A robust understanding of what lies underneath the surface of the earth is fundamental to mining, and nowhere more so than in resource-rich Western Australia. However, it’s typically a complex and time-consuming endeavour.

Building on a research partnership that spans the past 10 years, a new $6.1 million four-year research partnership between UWA and Rio Tinto Iron Ore (RTIO) is set to change that by developing innovative data science solutions for automated geology logging.

^{UWA’s Centre for Data-driven Geoscience team (from left to right): Dr Tom Horrocks; Mr David Nathan; Dr Minh Tran; Prof Eun-Jung Holden; Dr Chris Gonzalez; Mr Luke Smith; Dr Daniel Wedge; and Mr Tasman Gillfeather-Clark.}

For the vast scale of world-class iron ore mining operations in the Pilbara, geological information was routinely logged at mine sites, Dr Daniel Wedge from UWA’s Centre for Data-driven Geoscience(CDG) at the School of Earth Sciences explained.

“Rock samples from drillholes, in particular, provide crucial details about the potential size and shape of the ore body, making the accuracy of logging these essential for mine operations and planning,” Dr Wedge said.

“Until recently, geologists, metallurgists and geotechnical engineers have had to manually interpret and record materials found in drill core samples.

“The visual and textural ambiguity of some mineralogical types in small chip samples means that this logging has relied heavily on geologists’ ability and subjective biases, meaning different geologists produce different results.”

The new project will use machine learning, computer vision, spatial modelling and optimisation techniques to integrate diverse drill hole data including spectroscopy, photographic imagery, geochemistry and geophysics data in order to model material compositions, geomechanical proxies and their spatial distribution, ultimately leading to improved mining practices.

It’s yet another example of how data science, which unlocks value and meaning from vast volumes of information, is being adopted to assist industries across the world.

^{Dr Minh Tran from the UWA Centre for Data-driven Geoscience demonstrating her virtual reality software}

“The new project involves entwining advanced data science, geological understanding of the data, and Rio Tinto Iron Ore’s extensive knowledge of the mining area."

Building on a strong 10-year relationship

Dr Angus McFarlane, RTIO Principal, Ore and Product Characterisation, said past partnerships between the two organisations resulted in UWA’s commercialisation of automated downhole image analysis software,and three RTIO-driven joint patent applications on machine learningbased modelling of geology.

“The UWA team has already successfully developed machine learning-based methods and tools for the analysis of stratigraphy and their material compositions for resource evaluation,” Dr McFarlane said.

“The latest engagement will adapt and extend some of these advances for mining, the next stage of the industry workflow from resource evaluation.”

Professor Eun-Jung Holden, who leads CDG and UWA’s Data Institute, said, the new project involved entwining advanced data science, geological understanding of the data, and RTIO’s extensive knowledge of the mining area.

“Applying machine learningbased solutions tailored for industry is a big challenge,” said Professor Holden, who recently received the top honour in the Artificial Intelligence Applied to Mining category at the Women in AI Awards 2022 (Australia and New Zealand).

“As a research team, we benefited greatly by being integrated into our sponsor’s teams, to get experience of their current day-to-day practices and geological knowledge.”

Dr Tom Horrocks from CDG said that knowledge was crucial.

“Improving the team’s understanding of what end-user geoscientists do with the data at various stages, what they want to do but currently cannot, and how best to integrate these new solutions into existing infrastructure is critical for building industry applications such as these,” Dr Horrocks said.

RTIO Manager Geoscience & Water, Research, Development and Technology, Tom Green, described the project as important for the future.

“UWA and RTIO teams have developed a respectful and collaborative culture that returns mutual benefits,” he said.

“We’re now are at the forefront of transforming our geological interpretation using data science through this partnership.”

Download a print copy of Uniview Winter 2022 to read the full edition. An accessible version is also available.