No description available.

Flotation is the separation process most used to recover valuable minerals from sulphide ores. The design of industrial flotation plants is a complex process involving many stages. Current design practice involves performing laboratory scale grinding and batch flotation tests, followed by a circuit design based on the scale-up of the laboratory kinetics and recovery-grade data. A pilot plant is then operated to evaluate the performance of the circuit based on recovery and grade, usually in the configuration of the intended full-scale plant. The circuit design is refined and economically evaluated after which the design of the full-scale plant is performed. A new approach to full-scale flotation plant design has been proposed by the Julius Kruttschnitt Mineral Research Centre and the Mineral Processing Research Unit of the University of Cape Town, as part of the Australian Minerals Industries Research Association (AMIRA) P9 Project. In this new methodology, the effects of the ore on plant performance are decoupled from the circuit effects. The ore properties are characterised by operating the pilot plant in as simple a configuration as possible. The pilot plant units are configured to perform a similar duty (in terms of mineral content and particle size) to the full-scale operation and their response measured. An important factor in the success of the methodology is having a pilot plant that is capable of accurately characterising the ore properties. For this purpose, the Wemco® Floatability Characterisation Test Rig (FCTR) was built. The FCTR is a self-contained, highly instrumented mobile pilot plant designed to develop and validate the new flotation plant design methodology. The aims of this thesis were to propose, develop and validate a methodology for using the FCTR to design industrial flotation plants. The hypothesis was that full-scale flotation plant design could be accurately performed using the P9 flotation model and modelling and scale-up methodologies, in conjunction with the FCTR. Test work was performed in three main areas: calibration of the ore characteristics and model parameters for the flotation model currently used by the AMIRA P9 Project; validation of the flotation model and modelling methodology; and prediction of full-scale plant performance using parameters determined on the FCTR. The ore floatability characteristics were calibrated using four FCTR circuits of increasing complexity. The ore floatability characteristics were determined for various models derived from data from one, two, three and four calibration circuits. Validation of the flotation model and modelling methodology was performed using three validation methods: internal validity tests, parameter sensitivity tests and predictive validation. From the internal validity tests, some of the models did not meet the validation criteria. The parameter sensitivity tests used Monte Carlo simulations to determine the sensitivity of the regressed model parameters. The tests produced small differences in the values determined from the models and the average values from the Monte Carlo simulations. The floatability characteristics appeared to be stable and unique. Predictive validation was performed using five FCTR circuits, different in configuration to the calibration circuits. The predictive validation was performed using the floatability characteristics determined from each of the calibration models, in conjunction with estimates of the model parameters. Overall, the predictions of the circuit performance were accurate and within experimental standard deviations for most streams in the circuits. The predictions of the key parameters of pentlandite and chalcopyrite recovery were accurate, especially for the final concentrate. The prediction of pyrrhotite recovery produced the largest errors. The prediction of pyrrhotite recovery appeared to be dependent on the addition of depressant to the cleaner and recleaner circuits of the circuit to be predicted. When the depressant addition rates were significantly different from those used in the calibration circuits, the prediction of pyrrhotite recovery was inaccurate. These errors were however reduced when experimental water recovery values were used. The extensive and robust validation of the flotation model and modelling methodology has shown that the flotation model and modelling methodology are valid under certain conditions. This test work represents the first comprehensive validation of the flotation model and modelling methodology incorporating changes in circuit configuration. Using the proposed modelling and scale-up methodology in conjunction with the FCTR, the metallurgical performance of three industrial flotation circuits were predicted. The predicted results were then compared to the experimentally determined results around the industrial circuits. In each case, a scale-up factor between the ore floatability characteristics determined on the FCTR, and the full-scale floatability characteristics, was required to achieve an accurate prediction. The scale-up factor ranged from 0.17 to 0.97 for the case studies investigated. In light of the results from each stage of test work, the proposed flotation plant design methodology was refined. With this methodology and the continual development of techniques for the measurement and prediction of the P9 flotation model parameters, accurate industrial plant design using the FCTR will become possible. With the addition of other unit operations, such as comminution, into the flowsheet, this methodology will eventually lead to the achievement of the ultimate goal of accurate plant design of green-field sites.

