creates AI algorithm to predict thickener performance

A UK-based startup says it has devised a machine learning-algorithm that can help mining companies predict how thickeners will operate an hour into the future., which has been helped along the way by Digital Catapult (an agency for the early adoption of advanced digital technologies) and the UK’s Department for International Trade, said it wanted to help the mining industry become more efficient and sustainable by harnessing the power of artificial intelligence.

“Traditional operations technology cannot handle dynamic conditions, so is focused on using advanced digital technologies to create a platform that can predict varying conditions and is, therefore, far more responsive to change,” it said.

This led the company to develop an application to control thickeners in mining operations, which, says, would provide three key benefits:

  • Less water would be needed to complete the thickening process;
  • More water could be recycled, resulting in less wastewater;
  • Reduced power would be consumed as less water would be pumped into the thickener.

To create an algorithm, needed to analyse three years’ worth of data from six thickeners, each measuring roughly 800 different metrics collected every minute.

“This represents a volume of data that would only be possible with a significant amount of computer power and specialist expertise,” the company said. This led to applying to join Machine Intelligence Garage, Digital Catapult’s AI programme that helps businesses access the computation power and expertise they need to develop and build machine learning and artificial intelligence solutions.

Thanks to this assistance, has devised an AI tool that ingests these 800 different metrics every minute and can, according to the company, “predict how thickeners will operate an hour in the future”.

“This invaluable knowledge will make mining more efficient and sustainable, and provides optimum operating condition recommendations to maximise output,” the company said.

The thickener algorithm has since been applied in an optimisation stability project at a gold-copper mine in Chile where the miner in question had seen low underflow percentage solids and water recovery, and high flocculant consumption.

The implementation of the Thickener Circuit Optimisation application at the mine, which integrated data from SCADA and other control systems with advanced statistical data modelling and machine learning algorithms and first principle models, came up with a solution.

This has seen, among other benefits, decreased variability in the thickener circuit operation, enhanced water recovery at the thickener circuit and reduced equipment downtime due to stricter torque constraints.

The payback period has been less than 12 months with projected direct savings calculated at $400,000 in the first year alone, according to

The company has also signed a memorandum of cooperation with JSC AK Altynalmas, a gold producer in Kazakhstan. This involves the development of an AI system for predictive analysis and optimisation of the grinding process, according to

This agreement is part of a wider pact around the implementation of industry 4.0, says.