Freeport turns to artificial intelligence to raise copper output by 90,000 tonnes

Freeport-McMoran, one of the world’s biggest copper producers, is targeting a 90,000-tonne increase in annual output from the introduction of machine learning technology at its mines.

The US-listed company has been testing an artificial intelligence model at its Bagdad mine in Arizona, which it is now planning to roll out across all of its operations in the Americas. The move will lift its yearly copper production by about 5 per cent.

“We have set an aspirational goal of adding 200m pounds [90,000 tonnes] of copper from these initiatives with very little capital investment,” Richard Adkerson, chief executive, said in an interview in London.

Copper is critical to the switch away from fossil fuels to renewable energy, with the red metal used in wind turbines, batteries that power electric vehicles and charging points.

Typically a project to develop new capacity for 90,000 tonnes of copper, worth about $500m at current prices, would cost about $1.5bn to $2bn and involve buying new haul trucks, giant shovels and ore crushing equipment.

“It’s been a remarkable success,” Mr Adkerson said of the trial, which has boosted production at Bagdad by 9,000 tonnes this year.

Compared with many other sectors, the level of technological maturity in mining is still relatively low. But that is starting to change as the industry faces pressure to reduce carbon emissions and its physical footprint.

At ageing deposits such as Bagdad — where prospectors first staked their claims in 1882 and most of the best copper has been extracted — miners are having to crush more lower-grade rock just to sustain output.

The machine learning model, which Freeport developed with McKinsey, the consultancy, uses data from sensors around the mine and suggests new ways to improve the performance of its crushers and processing mills, according to the company.

The system found that the mine was producing seven distinct types of ore and that the processing method, which involves the use of large flotation tanks, could be adjusted to recover more copper by adjusting the PH level.

The machine-learning programme is one of three major initiatives Freeport says will boost copper production by 30 per cent. The other projects are the transition to underground mining at its giant Grasberg mine in Indonesia and the Lone Star, a new copper development in Arizona.

Concerns about the Grasberg underground project have weighed on Freeport’s share price, which has flatlined this year. However, Mr Adkerson, said the Grasberg project was very different to Oyu Tolgoi, the underground copper mine Rio Tinto is struggling to develop in Mongolia’s Gobi desert. “They are sinking a shaft and we are going in horizontally. And these are ore bodies we have mined . . . since the early ’90s. We know the rock,” said Mr Adkerson.

Once the switch to underground mining is complete in 2020, Grasberg will be producing more than 700,000 tonnes of copper annually, making it the second biggest mine in the world, as well as 1.8m ounces of gold.

“Those volumes are sustainable for 20 years and beyond,” said Mr Adkerson, adding that excess cash would be used to pay down debt and step up shareholder returns.

“The market is heavily discounting the risk associated with the underground transition. We are confident we can complete it but we have to do it,” he said.