Flotation is the separation process most used to recover valuable minerals from sulphide ores. The design of industrial flotation plants is a complex process involving many stages. Current design practice involves performing laboratory scale grinding and batch flotation tests, followed by a circuit design based on the scale-up of the laboratory kinetics and recovery-grade data. A pilot plant is then operated to evaluate the performance of the circuit based on recovery and grade, usually in the configuration of the intended full-scale plant. The circuit design is refined and economically evaluated after which the design of the full-scale plant is performed. A new approach to full-scale flotation plant design has been proposed by the Julius Kruttschnitt Mineral Research Centre and the Mineral Processing Research Unit of the University of Cape Town, as part of the Australian Minerals Industries Research Association (AMIRA) P9 Project. In this new methodology, the effects of the ore on plant performance are decoupled from the circuit effects. The ore properties are characterised by operating the pilot plant in as simple a configuration as possible. The pilot plant units are configured to perform a similar duty (in terms of mineral content and particle size) to the full-scale operation and their response measured. An important factor in the success of the methodology is having a pilot plant that is capable of accurately characterising the ore properties. For this purpose, the Wemco® Floatability Characterisation Test Rig (FCTR) was built. The FCTR is a self-contained, highly instrumented mobile pilot plant designed to develop and validate the new flotation plant design methodology. The aims of this thesis were to propose, develop and validate a methodology for using the FCTR to design industrial flotation plants. The hypothesis was that full-scale flotation plant design could be accurately performed using the P9 flotation model and modelling and scale-up methodologies, in conjunction with the FCTR. Test work was performed in three main areas: calibration of the ore characteristics and model parameters for the flotation model currently used by the AMIRA P9 Project; validation of the flotation model and modelling methodology; and prediction of full-scale plant performance using parameters determined on the FCTR. The ore floatability characteristics were calibrated using four FCTR circuits of increasing complexity. The ore floatability characteristics were determined for various models derived from data from one, two, three and four calibration circuits. Validation of the flotation model and modelling methodology was performed using three validation methods: internal validity tests, parameter sensitivity tests and predictive validation. From the internal validity tests, some of the models did not meet the validation criteria. The parameter sensitivity tests used Monte Carlo simulations to determine the sensitivity of the regressed model parameters. The tests produced small differences in the values determined from the models and the average values from the Monte Carlo simulations. The floatability characteristics appeared to be stable and unique. Predictive validation was performed using five FCTR circuits, different in configuration to the calibration circuits. The predictive validation was performed using the floatability characteristics determined from each of the calibration models, in conjunction with estimates of the model parameters. Overall, the predictions of the circuit performance were accurate and within experimental standard deviations for most streams in the circuits. The predictions of the key parameters of pentlandite and chalcopyrite recovery were accurate, especially for the final concentrate. The prediction of pyrrhotite recovery produced the largest errors. The prediction of pyrrhotite recovery appeared to be dependent on the addition of depressant to the cleaner and recleaner circuits of the circuit to be predicted. When the depressant addition rates were significantly different from those used in the calibration circuits, the prediction of pyrrhotite recovery was inaccurate. These errors were however reduced when experimental water recovery values were used. The extensive and robust validation of the flotation model and modelling methodology has shown that the flotation model and modelling methodology are valid under certain conditions. This test work represents the first comprehensive validation of the flotation model and modelling methodology incorporating changes in circuit configuration. Using the proposed modelling and scale-up methodology in conjunction with the FCTR, the metallurgical performance of three industrial flotation circuits were predicted. The predicted results were then compared to the experimentally determined results around the industrial circuits. In each case, a scale-up factor between the ore floatability characteristics determined on the FCTR, and the full-scale floatability characteristics, was required to achieve an accurate prediction. The scale-up factor ranged from 0.17 to 0.97 for the case studies investigated. In light of the results from each stage of test work, the proposed flotation plant design methodology was refined. With this methodology and the continual development of techniques for the measurement and prediction of the P9 flotation model parameters, accurate industrial plant design using the FCTR will become possible. With the addition of other unit operations, such as comminution, into the flowsheet, this methodology will eventually lead to the achievement of the ultimate goal of accurate plant design of green-field sites.

