Case Study

Significant recovery uplift with prescriptive analytics for gold processing

Case study: Significant recovery uplift with prescriptive analytics for gold processing

Significant recovery uplift with prescriptive analytics for gold processing

About the concept

Gold recovery from polymetallic ore is a complex process requiring crushing, screening, gravity separation, flotation and smelting. Application of reagents used in flotation depends on the mineral composition of ore, its gold content and impurities and the grind size. If ore properties change often, the output of the flotation circuit suffers from high variability. Analytics helps optimise flotation circuit set points to maximize gold recovery at the highest throughput while reducing variability on the outputs.

About our partner

This project was done together with Partners in Performance with DataStories team working remotely and Partners in Performance team working onsite with the client.

Industry

Mining

Market

Worldwide

Technology Used
  • DataStories A.I. Platform for modelling
  • DataStories SDK for model deployment, simulation
  • DataStories’ Virtual Processing Expert for implementation

The challenge

Our client, a major gold miner and processor hired us to assess and prioritise business improvement opportunities. Operational data was available and logged properly in a historian and custom Excel-based metallurgical accounting system for weekly and monthly reporting. However, the data sources were not connected and not used to drive improvements. We were asked to demonstrate the benefit of data and analytics on the most important profit driver - gold recovery.

The solution

  • We connected and analysed 550+ operation parameters to determine optimal tradeoffs between recovery and throughput and assess if recovery at flotation can be accurately predicted on a fine frequency.
  • We constructed prescriptive models for four flotation outputs - recovery of gold and the second base-mineral, and variability of both recoveries.
  • The models were validated by the experts and used to simulate the added value from dynamic parameter changes for any given head grade and grind size combinations.
  • Solutions were rated by implementation complexity since some set points were easier to change than others.
  • Resulting scenarios, implementation roadmaps and cost-benefit estimations were presented to the client to choose an implementation strategy.

The impact we created

  • 3.2p.p. In Au recovery improvement opportunity by changing two flotation set points
  • 61% reduction in the Au recovery variability (standard deviation)

Actual recovery over a 6 month period and optimised recovery using the Virtual Expert with dynamic changes of 3 set points Actual recovery over a 6 month period and optimised recovery using the Virtual Expert with dynamic changes of 3 set points

Change of recovery in percentages

Talk to us about how you can significantly uplift the recovery in your processing plant