Critical materials & strategic value chains for the Energy transitionDirectorate for Energy, Transport and Climate of the Joint Research CentreUnit C7 - Knowledge for the Energy Union

Developing new data assimilation methods to update high-order geostatistical simulations. These can incorporate incoming information from new digital technologies with different geometallurgical properties of a mining complex in non-Gaussian and non-linear settings, reducing the uncertainty of the resource estimation.

Ph.D. Thesis: Multivariate Data Assimilation for Geometallurgical PropertiesThe aim of my thesis wasto formulate and implement innovative multivariate data assimilation techniques that account for multivariate geometallurgical data for updating the resource model in short-term mine planning. Most commonly used geometallurgical properties involve characteristics of the elements (as chemical assay concentrations), minerals (as modal mineralogy), texture, hardness, fragmentation, and grain-sizes, among others. These properties have a high influence in later metallurgical processes like crushing, flotation, grinding, etc. For that reason, a proper quantification of these properties is of importance for an optimal resource estimation and mine planning. Some of these properties are represented by vectors of positive components often mutually linked through complex multivariate relationships such as non-linearity and heteroscedasticities (compositional data). On the other hand, the micro-structure and texture of the ores information is of crucial importance because it allows a description of the geometric spatial relations between the mineral grains and not just summaries of their composition (i.e., particle-size or mineral grain-size). These are curves of distributional data: a full characterization of these would require infinitely many abscissa-ordinate pairs. By accounting for compositional and distributional settings, my thesis provides innovative, robust and tailored updating algorithms to continuously fuse the data into the resource models decreasing the uncertainty of the resource models on real-time.