Reliability based multi-objective strategic and operational mine planning through parallel meta-heuristics

点击翻译按钮获取中文摘要

Mining is a cyclical business and experiences with periods of low prices. In Canada, mines also go deeper, which increases operation costs. These characteristics jeopardize the future of some mining operations. The suspension or termination of operations leads to increasing unemployment rate, loss of the investment and negative impact on the economies of local communities. Therefore, to stay in business, mining operations should be improved in such a way as to challenge the negative atmosphere. The advancements in the fields of computer science and operation research have an important potential to increase productivity, efficiency and profitability of mining operations. At the same time, mine planning software industry is also competitive and seeks to add new features to their variety of products to improve their product portfolio. In this scope, the proposed research aims to develop new optimization frameworks to increase the utility obtained from underground mining operations. The research will move forward from sequential to simultaneous optimization such that cut-off grade, stope layout and sequencing problems are concurrently solved to generate strategic and operational plans. The size of problem will inevitably grow. Therefore, the research will investigate various clustering, parallel and distributed computing techniques to deal with size issue. To solve the problem, a hybrid meta-heuristics using various properties of simulated annealing, genetic algorithms and tabu search will be developed. In addition to maximization of net present value, due to tactical or mine specific reasons, there could be different objectives such as recovery maximization, dilution minimization or capacity utility maximization. The problem formulation will be extended to consider multi-objective characteristics and non-dominated solutions will be generated. Hence, the decision-maker can pick up a solution suiting her/his preferences. Grade data used in underground mine planning are sparse and financial inputs depend on a series of unknown future events. Consequently, underground mine planning comprises significant degree of uncertainty. Therefore, the research will also address reliability-based optimization and risk quantification.