Managing mine water under extreme climate variability using a model predictive control approach

In Australia, growing water demand, intensifying competition with other water users, and significant climate variability have brought a pressure of securing water use to the mining industry, especially in years of droughts. However, during heavy rainfall periods, unregulated discharges of mine-affected water may put the livelihood of community and the environment at risk. Therefore, a careful balance needs to be struck between securing mine water use during water limited periods, on one hand, and mitigating environmental risk expressed through unregulated discharges when water is in excess, on the other. This paper designs a model predictive controller (MPC) to address this mine water management problem. The MPC uses a simulator of the mine water system as the internal model to make future predictions, on which an optimal control action sequence is determined by solving an optimisation problem at each control step. The cost function of the optimisation is to minimise the deviation between the water level and reference level, and to minimise the energy consumption that is required to implement the deviation. After solving the optimisation, a sequence of future control actions is obtained and only the first action is actually applied into the mine water system. Rainfall predictions are incorporated into the MPC to make predictions of the future water levels in the reservoirs. We applied the MPC into a coal mine in the Fitzroy Basin, Queensland, Australia. The results demonstrate that the designed MPC is able to ensure a 'fit-for-purpose' water use security and reduce the risk of unregulated discharges, even though in the condition of the existence of uncertainties in rainfall forecasts.