Flotation is the separation process most used to recover valuable minerals from sulphide ores. The design of industrial flotation plants is a complex process involving many stages. Current design practice involves performing laboratory scale grinding and batch flotation tests, followed by a circuit design based on the scale-up of the laboratory kinetics and recovery-grade data. A pilot plant is then operated to evaluate the performance of the circuit based on recovery and grade, usually in the configuration of the intended full-scale plant. The circuit design is refined and economically evaluated after which the design of the full-scale plant is performed. A new approach to full-scale flotation plant design has been proposed by the Julius Kruttschnitt Mineral Research Centre and the Mineral Processing Research Unit of the University of Cape Town, as part of the Australian Minerals Industries Research Association (AMIRA) P9 Project. In this new methodology, the effects of the ore on plant performance are decoupled from the circuit effects. The ore properties are characterised by operating the pilot plant in as simple a configuration as possible. The pilot plant units are configured to perform a similar duty (in terms of mineral content and particle size) to the full-scale operation and their response measured. An important factor in the success of the methodology is having a pilot plant that is capable of accurately characterising the ore properties. For this purpose, the Wemco® Floatability Characterisation Test Rig (FCTR) was built. The FCTR is a self-contained, highly instrumented mobile pilot plant designed to develop and validate the new flotation plant design methodology. The aims of this thesis were to propose, develop and validate a methodology for using the FCTR to design industrial flotation plants. The hypothesis was that full-scale flotation plant design could be accurately performed using the P9 flotation model and modelling and scale-up methodologies, in conjunction with the FCTR. Test work was performed in three main areas: calibration of the ore characteristics and model parameters for the flotation model currently used by the AMIRA P9 Project; validation of the flotation model and modelling methodology; and prediction of full-scale plant performance using parameters determined on the FCTR. The ore floatability characteristics were calibrated using four FCTR circuits of increasing complexity. The ore floatability characteristics were determined for various models derived from data from one, two, three and four calibration circuits. Validation of the flotation model and modelling methodology was performed using three validation methods: internal validity tests, parameter sensitivity tests and predictive validation. From the internal validity tests, some of the models did not meet the validation criteria. The parameter sensitivity tests used Monte Carlo simulations to determine the sensitivity of the regressed model parameters. The tests produced small differences in the values determined from the models and the average values from the Monte Carlo simulations. The floatability characteristics appeared to be stable and unique. Predictive validation was performed using five FCTR circuits, different in configuration to the calibration circuits. The predictive validation was performed using the floatability characteristics determined from each of the calibration models, in conjunction with estimates of the model parameters. Overall, the predictions of the circuit performance were accurate and within experimental standard deviations for most streams in the circuits. The predictions of the key parameters of pentlandite and chalcopyrite recovery were accurate, especially for the final concentrate. The prediction of pyrrhotite recovery produced the largest errors. The prediction of pyrrhotite recovery appeared to be dependent on the addition of depressant to the cleaner and recleaner circuits of the circuit to be predicted. When the depressant addition rates were significantly different from those used in the calibration circuits, the prediction of pyrrhotite recovery was inaccurate. These errors were however reduced when experimental water recovery values were used. The extensive and robust validation of the flotation model and modelling methodology has shown that the flotation model and modelling methodology are valid under certain conditions. This test work represents the first comprehensive validation of the flotation model and modelling methodology incorporating changes in circuit configuration. Using the proposed modelling and scale-up methodology in conjunction with the FCTR, the metallurgical performance of three industrial flotation circuits were predicted. The predicted results were then compared to the experimentally determined results around the industrial circuits. In each case, a scale-up factor between the ore floatability characteristics determined on the FCTR, and the full-scale floatability characteristics, was required to achieve an accurate prediction. The scale-up factor ranged from 0.17 to 0.97 for the case studies investigated. In light of the results from each stage of test work, the proposed flotation plant design methodology was refined. With this methodology and the continual development of techniques for the measurement and prediction of the P9 flotation model parameters, accurate industrial plant design using the FCTR will become possible. With the addition of other unit operations, such as comminution, into the flowsheet, this methodology will eventually lead to the achievement of the ultimate goal of accurate plant design of green-field sites.

