ChileChemicals and materials2020 – ongoingCodelco
Machine learning for ore processing at Chuquicamata
Codelco said machine learning added about 8,000 tonnes of copper a year at its century-old Chuquicamata mine
Problem
Chile's state copper company faces falling ore grades at ageing mines, so it must move and process more rock for the same copper, while water limits and labour disputes also cut output. New deposits are costlier and harder to develop.
Approach
In 2020 Codelco set up a digital data centre that applies machine learning to a continuous stream of data on extracted ore and uses it to optimise processing, for example by deciding how to blend ores. At Chuquicamata it also optimised the chain from mine to concentrator and installed advanced process control in grinding and flotation. Predictive maintenance at the Radomiro Tomic division was slated for extension to Chuquicamata and El Teniente.
Results
Takeaway
Codelco extracted more metal from ore it was already mining as grades declined. Its company-reported gains use different measures: tonnes, profit and a 2% rise in fine copper.
Categories
- Size
- Large enterprise
- Industry
- Chemicals and materials
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and MLOptimization
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
Bloomberg reported that Codelco set up a machine-learning digital data centre in 2020 to analyse ore data and optimise processing, for instance through blending. Codelco said the platform added about 8,000 tonnes of copper and roughly US$80 million in annual profit at Chuquicamata, and planned to extend AI to other operations. The company faced falling ore grades and a 9.3% drop in quarterly output.
The Chilean mining publication describes Codelco's 2021 AI and automation projects. It reports that optimising the mine-to-concentrator chain at Chuquicamata raised fine copper output by 2% a year and that advanced process control was installed in grinding and flotation. It says predictive maintenance generating about US$7 million a year at Radomiro Tomic was to be extended to other divisions.
Full details are in the original sources.
