Hydrodynamic performance of a biological sulphate reduction reactor treating high-sulphate acid mine drainage

Abstract

This master's degree was focused on investigating the evolution of the hydraulic behaviour of a packed bed reactor over time and the effects of these changes on the sulphate removal performance of the reactor. This was achieved by monitoring the hydraulic behaviour over the duration of the study with tracer tests to observe any changes in the flow patterns through the reactor. The physical structure of the packing material was investigated by stabilising and sectioning the packed bed. Computational fluid dynamic modelling of the flow through the reactors was done to further understand the experimental residence time distribution results. This work then enabled the determination of a potential link between the hydraulic behaviour and the sulphate reducing performance of the process, as the structure of the packed bed evolved over time. Eleven reactors were set up by using packing material that consisted of a specified mix of wood chips, wood shavings, hay, lucerne and kraal manure. The reactors had an average operating volume of 9.7 L. During the 30-day inoculation period, the sulphate concentrations were lowered on average to below 400 mg/L from initial concentrations of 3450 mg/L. The reactors' process parameters after the 30-day inoculation period led to the decision that adequate inoculation occurred, which meant that sufficient biofilms had established to commence continuous operation. The reactors were continuously operated at a 10-day hydraulic retention time at a constant temperature of 35 °C. The reactors were operated for different durations as several reactors were sent for physical analysis of the packed bed structure. Continuous operation lasted for 364 days, by which the end of the study was reached. The alkalinity, chemical oxygen demand, electrical conductivity, pH, sulphate and sulphide of the effluent were measured twice a week. The pH readings of the treated water were maintained close to neutral, which was in the expected range for a biological sulphate reduction (BSR) system treating mine water. The oxidation-reduction potential readings of the reactor's effluents were maintained below -240 mV throughout the study, and below -300 mV during the high-performance phase. The electrical conductivity values of the effluent remained relatively constant compared to the feed, which indicated that there was minimal removal of dissolved chemicals/salts and minimal production of carbonates throughout the process. All the reactors demonstrated similar sulphate removal performance trends exhibiting distinct phases. The lag phase took place from start-up to day 70, the high-performance phase was observed from day 71 to day 120, the crash phase occurred from day 121 to day 130, and the sustained phase lasted from day 131 to day 364. A peak sulphate reduction rate of 224 g/m3/d with an average rate of 116 g/m3/d was achieved during the high-performance phase. The average sulphate-removal rate in this study was similar to the removal rates achieved in the large-scale integrated management passive (IMPI) and cloSURE® plants, which were 192 g/m3/d and 233 g/m3/d, respectively. The lower removal rates achieved during this study were attributed to the lower amounts of organic carbon in the packing material, and the continuous addition of supplemental substrate in the large-scale tests. The similarity of the performance indicators of all the reactors indicated high reproducibility and demonstrated the possibility of repeating the results in an upscaling of the process in field operations. The sequential tracer tests indicated that the tanks-in-series model fitted the experimental tracer responses very well and was therefore used to observe the change of the flow behaviour over the operational periods of the reactors. In general, the mean residence time tended to decrease over time, suggesting that the flow regime developed gradually, indicating the possibility of channelling and bypassing in portions of the reactor. The observed ranges of the residence time distributions among the reactors suggested that the individual packing structures had a significant impact on their respective hydraulic behaviours. The structure of the packed bed was further investigated through physical analysis and computational fluid dynamic modelling of the process. The images produced by the physical analysis experiments showed the presence of channels and bypassing occurring in some sections of the packed bed. At the top end of the reactors, a relatively even distribution of flow was observed, with a reduction of the flow through the middle of the packed bed, as more flow occurred closer to the walls. The computational fluid dynamic modelled tracer response peaks did not completely overlap, but the shape of the tracer responses exhibited similar behaviours, with the one reactor's residence time distribution peaking earlier with a narrower spread, and the other reactors' residence time distribution curves peaking later with a wider spread. Sensitivity analysis of the diffusion coefficient showed that with a 50-fold increase, similar residence time distribution responses were achieved. These results highlighted the importance of investigating a tracer's properties in new applications, as this could have a large impact on modelling the predicted flows in a packed bed reactor. The results of this study indicated that the hydraulic behaviour and the permeability of the packed bed during the operational periods of the reactors had a negligible impact on the sulphate removal rates, whereas the availability of organic substrate was the determining factor for attaining high rates of sulphate removal. Providing sufficient organic carbon and ensuring effective transportation of the substrate to the bacteria are major considerations that need to be included as part of the design criteria of a biological sulphate reduction plant to ensure successful implementation of the process at scale.