Three tests, and a list of things we decided not to buy.

A deliberately unfashionable approach

Mining has a long history of expensive technology programmes that produced conference presentations and very little ore. The pattern is consistent: a novel system is deployed at one site by a corporate team, it works in the pilot, it is never adopted by the people who have to use it at three in the morning, and it is quietly retired two budget cycles later.

Quintel applies three tests before funding anything. Does it remove a person from a place where they could be killed? Does it reduce cost per tonne by an amount that survives a conservative estimate? Can the people at the operation maintain it without a specialist flying in? A project that fails the third test will fail eventually regardless of how well it passes the first two.

Nothing on this page is a first-of-a-kind deployment. Everything described here exists in commercial operation elsewhere in the industry, which is precisely why the group is willing to spend money on it.

Removing people from harm

Proximity detection and collision avoidance. Vehicle interaction is the fatal risk that appears most often in the group’s high-potential incident record and in the industry’s fatality statistics. Proximity detection is fitted to 94% of the mobile fleet across the group, with full coverage targeted by the end of 2026. The system provides operator warning and, on underground equipment at Mbengwe and Asankran, automatic slow-down.

Remote and automated drilling. Nine of twelve underground production drills at Mbengwe are operated from a surface control room. The operator is not in the stope. That is the entire justification; the productivity gain of roughly 11% is a secondary benefit.

Fatigue monitoring. Every haul truck in the open pit fleets carries an in-cab fatigue detection system with real-time intervention. Data is used to change roster design, not to discipline individuals — a distinction the group had to make explicitly and defend before the system was accepted by the workforce.

Slope and seismic monitoring. Radar slope monitoring at all open pits, and a fourteen-station microseismic array at Asankran underground with an automatic exclusion protocol triggered by events above magnitude 1.5. Two seismic events in 2024 caused localised damage in areas that had already been cleared by the protocol.

Energy and emissions technology

Underground fleet electrification. The Mbengwe conveyor decline approved in 2026 removes 41 diesel haul trucks from the underground ramp, cutting underground diesel consumption by an estimated 62% and reducing primary ventilation load by 4.2 MW. Ventilation is the largest single electrical consumer in an underground mine, and diesel is the reason it has to be that large.

Solar with storage. Three plants operating — 28 MW at Asankran, 45 MW at Mbengwe and 12 MW at Gamagara — with 22 MWh of battery storage at Mbengwe. A 20 MW installation at Fatala is approved for 2027 and a 32 MW plant is designed into the Bagoé build rather than planned as a retrofit.

Comminution optimisation. Crushing and grinding consume roughly 40% of the electricity at a concentrator. Mill circuit optimisation using online particle size analysis and blast fragmentation control has reduced specific energy consumption by 11% since 2021, which in absolute terms exceeds the output of the Gamagara solar plant.

Ventilation on demand. Airflow at Mbengwe underground is modulated by real-time personnel and equipment tracking rather than run at full volume continuously, saving an estimated 8.4 GWh a year.

Geoscience and the orebody

The most valuable technology in mining is usually the least visible: knowing more precisely where the ore is. A one percent improvement in grade control at Mbengwe is worth more than any automation project in the group.

Machine-learning grade estimation. Applied at Lubumba, where cobalt grade varies over metres rather than tens of metres and conventional kriging performs poorly. The model is used alongside, not instead of, conventional estimation and conditional simulation, and its predictions are reconciled against mined grade monthly. Where it has underperformed, that has been reported.

Portable XRF grade control. 1,400 samples a week at Lubumba, giving grade control decisions within hours rather than days and allowing selective mining at a scale the laboratory turnaround could not previously support.

Geophysics under cover. The Erongo programme depends entirely on airborne electromagnetics and gravity to map basin architecture beneath 40 to 120 metres of Kalahari sand. Historical explorers in that district could not see through the cover, which is the reason the ground was available.

Digital orebody models. Every operation maintains a single reconciled geological model used by geology, mine planning and metallurgy. This sounds administrative and is not: separate models maintained by separate departments is a documented cause of mine plan failure across the industry.

Where we have not gone, and why

Full surface haulage automation. Autonomous haulage is commercially proven at very large scale in Australian iron ore and Chilean copper. At Quintel’s pit sizes the capital and system integration cost is not recovered, and the technology requires a level of road and traffic control discipline that is a management problem rather than a technology one. The group has chosen to solve the management problem first.

Battery surface haul trucks. Not commercially available at 220 tonnes for our duty cycles and ambient temperatures. The group has committed not to purchase a new diesel surface haul truck after 2032 and participates in two OEM development programmes, but will not publish a target that depends on equipment nobody sells.

Blockchain traceability. Evaluated in 2022 and not adopted. The group’s chain-of-custody weakness is not the immutability of the record, it is the physical reconciliation of material at the point of transfer. A distributed ledger records a claim; it does not verify that the truck contained what the ticket said. Investment went into weighbridges, seals and GPS instead.