Simulated learning model for minable reserves evaluation in surface mining projects - SME Transactions 2013

The amount of information available for characterizing a deposit increases over time due to the continuous acquisition of data during mining. This additional information is collected from different sources, including geologic mapping, production data and infill drilling. Throughout the lifetime of a mining project, the block model and the mining sequence are periodically updated to account for this new data. In general, the acquisition of additional data increases the accuracy of the block model, reduces uncertainty in ore/waste limits and clarifies the optimal mining sequence. There has been extensive research on mine planning, but current techniques do not consider the decrease in uncertainty as additional information becomes available. Conventional paradigms assume either 1) the kriged model is correct and uncertainty due to multiple realizations does not change the mining sequence, or 2) the mining sequence is unrealistically adapted to each realization. A new paradigm is proposed for evaluating minable reserves of surface mining projects, which accounts for the acquisition of additional information in the design of the long term mine plan. Multiple scenarios characterizing the dynamic nature of mining and data collection are created. In the proposed methodology, the performance of the long-term mine plan depends on both mining and data acquisition strategies. Each scenario considers the same extracted volume for the first period, resulting in a set of scenarios that diverge from a common initial period, accounting for how the mine may develop over time as new data are acquired and provide a more realistic evaluation of reserves with an appropriate level of uncertainty. A synthetic example is presented to illustrate the implementation of the methodology and the benefits over conventional paradigms.