Abstract:
Stochastic mine optimisation requires the full characterisation of orebody uncertainty. This characterisation does not only include the simulation of grades and geometallurgical variables, but begins with the characterisation of geological uncertainty. The application of simulation techniques is often limited to the modelling of grade uncertainty. This results in an understatement of the total uncertainty that is transferred to mine optimisation and planning, with important repercussions for the evaluation of risks and opportunities.
Gaussian-based simulation techniques for categorical variables, such as Truncated and Hierarchical Truncated Plurigaussian simulations, are well established. These techniques, however, need to be adapted to account for the structural complexity of diverse deposit types.
This presentation shows case studies demonstrating how these techniques are adapted for modelling the lithological and structural controls of mineralisation in folded layered deposits, skarn deposits, and orogenic deposits. Global and local spatial transformations are presented as key to honouring the complex geometries of these controls.
Layered deposits are unfolded, and the thicknesses and elevations of the main layers are modelled in 2D using a compositional approach. The 2D simulated contact elevations and layer thicknesses are subsequently back-transformed to the folded 3D space in a manner that reproduces the stratigraphic sequence. A signed-distance approach is applied to the simulation of the main lithologies in a skarn deposit. The zero-value envelopes are used to model the uncertainty in the contacts between the different lithologies.
In complex orogenic deposits, where multiple structurally controlled mineralisation events are present, the original space can be locally transformed around the interpreted structural controls. Truncated Gaussian simulation methods can be applied to mineralised lenses within the zones of influence of individual structural controls. When secondary data, such as vein intensity from automated core logging, are available, they can be used in combination with grades for the modelling of the mineralised lenses. The simulated models are validated by local and global reproduction of categorical proportions and by reproducing the spatial continuity of the geological categories.
Presenters:
- David Machuca-Mory | SRK Consulting (Canada)
- Silvana Stipetich | SRK Consulting (Peru)