Abstract Screen modelling has traditionally been based on rudimentary empirical ‘factor’ methods, or semi-empirical data-fitting techniques. Both of these methods have significant limitations in practice, and industrial screen optimization remains somewhat of a ‘black art’. This thesis introduces the concept of dimensional analysis and scale model similitude to the problem of modelling vibrating screens. This leads to a new method of modelling industrial screens. A small-scale screen can be built at a more convenient scale in the laboratory, and used to predict the performance of a large industrial scale machine. Verification of the scaling theory is based on three separate methods: 1. Firstly, the scaling theory is developed by analysing particle-level forces acting in a dry granular system. It is shown that scale-up of granular systems can be achieved using Froude scaling: that is, if the geometry and operating variables in an experiment are scaled in a pre-defined manner, the behaviour of the particles in the full-scale can be predicted from the behaviour of the particles at the smaller scale. 2. Secondly, the scaling rules are applied to a number of idealized granular systems using numerical simulations via the discrete element method (DEM). The granular systems modelled include inclined vibrating feeders and inclined vibrating screens. It is found that simulations performed at different scales yield almost identical dimensionless responses when the geometry and operating conditions are scaled according to Froude scaling rules. 3. Thirdly, the scaling rules are applied to modelling physical screening data. A dataset obtained from a larger pilot-scale screen in a thesis by R. De Pretto (1992) is reproduced at a smaller scale in this thesis using a purpose-built laboratory-scale screen. The throughput, efficiency curves and cut size are shown to be predictable at all feed rates, despite the former dataset being based on a screen with a feed sample size of around 5000 kg per test and the latter dataset obtained using a feed sample size of less than 30 kg per test. The thesis also touches on modelling the screening efficiency curves. A fully dimensionless version of the proven and familiar Whiten screen model is proposed. Scale-up and modelling of industrial screens M. M. Hilden viii Finally, some of the further possibilities of this theory are discussed briefly in a section on further work; these include further applications of the Screening Physical Model and the application of Froude scaling to the modelling of other granular systems.

The role of mathematical models and simulators in describing the performance of mineral processing applications have had a large impact in optimising existing industrial plants and designing new plants over recent years. Before the development of sophisticated computer simulators, the design engineer used industrial “rules of thumb” to estimate the size and layout of plants. However, newly designed plants after commissioning, often do not meet the design product specification requirements and quite often years of “trial and error” optimisation is required. This process can be very costly especially with froth flotation processes where the complexity of the various stages of treatment makes “trial and error” optimisation very difficult to quantify and assess the benefits. Over the past 10 years, advances in the modelling of the flotation process have been conducted by many authors. The most significant flotation modelling advances in recent years have been provided by the AMIRA (Australian Mineral Industry Research Association) P9 project, whereby a new modelling methodology has been proposed. Within this methodology, the flotation response can be represented by a number of sub-processes including parameters describing the hydrodynamic, froth and ore characteristics. Although these parameters have been proposed, methods for measuring these parameters in large flotation cells are still developing, especially in the areas of the froth zone recovery and entrainment. In the light of this it was felt that the literature on froth recovery determination should be investigated to determine the most appropriate method for measuring froth recovery in large industrial flotation cells. It was found after the investigation of the literature that three techniques for measuring the froth recovery parameter stood out as potential methods for measurement within a large scale flotation cell. These were the methods decribed by Gorain et al (1998), Vera et al (1999) and Savassi et al (1997). It was decided that all three methods should be assessed on a quantitative and qualitative basis from data collected at the Mount Isa Mines (now Xstrata) Copper flotation circuit. In this assessment, all three methods were extensively trialed in a 2.8 m3 flotation cell which was operated in parallel to the main copper rougher flotation circuit. The cell could be operated at numerous operating conditions which allowed sufficient data to be collected. The conclusions from this work were that although the method proposed by Vera et al (1999) required significant amounts of data, the method appeared to be reliable in this scale of cell. The main recommendation from this work was to further test the Vera methodology in larger industrial flotation cells. A 100 m3 Outukumpu tank cell at the Mount Keith Nickel Concentrator was chosen for the further assessment of the Vera et al (1999) methodology and its applicability to large scale cells. This flotation cell was one of the largest flotation cells operating on a production scale at the time of the testwork. Numerous tests were conducted and data collected from this investigation showed that the Vera et al (1999) technique was applicable to this scale of flotation cell. Since the work at Mount Keith was conducted in a rougher flotation cell, it was decided to test the methodology with numerous cells of various sizes and duties at the Kambalda Nickel Concentrator. As with the previous investigation at Mount