Production planning decisions in the mining industry are affected by geological, geometallurgical, economic and operational information. However, the traditional approach to address this problem often relies on simplified models that ignore the variability and uncertainty of these parameters. In this paper, two main sources of uncertainty are combined to obtain multiple simulated block models in an iron ore deposit that include the rock type and seven quantitative variables (grades of Fe, SiO2, S, P and K, magnetic ratio and specific gravity). To assess the effect of integrating these two sources of uncertainty in mine planning decision, stochastic and deterministic production scheduling models are applied based on the simulated block models. The results show the capacity of the stochastic mine planning model to identify and minimize risks, obtaining valuable information in ore content or quality at early stages of the project, and improving decision-making with respect to the deterministic production scheduling. Numerically speaking, the stochastic mine planning model improves 6% expected cumulative discounted cash flow and generates 16% more iron ore than deterministic model.
Geostatistics offers a set of methods for modeling, predicting, or simulating geological domains in space. In addition of being an input of some of these methods, indicator direct and cross-variograms convey valuable information on the geometry of the domain layouts and on their contact relationships, in particular, on the surface area of a domain boundary, on the surface area of the contact between two domains, on the propensity for a domain to be in contact with, or separated from, another domain, and on the minimum and maximum distances between points from two domains. Accordingly, the indicator variograms inferred from sparse sampling data can be used to determine whether or not an interpreted model of the subsurface is consistent with the sampling information. The previous concepts are illustrated through a case study corresponding to a porphyry copper deposit.
Beach litter is a worldwide problem that has several negative effects. A first step in preventing an environmental hazard is to determine and model the level of contamination. In this paper, geostatistical simulation is used to model two main forms of beach litter (cigarette butts and sharp items) in one of the most contaminated beaches in Antofagasta, Chile. A hundred realizations of cigarette butts and broken glass are generated to emulate their joint spatial distribution. The simulation results are used to classify the beach into different areas with respect to the risk of injury by broken glass and the level of contamination by cigarette butts. The models obtained can be used by local authorities in beach clean-up programs and by visitors to beaches in choosing the safest and cleanest areas. The results demonstrate the capability of geostatistical simulation algorithms to model different types of beach litter.