Inter-comparison geochemical modelling approaches and implications for environmental risk assessments: A Witwatersrand gold tailings source term characterisation study

Abstract

Acid mine drainage (AMD) has long been recognised as a major environmental concern from mines associated with especially Fe-sulphide minerals with the ore and/or gangue. The generation of AMD and its associated geochemical risks, e.g. leaching of toxic metal(loid)s into the environment, is a major concern from such mines, as is the case in the Witwatersrand gold mines. Geochemical modelling, incorporating thermodynamic equilibrium and kinetic in modelling geochemical processes in the mineral waste, forms an important part of this process. However, a concern regarding the geochemical modelling is on the meaning of the results, especially when these models are used to make long-term projections, as complex scientific models cannot be validated. This study shows that inter-model comparison using existing data from specifically gold tailings pore water is a viable method to build confidence in geochemical models used for risk assessments. It also demonstrates for trace elements, specifically As, Co, Ni, Pb and Zn, that sources can be identified and the release rates estimated. The process of model confidence building is the most important in identifying problems with the conceptual model as well as determining which sources are important for specific trace elements. The Pb concentration in uraninite exceeds that of pyrite by 2 orders of magnitude; however, pyrite contains sufficient concentrations of Pb so that its’ more rapid dissolution rate renders it the more important source of Pb in the tailings pore water. The varying release rates for the trace metal(loid)s can also be used to estimate their mode of occurrence within source mineral structures. Quantification of the concentration of trace elements in specific host mineral phases should be a requirement of any environmental geochemical risk assessment and can be done using standard mineralogical analytical tools.