The ‘Digi-MIN’ team was made up of Gekko graduate process engineer Eliza Craig, graduate metallurgist Barry Tuncks and mechanical designer Matt Kurtze, as well as Monash University student Daniel Bechaz.

Digi-MIN picked up its first place ribbon after tackling an unplanned downtime and rate interruption event within Newcrest’s concentrators at the Cadia mine in New South Wales that caused major business and safety exposures.

Craig said the team developed a tool that flagged small variations in the correlation between two related variables, and were able to show the correlation between two variables in the plant’s high pressure grinding rolls changed a number of times before a single downtime event.

The algorithm developed was robust enough to be able to flag the changes more than 24 hours beforehand, and could be applied easily to any two related variables across any piece of equipment or site.

Craig said by flagging small changes in the correlation in real time, equipment and downtime could be inspected and planned to cut costs and reduce the impact of unplanned downtime on site.

The Hackathon awards $5000 cash and technology and co-working prizes to the winning teams.

According to Unearthed growth manager Mikey Kailis this was not the primary motivation for hackers at the event.

He said the real prize was the opportunity to work, and continue working with companies such as Newcrest after the event.

Unearthed Melbourne was held at coding skills school General Assembly, where more than 100 participants spent the 54-hours looking at solutions to real-world challenges using operational data sets provided by Newcrest.

Digi-MIN’s winning algorithm could also be applied to any related variables across any piece of equipment or site, to inspect equipment and plan downtime.

Second place went to Fossil Fools for its SpargeSense, an acoustic signature used to detect breakage of sparge tubes in an autoclave at Newcrest’s Lihir mine.

Fossil Fools team member Chris Mummery said the acoustic response for an intact tube was captured and machine learning routine established a three-dimensional acoustic signature based on frequency response, harmonics, decay and amplitude.

The signature was compared against successive measurements and any deviation from the standard signature was flagged as a potential tube breakage.

Third prize went to RoxOn, RoxOff, which developed adaptive driver performance training to optimise Cadia underground loaders.

The People’s Choice Award went to Team Neptune for its real-time autoclave performance optimiser and Team Monash for its prototype Uptime, which accurately predicted 16% of downtime failures five minutes before they happened using machine learning.

Newcrest chief information officer Gavin Wood said the company saw tremendous value in open innovation initiatives and was impressed by the calibre of talent on display at the event.

He said the fourth sponsorship of Unearthed was a great success, with the teams coming up with some insightful and interesting approaches to the four challenges.

Woods congratulated Digi-MIN for its win, and said the company looked forward to working with a selection of the teams to further explore the great ideas they generated.