Extracting Value from Simulation Models: Feasibility Grade Control

This paper discusses the use of Conditional Simulation (CS) models to provide mill feed grade profiles for different time periods. The conventional approach of classifying block estimates as ore or waste generally does not adequately represent dilution, lost ore and the anticipated grade control data, including variability of mill feed from the pit and possible stockpiles. The detailed simulation of grade control practices can be done, but at the Pre- or Feasibility stage of the project may be impractical, since drawing dig limits on multiple benches is tedious and time consuming. Using the Feasibility Grade Control (FGC) method as developed by the Centre for Computational Geostatistics (CCG, University of Alberta-Edmonton), the CS models can be processed quickly to establish dilution and lost ore, and predict daily or short-term feed to the mill. The simulated nodes can be aggregated into selective mining units with adequate representation of likely geometries to be recovered from the pit, to mimic anticipated selectivity. An example from a large copper deposit is shown. The study provides an estimated stream of tonnages and grades fed to the plant by mining periods of interest and based on a specific mine design and schedule. The resulting cash flows support the economic and financial analyses of the Feasibility Study; also leads to improving mine schedules; and aids in developing blending schemes if necessary. INTRODUCTION The conventional approach of separating mineralized material into ore or waste categories does not always include a realistic amount of dilution and ore loss. Also, often it does not provide a good estimate of the grade that will be fed to the mill. If the grade control estimates are unbiased, they will over the long run represent the long-term average of grades fed to the mill; but the anticipated variability of mill feed from the pit and stockpile sources may be misrepresented. Some references on traditional grade control implementations can be found in Chunnett (1982); Davis et al. (1989); Srivastava et al. (1992); and Nowak (1992). Alternative methods based on economic variables improved significantly on the purely grade-based control methods, see Douglas et al. (1994). Further developments involved the use of geostatistical conditional simulations (CS) and profit models, which in several cases have been successfully implemented (Aguilar and Rossi, 1996; Rossi, 1997; Deutsch et al., 2000; Norrena and Deutsch, 2002; Neufeld, 2005; Jewbali and Dimitrakopoulos, 2009). But they can be impractical, mostly due to the lack of understanding of the methodology involved; even if semi-automatic dig limit drawing is implemented, there is always a need for is tedious and time consuming checking and corrections, even if using a single profit model derived from the simulated grades.