Groundwater contamination as a result of Light Non Aqueous Phase Liquid (LNAPL) releases into the subsurface is a widespread occurrence across South Africa which threatens current and future water resources within the country. Groundwater contaminant fate and transport modelling are common elements of hydrogeological investigations and remedial design methodologies in many developed countries where the models are used as management and decision making tools. In South Africa this is not the case, with contaminant flow and transport modelling rarely being employed as part of LNAPL contamination investigations. Over the last three years, the Beaufort West study area has had extensive investigative work carried out with regards to the determination and delineation of LNAPL related groundwater contaminant plumes which are present underneath a significant portion of the town. As a result, an extensive data set has been generated with regards to aquifer geometry, fracture network distribution, aquifer parameters and contaminant plume concentrations and extent. The dataset should in theory provide an opportunity to construct a groundwater contaminant fate and transport model for the area as a remedial management tool. By means of collating previously existing data through a comprehensive desktop study, and supplementing this data with a toolkit of field investigation techniques such as diamond barrel core drilling, percussion drilling, electrical conductivity profiling, fluid electrical conductivity profiling, aquifer pump testing, and low flow inorganic and organic groundwater sampling, the conceptual model of the study area was updated and refined to a point where the feasibility of constructing a groundwater contaminant fate and transport model could be assessed. Based upon the conceptual understanding of the study area as defined in the conceptual model developed in the study, a groundwater contaminant fate and transport model is not considered feasible for the study area with body of data currently available. This is attributed mainly to the high level of complexity of the observed natural environment and the challenges in acquiring acceptable quality field data such as aquifer parameters given the uncontrolled pumping environment which is present due to the high number of private groundwater users. Potentially an even greater detractor to the construction of a model is that considering the conceptual understanding of the study area, there are very few questions of significance whose answers could be provided by a model, and this would indicate that a model would not be an effective remedial management or decision making tool in the current scenario

Eight of the 21 Ramsar-designated wetlands in South Africa are located in similar geohydrological settings as Groenvlei, a 359 ha lacustrine wetland found east of Sedgefield in the southern Cape. Groenvlei is unique as it is isolated from the sea and neither fed nor drained by rivers. Consequently, the lake is fed only by rainfall and groundwater inflow. Losses comprise evaporation and groundwater outflow. These characteristics result in a relatively uncomplicated hydrological system that allows for the geohydrological component to be quantified and understood. Using climatic and lake data monitored by the Department of Water Affairs and geohydrological data collected over a period of a decade, research was conducted to quantify the groundwater contribution to the system and develop an improved understanding of the hydrology of Groenvlei. A daily water balance based on rainfall, adjusted S pan evaporation data and lake levels was used to compute that the nett groundwater contribution to Groenvlei amounted to about 0.3 mm/d. It was shown that S pan evaporation data adjusted by coefficients prescribed by Midgley et al. (1994) should be used to quantify lake evaporation, and that the reed collar transpired 10% to 30% more during summer than evaporated from open water. No water is transpired by the reed collar in winter as the vegetation is dormant. Integrating the water balance results with steady-state Darcian flow calculations and a chemical mass balance indicated direct rainfall (71.6%) and groundwater inflow along the western and northern boundaries of the lake (28.4%) constituted inflow into the system. This is balanced by evaporation from open water (61.7%), transpiration from the reed collar in summer (21.4%) and groundwater outflow along the southern boundary (16.9%). This latter component invalidates claims that Groenvlei is endorheic in character. Recharge to the Eden Primary aquifer was estimated to be in the order of 20% MAP. It was calculated only 5.7% of rainfall in the lake catchment discharges into the lake. The balance of rain entering the subsurface is lost through terrestrial evaporation or discharges into the sea via the deeper part of the aquifer. It was interpreted that the deceptively thick vadose zone plays a buffering role in the hydrology of the area and that evapotranspiration losses are appreciable. The importance of the reed collar was further exemplified by the retention of salts in the vegetative fringe. Salts are assimilated by the vegetation and retained in the hyporheic zone until re-entrained into the main water body through wind