Development of new metaheuristic tools for long term production scheduling of open pit mines

Long term production scheduling of open pit mines is a large scale and complex optimization problem that has been extensively discussed in the technical literature since 1960s. It seeks to specify such an extraction sequence of ore and waste materials from the ground that maximizes the Net Present Value (NPV) of the operation while satisfying a set of physical and operational constraints. Block model representation of the orebody is commonly used as a basic input for this purpose. The block model discretize the ore deposit into a three dimensional array of regular sized blocks. A real sized open pit mine may contain thousands to millions of blocks of these blocks that may be needed to be scheduled over a time horizon typically ranging from 5 to 30 years which makes it a large combinatorial optimization problem. This thesis presents a framework that aims to handle the above mentioned computationally expensive problem of the open pit mines with low to moderate computational cost. To handle the scheduling problem more efficiently the proposed framework converts it into optimum depth determination problem. This so called optimum depth determination problem aims to find the optimum depth to be mined along a particular column of the block model in a certain period. In this way this framework helps to avoid computationally expensive scheduling decisions making process on the block level. The framework then uses a real valued / continuous population based metaheuristic technique to search the solution space for finding optimum or near to optimum solution of this so called optimum depth determination problem and consequently of the production scheduling problem. Different framework specific operators such as solution encoding, back transform, slope normalization etc. are also used during this process. The proposed framework can handle the production scheduling problem with or without the condition of grade uncertainty. Three different case studies have been carried out using Particle Swarm Optimization (PSO), Bat Algorithm (BA) and Differential Evolution (DE). The aim of these case studies was to determine the capabilities and efficiency of the proposed framework and of the respective metaheuristic technique along with of their different variants. By making comparison with the iv results obtained using CPLEX in terms of computational time and solution quality it was learnt that the proposed procedure can produce results of reasonable quality in relatively shorter period of time with smaller % gap and standard deviation. The following research papers have already been published or in the review process using some parts of this work: • Asif Khan and Christian Niemann-Delius, “Production Scheduling of Open Pit Mines Using Particle Swarm Optimization Algorithm” , Advances in Operations Research, vol. 2014, Article ID 208502, 2014. DOI:10.1155/2014/208502 • Asif Khan and Christian Niemann-Delius, “Long Term Production Scheduling of Open Pit Mines using Particle Swarm and Bat Algorithms under grade uncertainty”, The Southern African institute of Mining and Metallurgy (In review).