Why the SAG runs scared
Fear of overload keeps mill load conservative. The safety margin is unmilled tonnes paid for in lost throughput every shift — and a smart instrument can give that margin back.
Read more →Smart instruments built for mineral processing — Rock Analyzer, Froth Analyzer, Ball Counter and Fill Analyzer — give Model Predictive Control a head start, so your plant produces more, with care for the earth. Live in weeks.
Fear of overload keeps mill load conservative. The safety margin is unmilled tonnes paid for in lost throughput every shift — and a smart instrument can give that margin back.
Read more →
Bubble structure and surface velocity shift minutes before the assay catches up. A camera that reads the froth gives the controller a head start on grade.
Read more →
Pebble recycle is the leading indicator of ore hardness. Measure it well, and the SAG and crusher stop fighting each other for tonnes.
Read more →
Impact-rate signatures hide in the vibration of the shell itself — live mill fill and liner health, legible without new sensors. Our Fill Analyzer prototype pulls them out with a digitizer-side DSP pipeline.
Read more →
A decade of froth-imaging research has tightened the link between bubble-size distribution and recovery. Here is where the literature now stands — and where our camera picks up the baton.
Read more →A non-contact camera over the SAG feed belt sizes every parcel of ore as it passes — full PSD reported every few seconds.
Two-class deep network on the feed and discharge counts whole and fractured grinding balls as they enter and leave the mill — quantifying ball flux, wear, and the charge held inside the drum.
A shell-mounted vibration sensor turns the mill’s own noise into a live measurement — charge fill, toe and shoulder angles, and whether the cataract is landing where it should.
Dense optical flow plus bubble segmentation on the flotation cell surface — quantitative velocity, direction, and bubble-size distribution at 250 ms latency.
Multivariable MPC that predicts every CV 15–30 minutes ahead, sees an approaching constraint before it bites, and plans the MV trajectory that keeps the circuit safely inside its limits.
A continuously calibrated model of your full circuit that stays in sync with the plant — the trustworthy picture the controller acts on.
Most APC projects don’t stall on modelling; they stall at the integration with the plant’s DCS, SCADA, PLC, and historian stack. Our team has spent years inside those systems — OPC-UA bridges, PI historians, vendor-specific tag conventions, OT-network constraints — so the path from kick-off to running setpoints in production is measured in weeks, not quarters.
OPC-UA / DCS / historian wired up with our pre-built bridge. Plant tags mapped to the model, with the OT team in the loop. No new sensors.
Advanced instruments and MPC tuned and validated against your historical and live data, reviewed against soft-sensor benchmarks and walked through with your metallurgists and process engineers.
MPC online as operator advice first, then closed-loop on your terms. Operators stay in command; Aeterna holds the limit.
SAG, ball mill, pebble crusher, and classification coordinated as one optimisation problem. The circuit holds the true bottleneck instead of an operator’s safety margin, and P80 stays on target as ore hardness drifts.
Live signals from the froth and grade-aware control of reagent, air, and pulp level — recovery held against ore drift, grade tightened where it pays. The plant stops chasing the assay and starts leading it.
Lower energy per tonne, tighter water-loop control, and less rework all compound into a smaller footprint per gram of metal. Production grows; the impact does not.
Mineral processing, automation, and embedded-system engineers in one team. We understand the process end to end — from feed size to froth — and build instruments that answer the questions metallurgists already ask.
ProcessOpt advises your existing DCS rather than replacing it. Safety and interlock logic stay exactly as commissioned, and your team decides when to move from advice to closed loop.
Every product runs on the ProcessOpt platform, so instruments, the digital twin, and MPC share one data model. Our own automation hardware is in development for plants that want the whole stack from one team.
A 5 Mt/a gold plant, valued at the current spot gold price. The figure is the incremental revenue from the modelled throughput uplift — lower energy per tonne and recovery gains are additional upside.
Gross incremental gold revenue at the live spot price; excludes operating cost per tonne and additional upside from energy and recovery gains. A site study produces a net, plant-specific figure.
Get 7 days of full-licence access to explore the software. Provide 60–90 days of process data and we build an initial process model, simulate how Model Predictive Control would engage on your operation, and show you what it would improve — before any commitment.
Tell us about your operation and we'll give you a straight answer — no pressure, just a conversation.