and wave action. This results in only part of the salt load leaving the lake along the southern boundary and affecting groundwater quality between the lake and the sea. Further research is required to confirm this. The results of the research allowed for tools to be developed to assess the impact of groundwater abstraction from the lakeâs catchment on lake levels and water quality. These tools could also be used to demonstrate Groenvlei has long since lost its connection to the marine or estuarine environments, with a new equilibrium being reached within 120 years of disconnect. The young lake is dynamic in character and rapidly responds to hydrological change. In its short history, Groenvlei has adapted and responded to changes in both sea level and climate, collectively resulting in the present-day system. In addition to highlighting the importance of sound conceptualisation, data quality and a convergence of evidence, the outcomes of this study challenged the findings of Roetsâ (2008) PhD research and found no scientific evidence to support his contention that Groenvlei is sustained by underlying Table Mountain Group aquifers. It was also found that the permeability south of Groenvlei is not low and the extent of the lake catchment is 25 km2. Past research of Groenvlei has resulted in a number of misconceptions and it was argued a need exists to link hydrologists and ecologists to better understand wetlands, with each contributing specific skills and knowledge. An important contribution of the research documented in this thesis is that the approach used can be applied to similar wetlands where the role of groundwater might be less obvious because of river flows and tidal exchange. The importance of sound conceptualization and direct rainfall onto wetlands, quantification of evaporative losses using S pan data and coefficients prescribed by Midgley et al. (1994), and the relationship between open water losses and transpiration losses are three aspects that could improve the understanding and quantification of lake â groundwater interaction elsewhere. A limitation to understanding the geohydrology of Groenvlei is the lack of information pertaining to aquifer thickness. It is therefore recommended four boreholes be drilled to either bedrock or at least 100 m in depth (whichever is reached first) to quantify the thickness of the aquifer. Other limitations that require attention include: ï· Uneven spatial distribution of the geohydrological data; ï· Lack of information of losses from open water and the reed collar; and ï· Absence of monitored groundwater data needed to address temporal relationships between groundwater and the lake.

The significance of a reliable groundwater resource assessment is of growing importance as water resources are stretched to accommodate the growing population. An essential component of a groundwater resource assessment is the quantification of surface water â groundwater interaction. The insufficient amount of data in South Africa and the apparent lack of accuracy of current estimates of the groundwater component of baseflow lead to the investigation of a new methodology. The applicability of the Mixing Cell Model (MCM) to quantify the groundwater contribution to baseflow is examined to determine whether the method would be of use in groundwater resource assessments. The MCM simultaneously solves water and solute mass balances to determine unknown inflows to a system, in this application the groundwater component of baseflow. The incorporation of water quality data into the estimation of the surface water â groundwater interaction increases the use of available data, and thus has the ability to decrease the uncertainty of the estimation process. The balance equations are equated to an error term which is used in the quadratic programming solution of minimizing the square error sums in order to determine the unknown inflows. The mixing cell model is applied to datasets from the surface water â groundwater interaction test site developed by the University of the Free State, in addition to data collected along the middle Modder River during a fieldwork survey. The MCM is subsequently applied to a set of quaternary catchments in the Limpopo Province for which there are available calibrated estimates of the groundwater component of baseflow for the Sami and Hughes models. The MCM is further applied to the quaternary catchment D73F, located in the semi-arid Northern Cape, to assess the applicability of the mathematically based MCM in terms of a flow system located within a regionally-defined zero groundwater baseflow zone. The MCM results for each study area are assessed in comparison to groundwater baseflow volumes determined by the Pitman, Sami and Hughes models. A chemical hydrograph separation method which also incorporates water quality data is additionally reported for the study areas to further validate the MCM. The results indicate that the mixing cell model can reliably estimate the groundwater component of baseflow to a river. This application of the mixing cell model could contribute to increase and evaluate the accuracy of current groundwater baseflow estimates in South Africa, which will in turn ensure the responsible and sustainable use of the countries water resources.

