured using a bubble size analyser. An experimental programme is described in which bubble sizes were measured in two and three phases and the effects of different frothers on bubble size investigated. In two phase operations (liquid/gas), the bubble sampler was found to supply accurate and reproducible measurements when compared with measurements made by the bubble size analyser without the sampler fitted. The frother type was found to have the greatest influence on bubble size. With the addition of solids (gold plant tailings) to form a three phase system, the mean bubble size was found to increase and only the further addition of frother could reduce it to its former value. Different frothers were found to have very different effects on the size of bubbles produced in slurries compared with those produced in two phase systems. Of the frothers tested, MIBC was found to be the best all-round performer in terms of bubble size reduction with increasing solids concentration., A novel bubble sampler is described which allows the measurement of bubble size distributions in flotation cells containing slurries with high solids concentration. The device separates bubbles from a slurry into a water solution where they can be measured using a bubble size analyser. An experimental programme is described in which bubble sizes were measured in two and three phases and the effects of different frothers on bubble size investigated. In two phase operations (liquid/gas), the bubble sampler was found to supply accurate and reproducible measurements when compared with measurements made by the bubble size analyser without the sampler fitted. The frother type was found to have the greatest influence on bubble size. With the addition of solids (gold plant tailings) to form a three phase system, the mean bubble size was found to increase and only the further addition of frother could reduce it to its former value. Different frothers were found to have very different effects on the size of bubbles produced in slurries compared with those produced in two phase systems. Of the frothers tested, MIBC was found to be the best all-round performer in terms of bubble size reduction with increasing solids concentration.

Results are presented which show that when mixtures of dithiocarbamates are used to float pyrite, each component has a synergistic effect on the performance of the others. Best results were obtained when a 90:10 mole% ratio of di-n-propyl dithiocarbamate and cyclo-hexyl dithiocarbamate respectively was used. The mixtures yielded higher recoveries and grades and also resulted in a greater mass pull, greater recovery of coarse particles and enhanced selectivity. The increased recovery may be due either to the selective adsorption of different collectors on particular sites or the orientation of the alkyl chains causing better surface coverage. Increasing the dosage in the case of pure collectors and mixtures increased the recoveries and grades and effectively masked the synergistic effect. When mixtures of dithiocarbamates and other thiol collectors were used it was clear that the dithiocarbamates were the weakest collectors since the other thiol collectors all performed better than the pure dithiocarbamates or mixtures, irrespective of total dosage. Best results were achieved with a 90:10 mole% mixture of n-butyl xanthate and cyclo-hexyl dithiocarbamate. The method of collector addition (by premixing or simultaneous addition) made no difference but addition of the weaker collector before the stronger collector slightly enhanced performance. The addition of copper sulphate to various collectors marginally increased recoveries, but in the case of cyclo-hexyl dithiocarbamate resulted in a dramatic increase in recovery and grade., Results are presented which show that when mixtures of dithiocarbamates are used to float pyrite, each component has a synergistic effect on the performance of the others. Best results were obtained when a 90:10 mole% ratio of di-n-propyl dithiocarbamate and cyclo-hexyl dithiocarbamate respectively was used. The mixtures yielded higher recoveries and grades and also resulted in a greater mass pull, greater recovery of coarse particles and enhanced selectivity. The increased recovery may be due either to the selective adsorption of different collectors on particular sites or the orientation of the alkyl chains causing better surface coverage. Increasing the dosage in the case of pure collectors and mixtures increased the recoveries and grades and effectively masked the synergistic effect. When mixtures of dithiocarbamates and other thiol collectors were used it was clear that the dithiocarbamates were the weakest collectors since the other thiol collectors all performed better than the pure dithiocarbamates or mixtures, irrespective of total dosage. Best results were achieved with a 90:10 mole% mixture of n-butyl xanthate and cyclo-hexyl dithiocarbamate. The method of collector addition (by premixing or simultaneous addition) made no difference but addition of the weaker collector before the stronger collector slightly enhanced performance. The addition of copper sulphate to various collectors marginally increased recoveries, but in the case of cyclo-hexyl dithiocarbamate resulted in a dramatic increase in recovery and grade.

It was shown that the degree of mixing in both the froth and pulp zones of a column flotation cell increased with feed rate and air flow rate and decreased with increased frother concentration, pulp density and froth height. An increase in the wash water flow rate reduced mixing in the froth zone and increased it in the pulp zone. Factors which increased the degree of mixing in the froth zone resulted in decreased concentrate grade, while concentrate recovery of pyrite was essentially independent of mixing in either zone. The effect of each of the variables studied was consistent with the findings of other investigations., It was shown that the degree of mixing in both the froth and pulp zones of a column flotation cell increased with feed rate and air flow rate and decreased with increased frother concentration, pulp density and froth height. An increase in the wash water flow rate reduced mixing in the froth zone and increased it in the pulp zone. Factors which increased the degree of mixing in the froth zone resulted in decreased concentrate grade, while concentrate recovery of pyrite was essentially independent of mixing in either zone. The effect of each of the variables studied was consistent with the findings of other investigations.

