Underground mine water rebound prediction in its simplest form can be simulated linearly by comparing the volume of the mined ore with long-term average recharge rate to obtain an estimate of the time which will elapse before the workings are full to their decant elevation.
This type of linear interpolation of rising water levels can lead to an over estimation or an underestimation of the date when mine voids will flood to the critical levels. This is due to the fact that this method cannot account for the variability and interconnection between different mine voids and also does not consider the change in storage over time which is an important factor. In an abandoned underground water environment, water is stored in flooded mine stopes (tanks) and flows through a network of haulages (pipes). Due to the dip and strike of the ore body, the mined stopes are extensively interconnected on multiple levels and bounded by faults and dykes, so that water rising within any one tank will display a common level throughout that tank. At certain elevations, adjoining tanks may be connected via a discrete "overflow point", which may be a holding or permeable geological features. Water level rise during flooding is a function of head-dependent inflows from adjoining mine aquifers and/or other tanks, and the distribution of storage capacity within the tank.
The process of flooding occurs independently in two (or more) adjoining tanks until such time as the water level in one or more of the tanks reaches an overflow point. Inter-tank transfers of water will then occur until the difference in head between the two tanks either side of each overflow point is minimised. To apply the conceptual model stated above, EPANET 2 was used to predict the risk of flooding of a mine shaft, in the Free State Goldfields, if dewatering is discontinued. Considerations on stope volumetric calculations, haulage interconnections, modelling assumptions and predictions, are presented.