The Mipac team spent three days on the stand at the AusIMM Mill Operators Conference in Brisbane, 19 to 21 October. Here’s what plant managers, metallurgists and control engineers told us, and what we took from the program.
"My GM wants APC and AI"
We heard some version of this line more than any other. Our first question back was usually simple: how’s your base layer? Too often the honest answer was “not great”. Instruments needed attention, loops weren’t performing, and operating data wasn’t being tracked well. By base layer, we mean the fundamentals: reliable instrumentation, well-performing regulatory control loops and trustworthy operational data.
"A lot of the conversations started with APC or AI. Our first question was usually: how's your base layer? If the instruments, PID loops and data aren't right, that's where we'd start. Advanced control needs solid foundations underneath it."
For sites where data capture is the gap, the first step can be practical rather than glamorous. Digital logsheets were a recurring suggestion in those conversations.
The automation pyramid on our stand turned out to be the best conversation starter. Visitors could point straight to the layer where their problems sat, and the discussion went from there
AI: understand the problem before applying the tool
AI was a big theme of the program. The opening panel covered AI and data visualisation, and Chris Greet’s paper asked plainly whether AI is real. MMG Dugald River reported 18 months of closed-loop AI flotation control, including a randomised on/off trial that showed a 0.53 per cent improvement in zinc recovery.
On the floor, the mood was more cautious. Many metallurgists we spoke with see real value in AI as an analysis tool. Handing it direct control of the plant is another matter.
Some people are hesitant to trust AI directly, but a lot of metallurgists understand its value as a tool. There’s still caution around letting it directly control the plant, and context matters. A recommendation can look sensible for one part of the process but be completely wrong when you consider the behaviour of the wider circuit. AI needs that context too'd start. Advanced control needs solid foundations underneath it."
Why APC gets switched off
Agnico Eagle Mine’s Goldex operation presented grinding and flotation APC that prevents SAG overloads, and Palabora presented layered flotation control. Both reinforced a point the team kept coming back to: sustained APC performance depends on maintaining the systems underneath it, including the models, instrumentation and PID loops. When those deteriorate, APC performance can deteriorate with them and operators can lose confidence in the system.
Maintainability matters too. Sites want APC solutions their own teams can understand, support and sustain as process conditions change. If every model adjustment requires specialist vendor intervention, maintaining performance over the life of the system becomes more difficultAgni.
"When models degrade, instruments become unreliable or the underlying control loops aren't maintained, APC performance can deteriorate and operators can lose confidence in the system. Once operators lose confidence, APC can spend more time switched off than delivering value."
Cyclones don't grind
Brian Putland’s ‘Cyclones don’t grind’ keynote challenged conventional thinking around comminution and overall circuit performance. It’s a theme our team recognises.
"Control philosophy is incredibly important, and it should go hand in hand with the metallurgical operating philosophy of the circuit. Focusing on one control loop and optimising it in isolation won't necessarily optimise the performance of the overall circuit."
That intersection between process understanding and control design is where Mipac works.
New sensors: measure what you can act on
The program featured PGNAA for ore blending control at Kanmantoo, CSIRO’s online slurry XRD/XRF, gas hold-up measurement in flotation cells, and acoustic monitoring of grinding mills. The team’s view was mixed. From our team’s perspective, CSIRO’s online XRD/XRF looks promising, with reliability in slurry service an important consideration for any operating deployment. PGNAA is an established technology, but the broader point is that live assay data becomes increasingly valuable when it can feed directly into advanced control and optimisation.
The bigger question is what you do with a new measurement once you have it.
"Measuring a variable is only the first step. The real value is whether the measurement can be used to improve control and deliver meaningful performance outcomes. Any hardware introduced to site should provide actionable insights that support better decision-making and drive measurable production improvements, rather than simply adding another source of data to the historian."
The takeaway
The appetite for AI and APC across the industry is real. What we heard at MillOps is that, in many cases, the foundations still need attention. Get the base layer right, keep the whole circuit in view, and only add measurements you can act on.
If you’re considering APC or AI, the first question may not be which technology to choose, but whether the control foundations beneath it are ready.
Thanks to AusIMM, the Advisory Committee and everyone who stopped by the stand and shared their experience.