Keith, it was observed that the Vera method was able to measure froth recoveries in all cells measured at Kambalda (within typical operating ranges). However, the technique was not applicable at shallow froth depths since it does not take into account the effect of the pulp-froth interface within the froth recovery parameter estimation. The pulp froth interface and close to it is where a significant proportion of dropback occurs within the froth zone. In addition to this problem, the methodology required large numbers of samples and disturbed downstream processes which made the technique unpractical for operating industrial flotation plants. Hence, a new technique for measuring froth recovery in large flotation cells was required. For the technique to be successful on an industrial scale it required the following: • minimum disturbance on the process, • take into account the pulp froth interface within the froth recovery parameter, and • require a minimum amount of samples. To meet these needs a new technique was developed based on the Savassi et al (1997) technique and combining it with recent work by authors including Vera et al (2003). The methodology involves taking samples of the feed, concentrate and tail as per a typical flotation survey and combining them with two new samples: the air hold-up sample and the top of froth sample. With the addition of these samples, a mass balance across the pulp and froth phase could be conducted and the froth recovery parameter derived. In addition, the new method provided measurements of the pulp zone average bubble load and the amount recovered by the entrainment mechanism. The proposed method has a simple procedure which allows the technique to be used by academics and mill operators alike. The proposed froth recovery measurement technique was tested in numerous cells of various types (i.e. Wemco, Outokumpu, Dorr-Oliver etc), various sizes (up to 150 m3 in size), various duties (rougher, scavenger, cleaner, recleaner, etc) and various plants. In most cases the methodology proved to be a reliable measure of the froth recovery parameter. In addition, at the Century Zinc Operation, the methodology was compared directly with the original Vera et al (1999) technique and the results showed that there was a good comparison between the results with the off-set of the pulp-froth interface. A number of contributions to both the research and industrial areas have been provided from the outcomes of the thesis. The main contributions include: • A full assessment of the three current methods for measuring the froth recovery parameter within large flotation cells. With recommendations of developing a new technique. • The development of a froth recovery measurement technique which can be used in large cells to understand the impact of the froth zone in an individual cell, use within the AMIRA P9 modelling methodology and plant diagnostics. • The new method also allows the estimation of the average bubble load and quantifies the amount of material recovered by the entrainment mechanism which is invaluable to metallurgists in assessing the performance of a flotation circuit (plant diagnostics). Finally, the results of this thesis will provide practising metallurgists both within the research and operating fields, techniques to improve the profitability of flotation circuits worldwide. Metallurgists can quickly assess the performance of large flotation cells in terms of froth performance, bubble load and entrainment which has not been available before. In addition, the results from this thesis will also allow metallurgists to mathematically represent their plant through flotation models better and improve their understanding of their flotation circuits.

The role of mathematical models and simulators in describing the performance of mineral processing applications have had a large impact in optimising existing industrial plants and designing new plants over recent years. Before the development of sophisticated computer simulators, the design engineer used industrial “rules of thumb” to estimate the size and layout of plants. However, newly designed plants after commissioning, often do not meet the design product specification requirements and quite often years of “trial and error” optimisation is required. This process can be very costly especially with froth flotation processes where the complexity of the various stages of treatment makes “trial and error” optimisation very difficult to quantify and assess the benefits. Over the past 10 years, advances in the modelling of the flotation process have been conducted by many authors. The most significant flotation modelling advances in recent years have been provided by the AMIRA (Australian Mineral Industry Research Association) P9 project, whereby a new modelling methodology has been proposed. Within this methodology, the flotation response can be represented by a number of sub-processes including parameters describing the hydrodynamic, froth and ore characteristics. Although these parameters have been proposed, methods for measuring these parameters in large flotation cells are still developing, especially in the areas of the froth zone recovery and entrainment. In the light of this it was felt that the literature on froth recovery determination should be investigated to determine the most appropriate method for measuring froth recovery in large industrial flotation cells. It was found after the investigation of the literature that three techniques for measuring the froth recovery parameter stood out as potential methods for measurement within a large scale flotation cell. These were the methods decribed by Gorain et al (1998), Vera et al (1999) and Savassi et al (1997). It was decided that all three methods should be assessed on a quantitative and qualitative basis from data collected at the Mount Isa Mines (now Xstrata) Copper flotation circuit. In this assessment, all three methods were extensively trialed in a 2.8 m3 flotation cell which was operated in parallel to the main