Development And Testing Of Real-Time Shovel-Based Mineral Sensing Systems For The Enhanced Recovery Of Mined Material

In modelling, geological data is treated to define a contiguous orebody, which is then broken down into blocks based on grade. Blocks above a particular cutoff are scheduled for extraction to the mineral process; blocks below a particular cutoff are either left in situ, or scheduled for extraction to waste. In mine operations, material is extracted and routed to the destination assigned in the original plan. Sparseness in the underlying geological information, and consequently inaccurate models and plans, leads to poor execution in reality, with up to 30% of material dispatched being misclassified. This leads to a combination of increased cost due to waste dilution, as well as value destruction relating to ore losses. Present sample/assay-based approaches to grade control and ore routing are either inaccurate (due to poor representivity of underlying samples), or do not deliver results in time to support improved decision making during loading or dispatch operations. A new approach to classification of mined materials is proposed where suitable mineral sensors are installed on loading assets in mines such as shovels (surface), scoops (underground) and loaders, whose outputs are integrated with mine planning and ore routing systems. This approach can significantly improve ore/waste classification and dispatch to correct destinations, reducing both losses and dilution. MineSense has over the past 2 years worked on the development of fast, robust XRF sensors with extended range that are compatible with shovel-based systems used in mining. Work has included sensor and signal chain development, testing on applicable ores for algorithm development, and piloting on a 10t loader with ores. This paper covers the development and testing of the sensor to date, and outlines a use case and impacts of implementation at a copper-gold mine with similar mineralogy to the ores used in testing.