The residence time distribution characteristics were determined for solid and liquid phases in a laboratory column and two industrial columns, using radioactively-labelled solid and liquid tracers. For columns with 10-150 cm diameters the dispersion model was best, while larger columns could be modelled using the tanks-in-series model and either model was appropriate for small-diameter columns. Feed and gangue material produced similar tailings age-distribution curves. The solid dispersion coefficient was larger than the liquid dispersion coefficient, increasing with particle size in the laboratory column but remaining constant in the large columns; a modified correlation is proposed to predict the solid dispersion coefficient in the latter., The residence time distribution characteristics were determined for solid and liquid phases in a laboratory column and two industrial columns, using radioactively-labelled solid and liquid tracers. For columns with 10-150 cm diameters the dispersion model was best, while larger columns could be modelled using the tanks-in-series model and either model was appropriate for small-diameter columns. Feed and gangue material produced similar tailings age-distribution curves. The solid dispersion coefficient was larger than the liquid dispersion coefficient, increasing with particle size in the laboratory column but remaining constant in the large columns; a modified correlation is proposed to predict the solid dispersion coefficient in the latter.

Modelling of the tailings residence time distribution (RTD) in flotation columns has been described in a number of studies and the axial dispersion model has been extensively used. The model has three solutions depending on the boundary conditions of the system. The solutions can be fitted to experimental data using moments matching or direct search methods. The solutions of the various boundary conditions and the limitations in the application of the method of moments matching are discussed. Inconsistencies found in previous studies are described. The model can provide valuable information if used correctly., Modelling of the tailings residence time distribution (RTD) in flotation columns has been described in a number of studies and the axial dispersion model has been extensively used. The model has three solutions depending on the boundary conditions of the system. The solutions can be fitted to experimental data using moments matching or direct search methods. The solutions of the various boundary conditions and the limitations in the application of the method of moments matching are discussed. Inconsistencies found in previous studies are described. The model can provide valuable information if used correctly.

The flotation of pyrite is of great importance to the gold mining industry in South Africa and Australia. The flotation plants fall into a number of categories: flotation after milling, flotation of current residues and reclaimed residues, flotation of refractory ores and the flotation of gold-copper ores. These processes are discussed with particular attention being given to the differences in the physical and chemical operating parameters, the different circuits used on existing plants and the application of new cell technologies and flotation reagents., The flotation of pyrite is of great importance to the gold mining industry in South Africa and Australia. The flotation plants fall into a number of categories: flotation after milling, flotation of current residues and reclaimed residues, flotation of refractory ores and the flotation of gold-copper ores. These processes are discussed with particular attention being given to the differences in the physical and chemical operating parameters, the different circuits used on existing plants and the application of new cell technologies and flotation reagents.

Solid phase residence time distribution data for both the froth and pulp phases is reported for a laboratory column flotation cell. From this data, residence time distribution models for the pulp and froth phases have been developed. A plug flow reactor with recycle was found best to describe the behaviour of the solids in the froth phase and a tanks-in-series model was most suitable for the solid in the pulp. The feed region was found to be highly turbulent and was modelled using a mixed reactor with dead volume., Solid phase residence time distribution data for both the froth and pulp phases is reported for a laboratory column flotation cell. From this data, residence time distribution models for the pulp and froth phases have been developed. A plug flow reactor with recycle was found best to describe the behaviour of the solids in the froth phase and a tanks-in-series model was most suitable for the solid in the pulp. The feed region was found to be highly turbulent and was modelled using a mixed reactor with dead volume.

