Case Study - Flotation Circuit (Cu/Au)- Machine Learning Guidance/Recommendation System

CLIENT OVERVIEW 

Our client operates a copper-gold processing plant and was seeking innovative solutions to enhance flotation circuit efficiency. With complex ore variability and multiple interacting process variables, the business aimed to unlock greater value through AI and advanced data analytics.

THE CHALLENGE 

The key challenge was to optimise the rougher flotation pull rate system to reduce metal loss to tails. This required dynamic optimisation of a dependent KPI using a range of upstream and controllable variables such as ore type, feed grade, milling conditions, and reagent setpoints without disrupting existing operations.

APPROACH 

The project began with large-scale data engineering efforts, enabling live ingestion of process data from the client’s site into Interlate’s analytics platform. Our team of data analysts and scientists then developed machine learning models using historical and live operational data. These models were designed to predict outcomes and generate real-time recommendations, tailored to the current feed and circuit operating conditions. To ensure usability, an intuitive interface was developed, allowing operators to view and understand the system's guidance in real time. Additionally, a parallel flotation circuit was maintained as a control group to directly benchmark the performance improvements resulting from the implementation.

SOLUTION 

An AI powered decision support system was implemented to automate and optimise flotation setpoints. The system provided real time process control guidance that adapted to ore variability, leading to improved recovery and more consistent operations.

IMPACT 

The AI powered optimisation system led to a 0.6% increase in full-plant metal recovery, delivering an estimated annual value of AUD $5 million. Operational strategies identified by the system were adopted across both circuits, further enhancing productivity. The solution not only improved immediate process performance but also influenced broader operational philosophies by highlighting new ways of running the plant more efficiently.