Data-driven strategies to optimise Ma’aden Barrick Copper Company (MBCC) flotation circuit − The power of visualisation and machine learning in data mining-

• Novel approach integrating machine learning with traditional process optimization methods. • Enhanced prediction accuracy for mineral processing plant performance. • Application of advanced data analytics and visualization to optimize recovery and efficiency. • Comparative analysis showcasing improvements over conventional models. • Practical insights for engineers to streamline operations and maximize output. In this work, we present a case study demonstrating the application of Optimaviz, an advanced data analytics platform, in addressing the challenges of flotation optimization in the mining industry. The study utilizes historical data from the Jabal Sayid underground copper mine, operated by Ma’aden Barrick Copper Company (MBCC), a joint venture between Ma’aden and Barrick Corporations. Our findings reveal that maintaining cyclone feed slurry percentage solids within the range of 65–71 wt% is crucial for achieving high plant performance (rougher tails copper grade < 0.15 wt% and combined concentrate grade > 24 wt%), with deviations from this prescribed range resulting in a significant decrease in performance. Additionally, we observe the impact of the first concentrate grade (rougher concentrate grade) on overall plant performance, requiring first concentrate grade to be above 26 wt% to achieve high performance, highlighting the importance of the efficiency of the first three rougher cells on the overall plant performance. Furthermore, the study underscores the significance of the balance of power between the SAG mill and Ball mill in achieving a finer rougher feed size distribution that is required to promote flotation performance. This study demonstrates how Optimaviz can be used to effectively optimize mineral processing circuits, providing process engineers and metallurgists with a robust tool to derive actionable insights and enhance plant performance without the need for coding knowledge.