copper rougher flotation circuit. The cell could be operated at numerous operating conditions which allowed sufficient data to be collected. The conclusions from this work were that although the method proposed by Vera et al (1999) required significant amounts of data, the method appeared to be reliable in this scale of cell. The main recommendation from this work was to further test the Vera methodology in larger industrial flotation cells. A 100 m3 Outukumpu tank cell at the Mount Keith Nickel Concentrator was chosen for the further assessment of the Vera et al (1999) methodology and its applicability to large scale cells. This flotation cell was one of the largest flotation cells operating on a production scale at the time of the testwork. Numerous tests were conducted and data collected from this investigation showed that the Vera et al (1999) technique was applicable to this scale of flotation cell. Since the work at Mount Keith was conducted in a rougher flotation cell, it was decided to test the methodology with numerous cells of various sizes and duties at the Kambalda Nickel Concentrator. As with the previous investigation at Mount Keith, it was observed that the Vera method was able to measure froth recoveries in all cells measured at Kambalda (within typical operating ranges). However, the technique was not applicable at shallow froth depths since it does not take into account the effect of the pulp-froth interface within the froth recovery parameter estimation. The pulp froth interface and close to it is where a significant proportion of dropback occurs within the froth zone. In addition to this problem, the methodology required large numbers of samples and disturbed downstream processes which made the technique unpractical for operating industrial flotation plants. Hence, a new technique for measuring froth recovery in large flotation cells was required. For the technique to be successful on an industrial scale it required the following: • minimum disturbance on the process, • take into account the pulp froth interface within the froth recovery parameter, and • require a minimum amount of samples. To meet these needs a new technique was developed based on the Savassi et al (1997) technique and combining it with recent work by authors including Vera et al (2003). The methodology involves taking samples of the feed, concentrate and tail as per a typical flotation survey and combining them with two new samples: the air hold-up sample and the top of froth sample. With the addition of these samples, a mass balance across the pulp and froth phase could be conducted and the froth recovery parameter derived. In addition, the new method provided measurements of the pulp zone average bubble load and the amount recovered by the entrainment mechanism. The proposed method has a simple procedure which allows the technique to be used by academics and mill operators alike. The proposed froth recovery measurement technique was tested in numerous cells of various types (i.e. Wemco, Outokumpu, Dorr-Oliver etc), various sizes (up to 150 m3 in size), various duties (rougher, scavenger, cleaner, recleaner, etc) and various plants. In most cases the methodology proved to be a reliable measure of the froth recovery parameter. In addition, at the Century Zinc Operation, the methodology was compared directly with the original Vera et al (1999) technique and the results showed that there was a good comparison between the results with the off-set of the pulp-froth interface. A number of contributions to both the research and industrial areas have been provided from the outcomes of the thesis. The main contributions include: • A full assessment of the three current methods for measuring the froth recovery parameter within large flotation cells. With recommendations of developing a new technique. • The development of a froth recovery measurement technique which can be used in large cells to understand the impact of the froth zone in an individual cell, use within the AMIRA P9 modelling methodology and plant diagnostics. • The new method also allows the estimation of the average bubble load and quantifies the amount of material recovered by the entrainment mechanism which is invaluable to metallurgists in assessing the performance of a flotation circuit (plant diagnostics). Finally, the results of this thesis will provide practising metallurgists both within the research and operating fields, techniques to improve the profitability of flotation circuits worldwide. Metallurgists can quickly assess the performance of large flotation cells in terms of froth performance, bubble load and entrainment which has not been available before. In addition, the results from this thesis will also allow metallurgists to mathematically represent their plant through flotation models better and improve their understanding of their flotation circuits.

The role of mathematical models and simulators in describing the performance of mineral processing applications have had a large impact in optimising existing industrial plants and designing new plants over recent years. Before the development of sophisticated computer simulators, the design engineer used industrial “rules of thumb” to estimate the size and layout of plants. However, newly designed plants after commissioning, often do not meet the design product specification requirements and quite often years of “trial and error” optimisation is required. This process can be very costly especially with froth flotation processes where the complexity of the various stages of treatment makes “trial and error” optimisation very difficult to quantify and assess the benefits. Over the past 10 years, advances in the modelling of the flotation process have been conducted by many authors. The most significant flotation modelling advances in recent years have been provided by the AMIRA (Australian Mineral Industry Research Association) P9 project, whereby a new modelling methodology has been proposed. Within this methodology, the flotation response