This paper presents a near optimal hoist scheduling and control program for rock winders found in South African deep level mines in the context of demand side management and time-of-use (TOU) tariffs. The objective is to achieve a set hoist target at minimum energy cost within various system constraints. The development of a discrete dynamic and constrained mixed integer linear programming model for a twin rock winder system is presented on which a half-hourly model predictive control(MPC)algorithm containing an adapted branch and bound methodology is applied for near optimal scheduling. Simulation results illustrate the effectiveness of the control program by minimising the energy costs through scheduling according to the TOU tariff and controlling output and ore levels within their boundaries even in the case of significant random delays in the system. Scheduling according to the TOU tariff shows a possible 30.8% reduction in energy cost while approximately 6h of delays in the system resulted in a mere 14 increase in energy cost.