The frother concentration, the air rate and the pulp density affect the sizes of the bubbles in the pulp. This in turn influences the nature of the froth and the mixing characteristics of the froth and the pulp. When the air rate and pulp density decrease and the frother concentration increases, small bubbles are formed in the pulp, greater plug flow behaviour is observed and entrainment of fine particles decreases. The small pulp phase bubbles result in a more stable froth and bubble coalescence is reduced. The froth also has more plug flow behaviour and grades and recoveries increase. Factors which result in the pulp and froth phases having greater plug flow behaviour are conducive to an improved metallurgical performance. Apart from the effect of air rate and frother concentration, the sparger system, the internals and the aspect ratio of the column are also expected to be critical factors in this regard., The frother concentration, the air rate and the pulp density affect the sizes of the bubbles in the pulp. This in turn influences the nature of the froth and the mixing characteristics of the froth and the pulp. When the air rate and pulp density decrease and the frother concentration increases, small bubbles are formed in the pulp, greater plug flow behaviour is observed and entrainment of fine particles decreases. The small pulp phase bubbles result in a more stable froth and bubble coalescence is reduced. The froth also has more plug flow behaviour and grades and recoveries increase. Factors which result in the pulp and froth phases having greater plug flow behaviour are conducive to an improved metallurgical performance. Apart from the effect of air rate and frother concentration, the sparger system, the internals and the aspect ratio of the column are also expected to be critical factors in this regard.

Modelling of the mixing characteristics of the pulp phase in a flotation column has been investigated. The literature of previous modelling in bubble and flotation columns has been reviewed. Liquid tracer residence time distribution studies were performed on a laboratory flotation column. A tanks-in-series model was more accurate than the previously used axial dispersion model in a flotation column scale-up. The assumption made by previous investigators that the liquid and solids dispersion coefficient are equivalent does not apply for all operating conditions in flotation columns. Bubble size was found to substantially affect mixing., Modelling of the mixing characteristics of the pulp phase in a flotation column has been investigated. The literature of previous modelling in bubble and flotation columns has been reviewed. Liquid tracer residence time distribution studies were performed on a laboratory flotation column. A tanks-in-series model was more accurate than the previously used axial dispersion model in a flotation column scale-up. The assumption made by previous investigators that the liquid and solids dispersion coefficient are equivalent does not apply for all operating conditions in flotation columns. Bubble size was found to substantially affect mixing.

The effect of temperature on the flotation of pyrite has been investigated. At temperatures below ambient, an increase in temperature results in an increase in the rate of flotation of fast-floating particles but has a minimal effect on slow-floating particles. At temperatures greater than ambient, the dominant effect of increasing temperature is to decrease the viscosity of the water, thus increasing the rate of elutriation of the gangue back to the pulp. In general, an increase in temperature resulted in an increase in the rate of pyrite recovery and the sulphur grades and a decrease in water recovery and size of bubbles in the pulp., The effect of temperature on the flotation of pyrite has been investigated. At temperatures below ambient, an increase in temperature results in an increase in the rate of flotation of fast-floating particles but has a minimal effect on slow-floating particles. At temperatures greater than ambient, the dominant effect of increasing temperature is to decrease the viscosity of the water, thus increasing the rate of elutriation of the gangue back to the pulp. In general, an increase in temperature resulted in an increase in the rate of pyrite recovery and the sulphur grades and a decrease in water recovery and size of bubbles in the pulp.

The recovery of arsenopyrite from an arsenopyrite/pyrite ore can be optimised using a two stage flotation process in which a dithiophosphate is added at pH 11 in the first stage and copper sulphate and a dithiocarbamate in the second stage. It was found that better separations were obtained when aged ore was used. It was possible to simulate this aging process by heating. Various alkyl functional groups of the collectors were tested and the best separations were obtained using sodium isobutyl dithiophosphate and a mixture of 10%:90% sodium cyclohexyl:n-propyl dithiocarbamate. This suite of collectors led to a recovery of 74.8% arsenopyrite and 8.4% pyrite with grades of 37.9% arsenopyrite and 11.6% pyrite respectively. The grades of gold in the feed, concentrate and tailings for these conditions were 8.10 g/ton, 271 g/ton and 2.55 g/ton respectively. This represented a gold recovery of 83.6%., The recovery of arsenopyrite from an arsenopyrite/pyrite ore can be optimised using a two stage flotation process in which a dithiophosphate is added at pH 11 in the first stage and copper sulphate and a dithiocarbamate in the second stage. It was found that better separations were obtained when aged ore was used. It was possible to simulate this aging process by heating. Various alkyl functional groups of the collectors were tested and the best separations were obtained using sodium isobutyl dithiophosphate and a mixture of 10%:90% sodium cyclohexyl:n-propyl dithiocarbamate. This suite of collectors led to a recovery of 74.8% arsenopyrite and 8.4% pyrite with grades of 37.9% arsenopyrite and 11.6% pyrite respectively. The grades of gold in the feed, concentrate and tailings for these conditions were 8.10 g/ton, 271 g/ton and 2.55 g/ton respectively. This represented a gold recovery of 83.6%.

