Abstract: |
Mining industry increasingly uses simulations to predict fluctuations in ore grades. However, in some cases grade fluctuations analysis is not enough for risk assessment, as in some cases ore recovery at the processing plant yield depends on the rock type rather than the grades. Lateritic nickel weathering profiles are highly complex in its composition. The dataset used in this study is divided into five geological domains, which present in many locations changes in the sequence or lenses of different material trapped inside a major geological domains. It is common the absence of certain domains at some borehole intercepts. The use of a deterministic or interpreted model for defining the orebody contacts in this dataset is unsatisfactory. Instead to use deterministic models to define this orebody, based only in the interpretation of the drill holes (which does not take into account the uncertainty of lithological domains in areas with no information) this paper presents plurigaussian simulation as an alternative to generate equally probable scenarios from the orebody. Combining these realizations it is possible to map the uncertainty in the volumes and contacts. The simulation results showed good reproduction of the conditioning data, both laterally and vertically. Thus, the result of each simulation is also a possible geological model, presenting the spatial variability of the phenomenon, without the smoothing shown by other techniques. One advantage in using plurigaussian simulation for lateritic nickel profiles is the reduction of working time in a very complex geological model, using the average of the simulations (E-Type) to mimic the reality and producing a set of different equally probable scenarios allowing to access the uncertainty in defining the geological domains. |
