Abstract: Grootvlei Proprietary Mines Ltd is discharging between 80 and 100 Ml/d underground water into the Blesbokspruit. This water is pumped out of the mine to keep the underground water at such a level as to make mining possible. The water is of poor quality because it contains high TDS levels (2700-3800 mg/l) including high concentrations of iron, manganese, sulphate, calcium, magnesium, sodium and chloride. This water will adversely affect the water ecology in the Blesbokspruit, and it will significantly increase the TDS concentration of one of the major water resources if not treated prior to disposal into the stream. Therefore, alternative water desalination technologies were evaluated to estimate performance and the economics of the processes for treatment of the mine water. It was predicted that water of potable quality should be produced from the mine water with spiral reverse osmosis (SRO). It was demonstrated that it should be possible to reduce the TDS of the mine water (2000-2700-3400-4500 mg/l) to potable standards with SRO (85% water recovery). The capital costs (pretreatment and desalination) for a 80 Ml/d plant (worst-case water) were estimated at US$35M. Total operating costs were estimated at 88.1c/kl. Brine disposal costs were estimated at US$18M. Therefore, the total capital costs are estimated at US$53M. It was predicted that it should be possible to produce potable water from the worst-case feed water (80 Ml/d) with the EDR process. It was demonstrated that the TDS in the feed could be reduced from 4178 to 246 mg/l in the EDR product (65% water recovery). The capital costs (pretreatment plus desalination) to desalinate the worst-case feed water to potable quality with EDR is estimated at US$53.3M. The operational costs are estimated at 47.6 c/kl. Brine disposal costs were estimated at US$42M. Therefore, the total capital costs are estimated at US$95.3 M. It was predicted that it should be possible to produce potable water from the mine water with the GYP-CIX ion- exchange process. It was demonstrated that the feed TDS (2000- 4500 mg/l) could be reduced to less than 240 mg/l (54% water recovery for the worst-case water). The capital cost for an 80 Ml/d ion-exchange plant (worst-case water) was estimated at US$26.7M (no pretreatment). Operational costs were estimated at 60.4 c/kl. Brine disposal costs were estimated at US$55.1M. Therefore, the total desalination costs were estimated at US$81.8M. The capital outlay for a SRO plant will be significantly less than that for either an EDR or a GYP-CIX plant. The operating costs, however, of the RO plant are significantly higher than for the other two processes. Potable water sales, however, will bring more in for the RO process than for the other two processes because a higher water recovery can be obtained with RO. The operating costs minus the savings in water sales were estimated at 17.2; 6.7 and US$8.6M/y for the RO, EDR and GYP-CIX processes, respectively (worst case). Therefore, the operational costs of the EDR and GYP-CIX processes are the lowest if the sale of water is taken into consideration. This may favour the EDR and GYP-CIX processes for the desalination of the mine water.

Abstract: The effectiveness of wetland treatment of acid mine drainage (AMD) was assessed using three measures of performance: treatment efficiency, area-adjusted removal, and first-order removal. Mathematical relationships between these measures were derived from simple kinetic equations. Area-adjusted removal is independent of pollutant concentration (zero-order reaction kinetics), while first-order removal is dependent on concentration. Treatment efficiency is linearly related to area-adjusted removal and exponentially related to first-order removal at constant hydraulic loading rates (flow/area). Examination of previously published data from 35 natural AMD wetlands revealed that statistically significant correlations exist between several of the performance measures for both iron and manganese removal, but these correlations are potentially spurious because these measures are derived from, and are mathematical rearrangements of, the same operating data. The use of treatment efficiency as a measure of performance between wetlands is not recommended because it is a relative measure that does not account for influent concentration differences. Area-adjusted removal accounts for mass loading effects, but it fails to separate the flow and concentration components, which is necessary if removal is first-order. Available empirical evidence suggests that AMD pollutant removal is better described by first-order kinetics. If removal is first-order, the use of area-adjusted rates for determining the wetland area required for treating relatively low pollutant concentrations will result in undersized wetlands. The effects of concentration and flow rate on wetland area predictions for constant influent loading rates also depend on the kinetics of pollutant removal. If removal is zero-order, the wetland area required to treat a discharge to meet some target effluent concentration is a decreasing linear function of influent concentration (and an inverse function of flow rate). However, if removal is first-order, the required wetland area is a non-linear function of the relative influent concentration. Further research is needed for developing accurate first-order rate constants as a function of influent water chemistry and ecosystem characteristics in order to successfully apply the first-order removal model to the design of more effective AMD wetland treatment systems.

Abstract: Rapid oxidation and accretion of iron onto high surface area media has been investigated as a potential passive treatment option for ferruginous, net-alkaline minewaters. Two pilot-scale reactors were installed at a site in County Durham, UK. Each 2.0m high cylinder contained different high surface area plastic trickling filter media. Ferruginous minewater was fed downwards over the media at various flow-rates with the objective of establishing the efficiency of iron removal at different loading rates. Residence time of water within the reactors was between 70 and 360s depending on the flow-rate (1 and 12l/min, respectively). Average influent total iron concentration for the duration of these experiments was 1.43mg/l (range 1.08-1.84mg/l; n=16), whilst effluent iron concentrations averaged 0.41mg/l (range 0.20-1.04mg/l; n=15) for Reactor A and 0.38mg/l (range 0.11-0.93mg/l; n=16) for Reactor B. There is a strong correlation between influent iron load and iron removal rate. Even at the highest loading rates (approximately 31.6g/day) 43% and 49% of the total iron load was removed in Reactors A and B, respectively. At low manganese loading rates (approximately 0.50-0.90g/day) over 50% of the manganese was removed in Reactor B. Iron removal rate (g/m3/d) increases linearly with loading rate (g/day) up to 14g/d and the slope of the line indicates that a mean of 85% of the iron is removed. In conclusion, it appears that the oxidation and accretion of ochre on high surface area media may be a promising alternative passive technology to constructed wetlands at certain sites.