This dissertation describes the results of field based investigations on groundwater flow at Ruashi Mine located in Katanga Province of Democratic Republic of Congo (DRC). The core objectives of the study were to simulate groundwater flow, estimate flow into the pits and ultimately design a dewatering strategy for the mine. In order to understand how groundwater flows into and through the mine, a detailed conceptual hydrogeological model was constructed as framework for numerical groundwater flow modelling. The numerical model was used simulate groundwater flow and predict pit inflow volumes. At the time of this research, mining at Ruashi was being carried out in three pits that are expected to reach terminal depth of 180 metres below ground level (mbgl) in 16 years of continued mining. The mine is located along a faulted overturned syncline composed of composed of Siltstones, Argillites, Sandstones and Shales and covered by Laterite. Based on aquifer hydraulic testing results, the transmissivity of the shallow aquifer was estimated to be 10 m2/d. The specific yield for the deep aquifer was estimated to be 1 x 10-5. The Chloride Mass Balance Method was used to estimate recharge to the groundwater system as 280 mm per annum (14% of Mean Annual Precipitation). Water levels vary from 1.02 to 62.5 mbgl. The general groundwater type was analysed to be calcium-magnesium-bicarbonate (Ca-Mg- HCO3), typical of young groundwater. The numerical groundwater flow model area is 15.7 km2 and comprises 5 layers, 17 240 elements and 10 614 nodes. The model results indicated that groundwater flow to the pits is unlikely to exceed 42 000 m3/d. Using the pumping capacity (15 000 m3/d) for year 2012, a maximum water level drawdown of 55 m was estimated. However, the numerical model demonstrated that the existing pumping boreholes can be augmented by an additional set of 16 boreholes pumping 2 000 m3/d per borehole. This pumping rate can lower the groundwater level to about 1 188 mamsl which is about two meters below pit terminal elevation. This study made significant contribution to understanding the hydrogeological properties of aquifers at the mine. The aquifer hydraulic testing data was used to estimate aquifer hydraulic parameters. However based on the field evidence, it is suggested that Packer testing could improve the estimates of aquifer hydraulic parameters for each aquifer. The numerical model demonstrated the typical aquifer response to different pumping scenarios. The different pumping scenarios were run in order to determine the optimum pumping rates to dewater the mine.

The vital role of water within the mining industry, both as an asset which generates value as well as a shared natural resource requiring responsible stewardship, have long been recognised. Due to extreme climate changes, an increasing population density and poor water management, securing of water has become a global challenge and water scarcity will continue to be one of the greatest challenges facing mine water management. There is no simple recipe for mine water management and regulations by environmental authorities, along with past polluting practices, are forcing mining operations to improve and prioritise their water consumption. Accordingly, this sector is expected to be increasingly required to demonstrate leadership through innovative water use management. A mine water balance is considered to be one of the most important and fundamental tools available for mining operations as management begins with a basic understanding of where water is sourced from, and where it is utilised. With this kept in mind, a mine water balance should be based on a holistic systems model approach with an appropriate relationship between the required level of complexity in the model structure and purpose. Excessive detail can cause the model to become clumsy and tend not to focus on strategic water management principles. Emphasis should be put on a system approach, taking into consideration the main interactions, feedbacks and functional relationships between the various parts of the whole system. An overall mine water balance that superimposes different water systems can be divided into a process water system and a natural water system. The natural water system is associated with the intrinsic hydrological cycle and is often disregarded due to uncertainties. It can however significantly impact on mine water usage and losses as indicated in this case study. Consequently, decision making and management options should be based on the evaluation of the system as a whole and inclusion of the natural system as a component of the mine water balance is imperative for accurate quantification. The natural system includes a surface water environmental circuit as well as a groundwater environmental circuit. Surface and groundwater resources have historically been managed separately, but more than ever before, interaction between these two systems are required to facilitate effective resource management. Mining activities have a major effect on the modification of the hydrological regime and the influence of increased hydraulic conductivity along with mining induced recharge, should be evaluated as part of the adapted mine water balance. Furthermore, mine dewatering predictions and climatic scenarios must be incorporated to reflect site conditions more accurately. As poor water management poses an operational risk to mining operations, this sector has developed novel ways to respond to water issues in differing circumstances and has illustrated the ability to turn risk into opportunity. Now, more than ever, special measures are needed to identify options for life-of-mine strategies and initiatives for water conservation and management. Future focus should be to continue investigation and implementation of the water use strategies in order to improve performance across operations and encourage engagement with other water users. Moreover, to share experiences, learn from others and contribute to water discussions and debate at local, national and international levels.

