In a constantly changing environment, like that of an iron ore mine, optimising production requires a constant readjustment of mining equipment to provide steady production, consistent quality and a reliably delivered product.
Add to that the challenges of scheduling and control of automated equipment in remote locations, operated from a centralised office more than 1500 kilometres away.
Rio Tinto Iron Ore has already invested billions of dollars in automated drilling equipment, trucks and trains as part of its Mines of the Future initiative to improve productivity at its Pilbara mining complex in Western Australia. And in 2016 it began deploying complex scheduling software developed in collaboration with the University of Melbourne to further optimise these operations.
At the university’s Department of Computing and Information Systems, Associate Professor Adrian Pearce has led the five-year software project, jointly funded by Rio Tinto and an Australian Government Research Council grant.
He says the aim of the Rio Tinto project has been to harmonise the operations of the robotic mining equipment in the Pilbara, which is controlled from an office in Perth. To do that his project team drew together researchers with skills in optimising multi-agent systems, constraint solving and optimisation, mathematics, computer science, software engineering and artificial intelligence planning.
“Mathematically it has been challenging, because each time you mine a block of ore, the shape of the mine changes and all the paths to the other ore blocks change. You have to determine how to sequence the operations – the different grades of ore, different locations within the mine – to keep the crushers fed and produce a blended ore that meets customers’ needs while maximising the life of the mine.
“It uses a lot of energy to mine and move ore, so you have to move only what you need to, when you need to, in order to keep the flow up to crushers to maintain productivity.”
Associate Professor Pearce says the long-term aim is to improve the productivity of the mines and get the best return on investment from both the ore resource and the mining equipment for company shareholders, and for the Australian economy.
While the programming challenge has originated in the mining sphere, he says it could be equally applied in other areas, from agriculture and infrastructure management to health care, defence and even potentially retail sales and e-commerce.
“It represents a systematic approach to optimise production over a complex supply chain, and we have an appetite to expand its application,” he says.