can be represented by a number of sub-processes including parameters describing the hydrodynamic, froth and ore characteristics. Although these parameters have been proposed, methods for measuring these parameters in large flotation cells are still developing, especially in the areas of the froth zone recovery and entrainment. In the light of this it was felt that the literature on froth recovery determination should be investigated to determine the most appropriate method for measuring froth recovery in large industrial flotation cells. It was found after the investigation of the literature that three techniques for measuring the froth recovery parameter stood out as potential methods for measurement within a large scale flotation cell. These were the methods decribed by Gorain et al (1998), Vera et al (1999) and Savassi et al (1997). It was decided that all three methods should be assessed on a quantitative and qualitative basis from data collected at the Mount Isa Mines (now Xstrata) Copper flotation circuit. In this assessment, all three methods were extensively trialed in a 2.8 m3 flotation cell which was operated in parallel to the main copper rougher flotation circuit. The cell could be operated at numerous operating conditions which allowed sufficient data to be collected. The conclusions from this work were that although the method proposed by Vera et al (1999) required significant amounts of data, the method appeared to be reliable in this scale of cell. The main recommendation from this work was to further test the Vera methodology in larger industrial flotation cells. A 100 m3 Outukumpu tank cell at the Mount Keith Nickel Concentrator was chosen for the further assessment of the Vera et al (1999) methodology and its applicability to large scale cells. This flotation cell was one of the largest flotation cells operating on a production scale at the time of the testwork. Numerous tests were conducted and data collected from this investigation showed that the Vera et al (1999) technique was applicable to this scale of flotation cell. Since the work at Mount Keith was conducted in a rougher flotation cell, it was decided to test the methodology with numerous cells of various sizes and duties at the Kambalda Nickel Concentrator. As with the previous investigation at Mount Keith, it was observed that the Vera method was able to measure froth recoveries in all cells measured at Kambalda (within typical operating ranges). However, the technique was not applicable at shallow froth depths since it does not take into account the effect of the pulp-froth interface within the froth recovery parameter estimation. The pulp froth interface and close to it is where a significant proportion of dropback occurs within the froth zone. In addition to this problem, the methodology required large numbers of samples and disturbed downstream processes which made the technique unpractical for operating industrial flotation plants. Hence, a new technique for measuring froth recovery in large flotation cells was required. For the technique to be successful on an industrial scale it required the following: • minimum disturbance on the process, • take into account the pulp froth interface within the froth recovery parameter, and • require a minimum amount of samples. To meet these needs a new technique was developed based on the Savassi et al (1997) technique and combining it with recent work by authors including Vera et al (2003). The methodology involves taking samples of the feed, concentrate and tail as per a typical flotation survey and combining them with two new samples: the air hold-up sample and the top of froth sample. With the addition of these samples, a mass balance across the pulp and froth phase could be conducted and the froth recovery parameter derived. In addition, the new method provided measurements of the pulp zone average bubble load and the amount recovered by the entrainment mechanism. The proposed method has a simple procedure which allows the technique to be used by academics and mill operators alike. The proposed froth recovery measurement technique was tested in numerous cells of various types (i.e. Wemco, Outokumpu, Dorr-Oliver etc), various sizes (up to 150 m3 in size), various duties (rougher, scavenger, cleaner, recleaner, etc) and various plants. In most cases the methodology proved to be a reliable measure of the froth recovery parameter. In addition, at the Century Zinc Operation, the methodology was compared directly with the original Vera et al (1999) technique and the results showed that there was a good comparison between the results with the off-set of the pulp-froth interface. A number of contributions to both the research and industrial areas have been provided from the outcomes of the thesis. The main contributions include: • A full assessment of the three current methods for measuring the froth recovery parameter within large flotation cells. With recommendations of developing a new technique. • The development of a froth recovery measurement technique which can be used in large cells to understand the impact of the froth zone in an individual cell, use within the AMIRA P9 modelling methodology and plant diagnostics. • The new method also allows the estimation of the average bubble load and quantifies the amount of material recovered by the entrainment mechanism which is invaluable to metallurgists in assessing the performance of a flotation circuit (plant diagnostics). Finally, the results of this thesis will provide practising metallurgists both within the research and operating fields, techniques to improve the profitability of flotation circuits worldwide. Metallurgists can quickly assess the performance of large flotation cells in terms of froth performance, bubble load and entrainment which has not been available before. In addition, the results from this thesis will also allow metallurgists to mathematically represent their plant through flotation models better and improve their understanding of their flotation circuits.

