Quantification and mitigation of segregation in the handling of alumina in aluminium production
This thesis addresses the development, evaluation and modelling of an anti-segregation system ("AS-System") for.use in larger silos handling alumina in the aluminium smelting industry. This work is unlike much of what has gone before because it is much more grounded in the technical and economic consequences of segregation for a particular manufacturing process. Segregation of particulates, i.e. separation of components due to differences in properties such as size, density etc., has an extensive literature going back as far as 1915, but this focuses mainly on the segregation process itself and largely ignores the context and the consequences. The consequence of segregation is loss of homogeneity; the impact that has on any given process, is generally not addressed very deeply. Surveying the literature on segregation, and studying the total processes of aluminium production, creates a basis for understanding the importance of segregation for aluminium production, as well as the importance of powder technology in general for this industry. A method for quantification of segregation in this production process, based on sampling, has been established. This was intended to give the fundamental information necessary for measuring the extent of the problem and the degree of improvement achieved. By placing sampling points along the logistic chain for the alumina, and by sampling these points for a long enough period, information about the influence of the various handling steps on the bulk solids can be identified. Segregation by particle size is the main type of influence, although the work has shown that attrition is another. Standardised statistical expressions have been used for analysing the bottlenecks of the logistic loops, and study of the results has led to a useful way of expressing the level of segregation, the change of segregation level in a handling step (silo filling and discharge), and improvements in this change. The degree of segregation when handling alumina has proven to be quite considerable in terms of effects on the production process. The effects on the efficiency of the aluminium smelting process, and the environment, have both been evaluated. Variations in the alumina due to segregation have been found to correlate with both dust concentration in the smelter pot room, and anode effects (an unwanted upset in the smelting process). An economical evaluation of an investment in anti segregation systems has been made. This evaluation has shown significant economic consequences, clearly justifying both the investigations of segregation, and the implementation of anti segregation measures. To remedy the effects of air current segregation in the aluminium industry, a complete Anti Segregation System (AS-System) based around Anti Segregation Tubes (AST) utilising a special inlet configuration has been developed. In a number of full size installations, this solution has proved itself capable of handling the variations and transients of process conditions which occur in the industry, with a large operational capacity range, due to the special inlet configuration. The effect of the full scale installed anti segregation systems have been measured, and compared to other systems. The AS-System clearly demonstrated a homogenising effect no matter how low the ingoing variations in particle size were. Other commercial systems which have been evaluated have turned out to be no more than Segregation Effect Damping Systems, since they only seem to reduce the segregation effect after it has happened, instead of trying to eliminate the problem by directly attacking the segregation mechanism itself which is what the AS-System does. A new scaled down test rig for the AS-System was developed, consisting of three ASTs fed from a central distributor. To test the potential effect of the AS-System, to determine its efficiency in countering segregation, tests with repeated filling and emptying of a scale silo were carried out. The results clearly showed that the AS-System very much reduced segregation, compared to conventional filling, even in a small-scale silo. Improved models have been developed for the function of the AST and these have been verified against measurements from the new test rig. The early version of the test rig for the AST used only one centre-mounted tube, with one pressure measurement in the top of the tube. The first models were based on the assumption of the pressure being linear, and assumed full dispersion of the falling material inside the tube; and calibrated from the single pressure measurement inside at the top of the tube. Although this simple model calculated very conservative values of the negative pressure, the model was used for the initial development and design of the AST, and later the ASSystem (Anti Segregation System). A second model was derived, where the material velocity was calculated based on free fall. This model was also based on the idea of full dispersion, but was in better agreement with measured values during further tests, which showed a considerable deviation from the original assumption of a linear pressure distribution once intermediate pressure measurements were available. When using the multi-phase-flow-simulation-program-code FLUENT to simulate the pressure distribution of the AST, the results were quite disappointing, however the FLUENT program was able to identify an initial positive pressure generated by the flow from the inlet box to the tube. Implementing this initial pressure into the simple non-linear model above, both the trend and values correspond quite well with the measured values. A single particle drag model was tried, but dismissed after calculating the maximum possible capacity for known tubes and finding the predictions to be unrealistically low. A new approach was introduced, modelling the fall of the powder in a continuous layer along the inner wall of the tube on one side, creating skin drag along the surface between the falling solid powder and the air. The length of the tube and the width of the chosen AST profile define this surface. This approach assumes that the powder falls like a layer along the tube wall. Previous theory for pressure drop in pneumatic conveying inspired this approach, but it had not previously been used for gravity flow in vertical tubes, and as a result the novel Solid Surface Body Drag Model (SSBDM) was developed. This analytical model gives very good correspondence with the measured data for the pressure distribution inside the AST, yet is extremely simple to use. When comparing the model with measured data, the SSBDM was able to predict the pressure distribution within the error boundaries of the test measurements. A method for design of the AST was derived from the SSBDM, using a dimensionless parameter function determined for the pressure drop model. The models giving the design indicate that the capacity is more than proportional to the cross sectional area of a chosen tube profile, which is in agreement with observations. The model suggests that the capacity is proportional to the cross sectional area in the power of 1.25. This model allowed the study of the effect of tube shape, which revealed that a square profile for the AST does not seem to be the optimum design; rather, a rectangular profile should be chosen for maximum capacity. The model suggests that the capacity is proportional to the width of the side of the tube along which the powder layer is falling, but proportional to the perpendicular side in a rectangular profile in the power of 1.5. The model gives an equation for pressure drop which can also be utilised to place the first valve on the tube. It also shows that for high capacities, and large silos, a system consisting of several ASTs should be chosen (AS-System). Predictions from the model have been tested against the measured capacities of full scale installed systems and give good agreement. Overall, the AS-System has been shown to be cost-effective in reducing segregation; results measured from the full scale installations show a homogenising factor (reduction in variation of the material being handled) of 1-1.5. As a result of these verifications and the simplicity of the model presented in this thesis, the plant engineer can confidently design a system which will function correctly and make a positive, predictable improvement in the homogeneity of the alumina in his plant.