Mineral resource deposit models currently rely on manual processes, from drilling and handling core samples to chemical analyses and lithological identification. These labor-intensive procedures extend to geostatistical analysis, where experts describe the spatial distribution of geo-domains and mineral properties through variogram analysis and numerous simulations to address uncertainty. In complex geological environments, numerical methods often fail to effectively characterize the spatial distribution of boundaries between geological domains. Consequently, geological experts manually delineate the morphology and boundaries of ore types. Estimations rely on sampling data collected during exploration and production, representing only a small fraction of the entire deposit and accompanied by significant uncertainty.

These steps require experienced geologists, are time-consuming, susceptible to variability, and limit the speed at which mineral resources are updated. This can result in significant delays between new sampling data availability, grade estimation, and decision-making in mining operations, often spanning several months and leading to outdated information and substantial bias in mined reserves.

This project proposes developing a digital twin (DT) representing a mineral resource, enabling effortless and dynamic integration of new information and updates. The approach adopts a 3D framework using Probabilistic Neural Networks for modeling, enabling simultaneous representation of geological boundaries and metal distribution while quantifying and incorporating their uncertainties. Additionally, the metal distribution estimation framework integrates XRF samples, enhancing the speed and accuracy of mineral resource modeling updates.

The primary objective is to achieve fully automated mineral resource estimation capable of integrating into a real-time framework. This integration enables fast and robust updates, providing timely and reliable support for decision-making, optimizing mine planning, and addressing challenges in real-time, ultimately enhancing the efficiency and sustainability of mining projects.