The role of mathematical models and simulators in describing the performance of mineral processing applications have had a large impact in optimising existing industrial plants and designing new plants over recent years. Before the development of sophisticated computer simulators, the design engineer used industrial “rules of thumb” to estimate the size and layout of plants. However, newly designed plants after commissioning, often do not meet the design product specification requirements and quite often years of “trial and error” optimisation is required. This process can be very costly especially with froth flotation processes where the complexity of the various stages of treatment makes “trial and error” optimisation very difficult to quantify and assess the benefits. Over the past 10 years, advances in the modelling of the flotation process have been conducted by many authors. The most significant flotation modelling advances in recent years have been provided by the AMIRA (Australian Mineral Industry Research Association) P9 project, whereby a new modelling methodology has been proposed. Within this methodology, the flotation response can be represented by a number of sub-processes including parameters describing the hydrodynamic, froth and ore characteristics. Although these parameters have been proposed, methods for measuring these parameters in large flotation cells are still developing, especially in the areas of the froth zone recovery and entrainment. In the light of this it was felt that the literature on froth recovery determination should be investigated to determine the most appropriate method for measuring froth recovery in large industrial flotation cells. It was found after the investigation of the literature that three techniques for measuring the froth recovery parameter stood out as potential methods for measurement within a large scale flotation cell. These were the methods decribed by Gorain et al (1998), Vera et al (1999) and Savassi et al (1997). It was decided that all three methods should be assessed on a quantitative and qualitative basis from data collected at the Mount Isa Mines (now Xstrata) Copper flotation circuit. In this assessment, all three methods were extensively trialed in a 2.8 m3 flotation cell which was operated in parallel to the main copper rougher flotation circuit. The cell could be operated at numerous operating conditions which allowed sufficient data to be collected. The conclusions from this work were that although the method proposed by Vera et al (1999) required significant amounts of data, the method appeared to be reliable in this scale of cell. The main recommendation from this work was to further test the Vera methodology in larger industrial flotation cells. A 100 m3 Outukumpu tank cell at the Mount Keith Nickel Concentrator was chosen for the further assessment of the Vera et al (1999) methodology and its applicability to large scale cells. This flotation cell was one of the largest flotation cells operating on a production scale at the time of the testwork. Numerous tests were conducted and data collected from this investigation showed that the Vera et al (1999) technique was applicable to this scale of flotation cell. Since the work at Mount Keith was conducted in a rougher flotation cell, it was decided to test the methodology with numerous cells of various sizes and duties at the Kambalda Nickel Concentrator. As with the previous investigation at Mount Keith, it was observed that the Vera method was able to measure froth recoveries in all cells measured at Kambalda (within typical operating ranges). However, the technique was not applicable at shallow froth depths since it does not take into account the effect of the pulp-froth interface within the froth recovery parameter estimation. The pulp froth interface and close to it is where a significant proportion of dropback occurs within the froth zone. In addition to this problem, the methodology required large numbers of samples and disturbed downstream processes which made the technique unpractical for operating industrial flotation plants. Hence, a new technique for measuring froth recovery in large flotation cells was required. For the technique to be successful on an industrial scale it required the following: • minimum disturbance on the process, • take into account the pulp froth interface within the froth recovery parameter, and • require a minimum amount of samples. To meet these needs a new technique was developed based on the Savassi et al (1997) technique and combining it with recent work by authors including Vera et al (2003). The methodology involves taking samples of the feed, concentrate and tail as per a typical flotation survey and combining them with two new samples: the air hold-up sample and the top of froth sample. With the addition of these samples, a mass balance across the pulp and froth phase could be conducted and the froth recovery parameter derived. In addition, the new method provided measurements of the pulp zone average bubble load and the amount recovered by the entrainment mechanism. The proposed method has a simple procedure which allows the technique to be used by academics and mill operators alike. The proposed froth recovery measurement technique was tested in numerous cells of various types (i.e. Wemco, Outokumpu, Dorr-Oliver etc), various sizes (up to 150 m3 in size), various duties (rougher, scavenger, cleaner, recleaner, etc) and various plants. In most cases the methodology proved to be a reliable measure of the froth recovery parameter. In addition, at the Century Zinc Operation, the methodology was compared directly with the original Vera et al (1999) technique and the results showed that there was a good comparison between the results with the off-set of the pulp-froth interface. A number of contributions to both the research and industrial areas have been provided from the outcomes of the thesis. The main contributions include: • A full assessment of the three current methods for measuring the froth recovery parameter within large flotation cells. With recommendations of developing a new technique. • The development of a froth recovery measurement technique which can be used in large cells to understand the impact of the froth zone in an individual cell, use within the AMIRA P9 modelling methodology and plant diagnostics. • The new method also allows the estimation of the average bubble load and quantifies the amount of material recovered by the entrainment mechanism which is invaluable to metallurgists in assessing the performance of a flotation circuit (plant diagnostics). Finally, the results of this thesis will provide practising metallurgists both within the research and operating fields, techniques to improve the profitability of flotation circuits worldwide. Metallurgists can quickly assess the performance of large flotation cells in terms of froth performance, bubble load and entrainment which has not been available before. In addition, the results from this thesis will also allow metallurgists to mathematically represent their plant through flotation models better and improve their understanding of their flotation circuits.