Research

The focus of our research is on predictive geometallurgical and geostatistical modeling.

Geometallurgy combines geological, mining and metallurgical information to create spatially-based predictive models for mining, mineral processing and metallurgy that can be used to optimize the decisions, given all other key project constraints such as environmental restrictions, water availability and energy efficiency.

Our goal is to create a system-based model to analyze problems related with the mining life cycle. Predicting process performance in mining, mineral processing and metallurgy, requires understanding the inputs, the processes and the outputs in each one of the stages of the system. Uncertainty plays a significant role in optimizing and improving processes. Therefore, the research can be divided in different components:

Geological characterization and modeling

This includes image analysis, integration of different sources of information, definition of spatial domains, spatial modeling with geostatistical tools, scaling on non-additive variables, among other issues.

Predictive Geometallurgy – Integration of in situ model and mining and metallurgical processes

Ore bodies are variable in nature, and we can only access a few locations, through sampling, to characterize their properties. This translates into uncertainty in these properties. Unfortunately, decisions about the mining method, plan, schedule, processing equipment and setting, and the mine operation are based on this very limited information. Important economic decisions must be made with fairly high levels of risk and these continue throughout the life of the assets.

Spatial prediction of the properties can be achieved by using geostatistical tools. However, having a model of the in situ resources does not guarantee a good prediction of the production performance. Several issues arise when the goal is to predict the performance of a specific process. Each process from blasting and loading to beneficiation is a complex combination of physical and/or chemical transformations, not fully understood. Predictive statistical methods based on machine learning and deep learning may replace the phenomenological understanding of these complex processes. Sample quality and spatial variability inject uncertainty in the final model. This uncertainty should be taken in consideration when making decisions, rather than basing these decisions on expected average values. This is critical when extreme values have large consequences in process performance.

Predicting performance requires an integrated view of the different stages of the mining process. The term geometallurgy has been coined initially to describe the integrated understanding of the materials geological characteristics and the mineral processing or metallurgical beneficiation performance of these ores. Currently, this definition can be expanded by saying that geometallurgy (geo-mining-metallurgy) combines geological, mining and metallurgical information to create spatially-based predictive models for mining, mineral processing and metallurgy that can be used to control the processes and optimize the decisions, given all other key project constraints such as environmental restrictions, water availability and energy efficiency. The materials characterization (ore and waste materials) must be modelled in space in terms of the in situ resources, but these characteristics must also be carried downstream through the scheduling, blending, and processing. Therefore, the question does not end at the ore body. The research goal is to propose new methodologies for predictive modelling, by using machine and deep learning techniques adapted to the spatial context and constrained by the geological knowledge of ore deposits.

The rapid development of artificial intelligence presents an opportunity for the minerals industry, as it may significantly improve the performance of mining projects by reducing resources consumption (water, energy, equipment and labor) thereby improving mining sustainability. Machine learning is already incorporated into Earth Sciences problems, but is not fully integrated into the geometallurgical workflows that model and optimize the performance of physical and chemical processes involved in the extraction (mining) and recovery (mineral processing and metallurgy) of minerals and metals. Deep learning is rapidly developing, but has not been explored in the mining field. In times of increasing water and energy scarcity, this research provides a unique and novel approach to significantly impact the decision-making processes in mining by offering systems workflows to predict and optimize the mining, mineral processing and metallurgical processes.

Modelling energy and water utilization in mining

Understanding the effect of rock properties in energy and water use is extremely important to improve the sustainability of mining projects. Energy is mainly consumed in blasting, crushing and grinding. The relation between in situ properties of the rock and the consumption must be understood to optimize the use of alternative energy soucres, such as photovoltaic energy. Sizing the photovoltaic plant and the battery energy storage system associated to it, can be done if the hardness of the rock and the variability in solar irradiance are integrated.

Similarly, water is lost in different stages of the process of the ore, but most of it goes to the tailings and can be recovered, or it evaporates or is lost by infiltration. Modeling these processes depend on properties of the rock such as mineralogy, presence of clays, and granulometry.