Coal Preparation » Environmental Improvement
The background for the project was that Bulga has installed the first paste thickener for coal tailings in Australia. The paste thickener was installed to thicken tailings sufficiently to handle it on the reject conveyor. During commissioning and operation it was determined that the short term fluctuations in thickener feed properties caused unstable and variable operation. The variations in thickener feed properties are caused by mining a large number of different sections and seams, both open cut and underground. The variable feed material caused the thickener product variations and at times it was not possible to obtain the required underflow densities for further handling which were above the design target of 50% underflow solids.
The project was designed to establish relationships between feed qualities and operating parameters that result in more stable paste thickener operation with high underflow densities.
The objective of the study was to explore scale-up possibilities, the effect of feed material and operational outcomes of applying different paste thickener sizes at various operating conditions to an Australian coal operation. Paste thickening testwork at different thickener scales (laboratory, mini-pilot, pilot) investigated the effects of tailings material properties, operational factors and floc dosage on dewatering characteristics to determine the scale sufficient for predicting and optimising full scale paste thickener performance.
Extensive thickening testwork on-site at Bulga with 0.1, 0.19 and 1m diameter thickeners were performed over an extended period in separate test series, each lasting a few days. Concurrently, coal tailings samples from different seams were analysed in regards to their chemical and mineralogical composition and their particle size and zeta potential determined.
Extensive analysis of the large amount of different thickener testwork series at different thickener scale, different dates and different feed material (coal seams) including linear and non-linear regression analysis could only reveal some general correlations but no quantitative description or prediction of paste thickener performance based on operational parameters. The bed depth seemed most significant with expectantly higher beds producing denser underflow. Increasing the bed depth on the 1m pilot thickener from 1.5 to 6m increased the underflow solids by average 4% points for the particular solids loading rates applied.
Some test series revealed further quantitative dependencies. Increasing the rake speed from 2 to 4rpm improved paste thickening on the 0.1m scale thickener, while the 0.19m showed an increase from 1 to 2 rpm and a drop in underflow solids thereafter. As expected, lower feed solids result in better paste thickening for all scales, assumable through better flocculation and/or increased residence time. Floc addition requires optimal dosage for maximising underflow solids, while no conclusive relationship could be derived.
Lower solids flux also improved underflow density on the 0.1m and 1m thickener sizes. However, not all findings were conclusive and even the extensive regression analysis could not produce conclusive quantitative relationships between individual operating parameters and the underflow solids. The variations in feed material caused considerable fluctuations of results with the time delay within the thickener for some operating parameters (such as feed solids and floc dosage) and provided further complications. Allowing for a residence time determined experimentally and by approximation of 12h in the pilot and full scale thickener did not improve the regression analysis results.
Bed rotation in particular at higher floc dosage was present in the smaller (0.1 and 0.19m) thickeners which results in lower underflow solids content. The 1m pilot thickener was limited through its underflow pump in producing a very viscous paste. Thus, scale up is only feasible with a paste thickener and its ancillaries, in particular the underflow pump, being able to handle the paste material. Still, shear during handling is higher on smaller scale, so that the yield stress for the full scale thickener can possibly only be predicted by removing a sample of compacted sediment carefully from a pilot scale thickener without applying too much shear.
The torque measured on the full scale paste thickener relates reasonably well to its underflow solids, which possibly provides a method for thickener control. Improvements in paste thickener operations are believed possible with enhanced dynamic control systems.
While the project resulted in qualitative results for the influence of operating parameters on the performance of paste thickeners for coal tailings, in particular underflow solids content, no qualitative relationships could be proven. Thus, pilot scale can predict the behaviour of coal tailings within a full scale thickener qualitatively but not accurately. The main reason is the fluctuation in feed properties. In particular the high underflow yield stress experienced in the full scale thickener could not been replicated on pilot scale.
While the feed properties could be controlled well on the smaller scale thickeners, they did not allow full scale bed depths and thus, had scale-up limitations. The larger pilot scale thickener (1m) allows sufficient bed depth but is depending on plant feed which is variable and thus, introduces significant fluctuations. Thus, non-steady-state operation and variation of input parameters make small scale testwork results almost impossible to interpret in regards to quantitative single parameter effects.