X-ray microcomputed tomography (µCT) as a potential tool in particle-based geometallurgy

Sammanfattning: In recent years, automated mineralogy has become an essential tool in geometallurgy. Automated mineralogical tools allows the acquisition of mineralogical, textural, and liberation information of ore samples. Such information is essential in the context of geometallurgy where it is needed for estimating the process response of the ores. Most automated mineralogical tools currently in application are based on two-dimensional (2D) microscopy analysis, which are subject to stereological error when analyzing three-dimensional (3D) object such as ore particles. Recent advancements in X-ray microcomputed tomography (μCT) have indicated great potential of such system to be the next automated mineralogical tool. μCT’s main advantage lies in its ability in monitoring 3D internal structure of the ore at resolutions down to few microns, eliminating stereological error obtained from 2D analysis. Aided with the continuous developments of computing capability of 3D data, it is only the question of time that μCT system becomes an interesting alternative automated mineralogical tool for ore characterization.This study systematically evaluates the applicability of μCT as an alternative tool for ore characterization in the context of geometallurgy. The focus has been to assess the potential strengths of 3D data generated from μCT as well as how such data can offer a new perspective in characterizing the ore. Some of the limitations of 3D μCT data in describing the ore were also discussed, with alternative methods proposed to address these limitations. The main hypothesis of the study is that 3D data generated from μCT can be of a value in a geometallurgical program. This study has been conducted in three different parts in order to systematically test the hypothesis. The first part of the study evaluated the use of μCT to obtain mineralogical characteristics of the ore. Mineralogy of the ore is the cornerstone information needed to proceed with further characterization of the ore. It is therefore important to establish whether μCT are capable to obtain such information. The study demonstrated the well-known limitation of μCT, namely its difficulty when dealing with minerals of similar attenuation. The study demonstrated how machine-learning based methods complemented with 2D data from automated mineralogy could address the limitation.The capability of μCT for ore texture characterization was evaluated in the second part of the study. The main strength of μCT for core scanning is highlighted in the study, in which the possibility of using μCT for automated drill core recognition was demonstrated. Some of the popular texture analysis methods in 2D such as Local Binary Pattern (LBP) and Association Index Matrix (AIM) were extended to 3D in order to capture the textural pattern in drill core samples. Furthermore, a classification scheme based on these textural characteristics was devised for the automated recognition of the drill cores. An accuracy of 84-88% was achieved in the classification scheme, illustrating the potential of μCT for such task.The last part of the study concerns the use of μCT for mineral liberation modeling. Combining both mineralogical and textural information obtained from the previous parts of the study, a liberation model to forecast particle population from the 3D ore texture was created. The model was based on various breakage types such as preferential, phase boundary, and random breakage. The contribution of each breakage type to the final particle population could then be adjusted with actual particles produced from experimental comminution. Accurate forecasting of particle population is one of the key componentin the particle-based geometallurgy, in which the particle carries the ore characteristics to the beneficiation process. Utilizing these particles in a process modeling and simulation would give some idea about the process response of the ore. The integration of 3D data from μCT with the liberation model could potentially complete the link from ore characteristics to the process behavior in the framework of particle-based geometallurgy.Combination of these three parts of the study can open up a 3D path of particle-based geometallurgy. The study has demonstrated the efficient extraction of crucial ore characteristics such as texture, mineralogy, and mineral liberation using μCT. Key limitations and potential measures to address them have also been discussed in the study. Coupled with a framework for process simulation using such ore characteristics as an input, the 3D path of particle-based geometallurgy can be realized. Future research should be dedicated to develop such framework, as the establishment of μCT as an alternative ore characterization tool should also be motivated by from the downstream processes, i.e.whether the 3D μCT data can unlock a new perspective in process modeling and simulation compared to the conventional 2D data. This new perspective can help to build more accurate process prediction and production forecasting, which can ultimately guide the decision-making process for efficient resource management as the essential core of a geometallurgical program.