Geotechnical and grain size assessment using Corescan automated core logging and complementary microanalytical techniques

Careful assessment and modelling of geotechnical and geometallurgical characteristics within an ore deposit are crucial to determining both the viability and profitability of a mining operation. Typically, a geotechnical model is constructed by combining manually collected geotechnical measurements from drill core with other geological observations. While this approach is successfully applied in the mining industry, it is often laborious and has the potential to generate inconsistent results. A geometallurgical model is the synthesis of comminution, liberation, and recovery data. While mineralogical proxies for comminution are widely used, proxy models for predicting liberation and recovery are inadequate for broad-scale implementation. The liberation, and ultimately, recovery of economic minerals is directly linked to the grain size, but current grain size assessment methods are costly and slow. This research investigates the potential to use data generated by a Corescan hyperspectral drill core logger for geotechnical and geometallurgical applications. Two primary aims are the focus of this thesis: (1) develop protocols to rapidly extract consistent morphological and mineralogical geotechnical parameters from Corescan data, and (2) test a number of currently available microanalytical techniques that could complement the current Corescan system in rapidly determining proxies for copper sulphide and gold grain sizes. These complementary techniques must operate on the same interval as a typical assay (1 to 2 metres) and collect measurements at a rate similar to Corescan data acquisition (3 minutes per metre). Corescan digital drill core data and rock samples from the Cadia East underground mine provide a case study to develop, test, and evaluate the protocols generated to address these aims. To extract ore, the Cadia East underground mine currently uses the block cave mining method. The site geotechnical model provides the foundation for the construction of stress models, caveability models, ground support design, and fragmentation analysis. While not all industry standard geotechnical parameters can be derived from Corescan data, this research focuses on calculating fracture orientation, fracture roughness, number of fracture sets, fracture spacing, fracture condition, and fracture alteration. A test dataset of 199 fractures from one Cadia East drill hole was used to develop and test a workflow to extract geotechnical parameters from Corescan data. The orientations of the 199 fractures were measured manually prior to Corescan analysis. After Corescan analysis, the manual measurements were compared to the calculated measurements derived from the Corescan data. Over 75% of the calculated orientations were within 25¬¨‚àû of their measured orientation value. Calculated fracture roughness, number of fracture sets, fracture spacing, fracture condition, and fracture alteration results were compared to expected site values and drill core photographs. Overall, the fracture protocols performed well and produced over 90% roughness values within the expected range for Cadia East, slightly underestimated the number of fracture sets and fracture spacing, and produced fracture condition and alteration values consistent with the fracture condition observed in photographs. One advantage of the automated protocols developed in this study is the ability to collect a higher density of data than is feasible by manual methods. This increased data density is collected rapidly and consistently, providing additional advantages over manual geotechnical logging methods. The second aim of this research is to search for a proxy dataset that would predict grain size at the critical size for recovery (less than 100 microns). The Cadia East geometallurgical model classifies ore types as they relate to mineral processing flowsheets and mill design. One of the key parameters used in the site geometallurgical model is the grain size of chalcopyrite, bornite, and gold. Accounting for the effects of grains size on liberation and recovery can optimise processing design. If rapid and broad-scale assessment of grain size can be utilised, robust proxies for liberation and recovery can be incorporated into the geometallurgical model to improve mine planning and design. A scoping study was designed to test multiple commercially available microanalytical techniques and determine which methods could provide reasonable proxies for copper sulphide and gold grain sizes. The objective of this study was to identify one or more microanalytical techniques that complement the current Corescan system, since the current sensors cannot directly detect sulphides less than 100 microns. Copper sulphide and gold grain sizes are commonly determined by the mineral liberation analyser (MLA) method. Microanalytical systems were tested using a set of twenty-six, 3 cm by 3 cm rock tile samples from Cadia East. Samples were analysed by portable x-ray fluorescence (pXRF), micro-x-ray fluorescence (˜í¬¿XRF), laser ablation inductively coupled mass spectrometry (LA-ICP-MS), laser induced breakdown spectroscopy (LIBS), and laser Raman. To determine the functionality of each technique, a grain size proxy derived from the microanalytical data was compared to the MLA dataset. Of the five microanalytical techniques tested, ˜í¬¿XRF, LA-ICP-MS, and LIBS were capable of producing reasonable grain size proxies for copper sulphide minerals. Determining grain size proxies for gold proved challenging due to complexities with sampling statistics, but, in general, when a sufficient number of gold grains were analysed, pXRF and ˜í¬¿XRF produced reasonable gold grain size proxies. LA-ICP-MS line scan analysis was unable to overcome the challenges of sampling statistics for gold, but this is the only technique with high enough precision to detect sub-micron gold grains (which could have broader geometallurgical applications). The laser Raman system was unable to detect gold and copper minerals under relevant conditions and further testing of this technique was terminated. Currently available ˜í¬¿XRF and LIBS technologies are able to collect a sufficient number of analyses to produce robust copper sulphide grains size proxies at a rate comparable to the Corescan system throughput. Recent developments in core scanning XRF systems indicate that this analysis will have data acquisition rates sufficient for gold grain size proxy calculations to become feasible. At the moment, LA-ICP-MS technology cannot be completed outside of an ablation cell, but, it could be used for the rapid scanning of selected core samples outside of the Corescan sample analysis stream. Of all the methods tested, ˜í¬¿XRF and LIBS show the most promising potential for future development and integration into the current Corescan system for geometallurgical grain size assessment. The use of current data outputs obtained from the Corescan automated core logging system allow for key properties that affect the geotechnical response of a rock mass to be rapidly and consistently estimated. Integration of the existing laser height and hyperspectral derived mineralogical data with a complementary system (such as ˜í¬¿XRF or LIBS, to quantify grain size) provides an opportunity to generate large volumes of consistent data. Through the methods developed in this thesis for geotechnical and geometallurgical grain size assessment, the underlying statistical support for rock mass characterisation and liberation and recovery modelling can be greatly increased. These outcomes have the capacity to substantially contribute to better geotechnical and geometallurgical models that will improve mine planning and ore recovery, which in turn will improve mine safety and profitability.