It has been shown that increasing the feed rate, decreasing the wash water rate, increasing the bubble size and decreasing the bubble bed height caused decreases in the grade, the particle size and the mean residence time in the column. Recovery is enhanced by optimising froth stability and froth crowding effects. Mean residence time in the froth was directly related to concentrate grade, but recovery was independent. A plug flow model with recycle is proposed for the froth phase and a tanks in series model for the pulp. The liquid residence time distribution was not indicative of the solids residence time distribution., It has been shown that increasing the feed rate, decreasing the wash water rate, increasing the bubble size and decreasing the bubble bed height caused decreases in the grade, the particle size and the mean residence time in the column. Recovery is enhanced by optimising froth stability and froth crowding effects. Mean residence time in the froth was directly related to concentrate grade, but recovery was independent. A plug flow model with recycle is proposed for the froth phase and a tanks in series model for the pulp. The liquid residence time distribution was not indicative of the solids residence time distribution.

A method for measuring bubble sizes in a laboratory column cell for two- and three-phase systems is described. An increase in viscosity, solid-particle size, pulp density and air flow rate and a decrease in temperature all resulted in larger bubbles. Bubble size decreased with an increase in the frother concentration and ionic strength of the solution. An increase in pH caused an increase in bubble size but the presence of xanthate collectors had no effect on bubble size. Reasons for these observations are proposed., A method for measuring bubble sizes in a laboratory column cell for two- and three-phase systems is described. An increase in viscosity, solid-particle size, pulp density and air flow rate and a decrease in temperature all resulted in larger bubbles. Bubble size decreased with an increase in the frother concentration and ionic strength of the solution. An increase in pH caused an increase in bubble size but the presence of xanthate collectors had no effect on bubble size. Reasons for these observations are proposed.

A survey of a number of pyrite flotation plants using copper sulphate showed that in correct dosages it was able to increase both grades and recoveries. This was confirmed in batch tests. Copper adsorbed irreversibly onto the pyrite when the latter was slurried in a copper sulphate solution. Copper sulphate had a significant effect on the maximum froth height obtained in static three-phase froth tests. Proposals to explain these observations are made., A survey of a number of pyrite flotation plants using copper sulphate showed that in correct dosages it was able to increase both grades and recoveries. This was confirmed in batch tests. Copper adsorbed irreversibly onto the pyrite when the latter was slurried in a copper sulphate solution. Copper sulphate had a significant effect on the maximum froth height obtained in static three-phase froth tests. Proposals to explain these observations are made.

The effect of the type of flotation cell (Denver, Wemco, Agitair and Leeds Autofloat) and the cell configuration (round or square) on both the kinetics of flotation as indicated by the recovery after 60 seconds and the final recovery was investigated. The Leeds cell was found to yield the most reproducible results over the entire flotation time and also performed best with respect to grades and recoveries. The use of a round cell caused slower flotation with poorer recoveries but higher grades., The effect of the type of flotation cell (Denver, Wemco, Agitair and Leeds Autofloat) and the cell configuration (round or square) on both the kinetics of flotation as indicated by the recovery after 60 seconds and the final recovery was investigated. The Leeds cell was found to yield the most reproducible results over the entire flotation time and also performed best with respect to grades and recoveries. The use of a round cell caused slower flotation with poorer recoveries but higher grades.

A novel method by which representative discrete samples can be obtained from the froth zone in a small cylindrical batch flotation cell is described. The particle size of the pyrite is generally smallest, and the grade the lowest, at the bottom of the froth. Most of the pyrite is present in the top and bottom layers of the froth. The pyrite in the bottom layer is thought to be due to recycling of ultra-fine pyrite., A novel method by which representative discrete samples can be obtained from the froth zone in a small cylindrical batch flotation cell is described. The particle size of the pyrite is generally smallest, and the grade the lowest, at the bottom of the froth. Most of the pyrite is present in the top and bottom layers of the froth. The pyrite in the bottom layer is thought to be due to recycling of ultra-fine pyrite.

Residue material from a gold mine in the Orange Free State, in which 50% of the kerogen was larger than 100 micro metres in size was tested. Although kerogen is naturally hydrophobic, observed flotation was poor, this could be due to occlusion of the kerogen by a slime coating and subsequent depression of the slime by guar gums. The recovery of gold from samples spiked with kerogen increased, with no significant reduction in grades, when a petroleum-based collector and tetrasodiumpyrophosphate were used., Residue material from a gold mine in the Orange Free State, in which 50% of the kerogen was larger than 100 micro metres in size was tested. Although kerogen is naturally hydrophobic, observed flotation was poor, this could be due to occlusion of the kerogen by a slime coating and subsequent depression of the slime by guar gums. The recovery of gold from samples spiked with kerogen increased, with no significant reduction in grades, when a petroleum-based collector and tetrasodiumpyrophosphate were used.