Optimization tools and simulation methods for designing and evaluating a mining operation

Abstract

Some of the major advances in the field of mining in the last three decades have referred to the development of new design and planning techniques for optimizing open-pit mining and the inclusion of a stochastic perspective in economic models that is more revealing than a purely deterministic perspective. These advances include the use of parametric techniques in the design and planning process, the formulation of criteria for establishing an optimum cut-off grade policy when the economic goal is to optimize net present value (NPV), and the introduction of economic risk analysis. This paper examines some of the difficulties involved in applying these techniques—arising largely as a result of a lack of knowledge of the spatial location and distribution of the deposit grades—and analyses how these difficulties can be tackled with the help of geostatistical simulation techniques that take probabilistic criteria into consideration during the optimization process. These techniques enable equally likely representations of the deposit to be obtained that reproduce the main dispersion features for the starting experimental data (covariance or variogram, as well as the histogram). Consequently, the uncertainty in regard to the deposit as well as its influence on the economic assessment of the deposit in risk terms can be evaluated. This paper also describes a simple method for introducing price and cost increases into the risk analysis via the Monte Carlo method and shows how geological, technical and economic uncertainty can be integrated in risk analyses. Although it is true that the relationship between prices and costs is maintained constant in mining planning based on using parametric techniques, it is no less true that the risk analysis requires the use of models in which the main parameters with a bearing on deposit economics are considered as stochastic variables. The proposed methodology simplifies the calculations and easily integrates the different sources of uncertainty.