Metallurgical process optimization using pulp chemistry data

点击翻译按钮获取中文摘要

The rejection of pyrrhotite from copper-nickel (Cu-Ni) concentrates has long been a pain point for the Canadian mining industry. The presence of pyrrhotite in the Ni-Cu concentrate takes up smelter capacity, adversely affects energy efficiency, and generates large quantities of sulphur dioxide. Canada produces ~6-7 % of the world's nickel and together, Vale Base Metals Canada and Glencore's Sudbury Integrated Nickel Operations mined and concentrated ~96 % of the nickel produced in Canada in 2020. It is well established that flotation pulp chemistry influences Cu-Ni mineral flotation performance and pyrrhotite rejection; however, many pulp chemistry parameters (such as dissolved oxygen and pulp potential) go unmeasured in industrial mineral processing plants. In plants where measurements are taken, the data are rarely used to predict or control the process, despite their known influence. Advances in online pulp chemistry sensors present the opportunity to leverage pulp chemistry data for the development of statistical and machine learning models that can better predict and control pyrrhotite behaviour in Cu-Ni flotation circuits. This project is a collaboration between Queen's University, automation specialist SK Godelius, Vale Canada Base Metals, and Glenore's Sudbury Integrated Nickel Operations. The short term objective is to develop models that can predict pyrrhotite content (and other ore characteristics) and corresponding flotation behaviour in the flotation of Cu-Ni ores using pulp chemistry data. Long term, this work will form the foundation of a plant optimization platform that uses process monitoring, modeling, and prediction at the industrial scale to improve plant performance and reduce the carbon footprint of Canadian nickel operations.