The study describes the effect of calculating a generalised mean transmissivity or hydraulic conductivity value for a region or aquifer system as it pertains to South Africa. Resource determination of an area is usually driven by the determination of the bulk flow parameters, such as hydraulic conductivity and storativity values. At this stage a decision is usually made on the basis of either maintaining the area under natural conditions (no pumping), or an abstraction (pumping) scenario is envisaged. In both instances water levels, hydraulic testing and distribution of the water resources (aquifer) are required. Since it is not possible to evaluate the total area for these parameters certain assumptions have to be made such as that an average bulk flow parameter for an area can be determined. In wide-ranging situations a simple average of observation points is assumed to be sufficient. A systematic research approach was followed in which a three-step process was used to evaluate methods of calculating these mean values. In the first instance a conceptual model approach was used, and all bulk flow parameters were generated by means of matrices to represent the natural system. Three typically employed mean values (arithmetic, geometric and harmonic) were calculated for two different dimensional matrices, i.e., N x N (N = 100 and 1000) with different hydraulic conductivity zones. In addition the relative difference between these hydraulic conductivity zones were steadily increased to mimic observed parameters in the field, i.e. typical hydraulic conductivity of shale (K = 0.01 m/d) versus a fracture zone (K = 100 m/d). In all instances the harmonic mean performed the best and as the number of sample sets were increased, a reduction in mean values were observed. As part of the conceptual model approach, two typically encountered scenarios were investigated, i.e. natural flow and forced gradient conditions. Under these two scenario conditions the harmonic mean performed the best to estimate the actual observed hydraulic conductivity value. Secondly, case studies were presented which highlighted the influence of sample size on observed parameters. Additionally, the effect of the differences between the low and high hydraulic conductivity zones on the calculated mean value as a function of sample size, was also reported. In all of these case studies the harmonic mean was the closest in approximating the observed hydraulic conductivity. It is evident from this section that the number of host rock (formation) hydraulic conductivity values plays a critical part in the mean value calculation since it is general practice in South Africa not to report low yielding borehole hydraulic test values. In the third step, the results were discussed in the context of a more general approach to the problem of calculating a regional mean hydraulic conductivity of transmissivity value. The estimation of representa-tive transmissivity values were discussed as seen from a stochastic modelling perspective as well as from the deterministic point of view. A comparison between main stream groundwater and oil industry specialists were noted in which both groups share the fundamental training but differ on the methodology of determining the observed transmissivity values. The impact of horizontal heterogeneities and different fracture networks was discussed and the influence these features have on the actual transmissivity value obtained, i.e. the influence of internal boundaries on hydraulic test data. Scale effects were also addressed from a regional perspective, with a focus on apparent scaling and the actual regional transmissivity value which should be obtained. The findings of this study are that in essence using geostatistical methods are not advised if regional transmissivity values are required from a South African perspective. The reason behind this statement is that the distribution of transmissivity values in an area does not follow the basic precepts that are required for these methods to work. In general the values are discontinuous in distribution and statistically skewed. Furthermore, the presence of transmissivity areas or points that differ significantly in magnitude, i.e. transmissivity values which differ by more than two orders, can be located within one meter from each other. The explanation of this phenomenon is the presence of dolerite dykes, which create baked-fractured zones with exceptionally large transmissivity values compared to the extremely low transmissivity ranges of the surrounding country rock (shales, mudstone and siltstone). In addition, the lack of data concerning low-yielding or âdryâ boreholes is a major source of concern since it influences the calculated mean value to a high degree.

Acid Rock Drainage (ARD) is expected to take place as soon as rock or coal that consists of sulphide minerals is exposed to the surface and comes into contact with oxygen and water. Water not only acts as catalyst, but also as a transportation medium of the yellow boy precipitates which in turn deteriorates the condition of the environment as well as water resources. Due to the long-term irreversible impact that ARD has, minimisation of this condition is an important factor to be looked at by mining companies. Worst cases have been reported by the Department of Water Affairs in and around coal and gold mines in the Witwatersrand and Witbank areas. This has brought about a necessity to explore pre-mining investigations into different lithologies to be aware of possible changes that might occur in the environment once mining commences. An evaluation of ARD was also done on the coal and rock samples to have better knowledge of what will be released in the environment and water resources as more mines are introduced in the Waterberg Coalfield.. Controlling ARD means to control the acid generating reactions of which mineralogy stands at the core of the whole reaction process. Coal deposits in South Africa largely consist of shales; mudstones, siltstones and sandstones which contain clay minerals, quartz, carbonates