Stope Sequence Optimization for Narrow Vein, Long Term Mine Planning

Long term mine scheduling is one of the most difficult optimization problems. The problem is said to be NP-complete in that the best possible solution cannot be found any quicker than checking all possible permutations and combinations. In current mining operations, it is important to report changes in long term plans as soon and as accurately as possible. Where geological data can change rapidly, such as in narrow vein deposits, the need to quickly identify and optimize different long term alternatives becomes a difficult task and can lead to many permutations of the long term mine plan. The question arises, is the new plan optimal or not? Can the company objectives be met while adhering to operational constraints? Although a number of optimization techniques have been used in the past, many include significant simplifications or fail to produce acceptable results within the required timeframe. Genetic Algorithms (GAs) are among the most promising optimization techniques within operations research. They are based on natural selection and evolution of solutions to complex problems. They have been known to quickly converge within a few percent of the optimal solution by examining only a fraction of the possible permutations. The GA itself contains very little problem specific information and, as such, in this study, has been linked to a discrete-event simulation (DES). The DES is able to evaluate each sequence proposed by the GA and determine the effectiveness and value of each schedule. The DES feeds information regarding the feasibility and quality of the solutions back to the GA during the optimization process. In this paper, GAs and DES are described and the coupling of the two methods is outlined. The GADES module is then used to ?optimize? a narrow vein theoretical case study and the results discussed in a broader mining context.