and sulphides. Problems that are associated with ARD (results) include decreased pH values and increased values of metals, acidity, sulphate and dissolved and suspended solids. The sulphate concentration is caused by sulphide minerals that are in the mining environment and undergo oxidation, thus bringing out the presence of sulphate. The investigation took place in four areas, namely Resgen, Grootegeluk (Exxaro), Sasol and Sekoko. A total of 214 samples were selected from the borehole core to conduct an Acid Base Accounting (ABA) (from the interburden, overburden and coal samples), 29 samples were selected for whole rock and mineralogical analysis, and 8 selected for kinetic tests. Results from the ABA and the kinetic tests leachate were further analysed for major and trace elements using the ICP-OES (Inductively Coupled Plasma- Optical Emission Spectrometer). ABA results show that the interburden and coal samples have a higher risk of producing acid upon oxidation when comparing it to overburden samples. They have a higher concentration of neutralising minerals that can neutralise the acid produced through sulphide minerals as oxidation takes place. In the Sasol samples, which were collected from the full succession that includes the Volksrust and the Vryheid formation, the closed Net Neutralising Potential (NNP) varied from -306.19 to 121.05 kg/tonne CaCO3. The results indicate that there is higher potential of acid production than neutralisation. Grootegeluk samples taken from the overburden showed a closed NNP that varied between -0.33 to 310.80 kg/tonne CaCO3 which classifies the samples as having more potential of neutralisation than acidification; the same case applies for Resgen with closed NNP that is between -68.96 and 57.74 kg/tonne CaCO3. Sekoko consists of the lowest number of samples that have a low risk of acid generation with more samples having a sulphide âS that is more than 0.3%. Mineralogical analysis indicates that there is a presence of pyrite, calcite and dolomite present in accessory to minor concentration. Minerals quartz and kaolinite are found in all the samples constituted in major to dominant concentrations. Results from the whole rock analysis which correlate with mineralogical analysis were SiO2 and Al2O3 which have the highest weight percentage and is similar to the presence of quartz and kaolinite. The average Fe2O3 that is 6.73 wt.% for Sasol and 3.28 wt.% for Resgen, is higher than the average CaO and MgO which is the result of the calcite and dolomite mineral. Leachate analysis from the ABA shows that metals become more soluble as acidity increases. An increase was noted in metals such as iron, magnesium and cadmium, which is a greater threat than the acidity that results from the drainage. Further tests were carried out after the completion of the ABA analysis; kinetic tests were performed on samples that gave inconclusive NNP results. The analysis showed the long-term behaviour of different samples with the EC and pH changing over time. Samples with a lower pH continue to produce more sulphate, while calcium continues to increase until it is depleted from the samples. The Waterberg Coalfield has not yet experienced major environmental deteriorations due to ARD; this is likely to change as more mines will be added to the area.

The engineering properties of a South African class F fly ash were studied as a potential base liner for a dry coal ash dump. In order to increase the unconfined compression strength, lime and gypsum were added to the fly ash while also aiding in reducing the hydraulic conductivity. Lime was added in the range of 1 to 10% while the gypsum amounts were varied at 1% and 3% per specimen. The constant head method was used to determine the hydraulic conductivity of compacted specimens in the laboratory. Gypsum was observed to have more influence in reducing the hydraulic conductivity as specimens with 3% gypsum had a more reduced hydraulic conductivity that those with 1%. The variations in lime percentages did not appear to reduce the hydraulic conductivity but rather displayed higher values than fly ash specimens without additives when higher percentages of lime were used. A fly ash admixture of 3% lime and 3% gypsum was found to have the lowest hydraulic conductivity of 2.27 x10-9 m/s after 60 days of percolating with brine water. The unconfined compression strength also appeared to be more influenced by gypsum than lime percentages as specimens with 3% gypsum obtained higher strength values than those with 1% gypsum added. Unreacted lime was observed in specimens with higher percentages of lime added and these specimens also presented lower strength values. The addition of lime and gypsum was observed to have limited the release of some trace elements from fly ash. The secondary mineral ettringite was detected and could have possibly precipitated and captured out these toxic elements. An attempt was also made to increase the plasticity index of fly ash using lignosulphonate and values recommended by the South African legislative guidelines for liner materials were obtained. The plasticity was however not retained with subsequent leaching. Two multi-layer liner systems were loaded under different compaction rates in permeameter cells with fitted inflow and outflow points. The primary liners of both systems were able to contain over 95% of leachate that percolated through a waste layer. Compaction rate was found to affect the liners performance as primary liners with a higher compaction rate had less seepage than primary liner compacted at a lower rate. An addition of lime and gypsum improved the overall engineering properties of fly ash to levels accepted by the South African legislative guidelines for a liner material that is able to line hazardous waste. Even though concentrations of some trace elements in fly ash were reduced by addition of lime and gypsum the level of some of these trace elements remain above the threshold set by South African legislative guidelines and therefore remains a health and environmental concern.