Rollout approach for autonomous haulage at an open-pit operation

An open-pit mine decides how to introduce autonomous haul trucks: a mixed-fleet pilot, conversion pit by pit, a single fleet-wide campaign, or full conversion of one pit with operation segregated rath...

Published content

1. node: Context. The organisation operates an open-pit metalliferous mine with three active pits and a haulage fleet of 42 trucks, on a site with an established fatigue management programme and a light-vehicle interaction control already in place. Haulage is the largest single operating cost and the largest source of vehicle interaction hazard on site. Procurement of the autonomous system has been approved, and the supplier needs the rollout approach fixed this quarter to schedule truck configuration. The site operates under a state mining safety regulator that expects a documented safety case and traffic management plan before autonomous equipment operates.
2. node: How should autonomous haulage be introduced at the operation: a small pilot in one pit, conversion pit by pit, conversion of the whole fleet in a single campaign, or full conversion of one pit with no mixed operation anywhere?
3. node: Rationale. The pilot tests the technology against this operation's own haul profile, which is the assumption most likely to be wrong in the published figures.
4. node: Rationale. Under phased conversion, lessons from the first pit inform the second and third, and capital is released in tranches against demonstrated results rather than all at once.
5. node: Rationale. A single campaign minimises the period of mixed operation, the state the industry guideline requires to be explicitly controlled, and reaches the benefit soonest.
6. node: Rationale. Segregating by pit removes the interaction hazard between changeover windows rather than controlling it throughout. The guideline's treatment of mixed operation as a state requiring explicit control is best satisfied by an arrangement that does not enter that state at all.
7. node: Decision-maker. The General Manager of the operation, with the capital release above USD 25 million requiring approval by the corporate investment committee. The Head of Health and Safety holds a separate approval on the traffic management plan and interaction control, which is not delegable to the operation.
8. node: Scope. The decision covers how autonomous haulage is introduced, not whether the technology is procured, which has been approved. It does not cover autonomous drilling or loading, which are separate proposals on a later cycle.
9. node: Rationale. A fully converted pit produces a throughput figure for this operation's own haul profile, unpaced by manned traffic, which is the number every other option assumes at the point it commits most of its capital.
10. node: Consulted. Haul truck operators and their union representatives under the enterprise agreement; mine planning on the consequences of losing inter-pit fleet flexibility; the technology supplier on configuration for the haul profile; and the mining safety regulator on the safety case and traffic management plan before the first changeover.
11. node: Risk. A pilot maximises the very state the guideline requires to be explicitly controlled: a small autonomous fleet mixed with a large manned one, for a long period, with the interaction control exercised constantly. It buys information about throughput at the cost of sustained exposure to the interaction hazard.
12. node: Risk. Thirty months of mixed operation is the longest exposure of any option to the hazard the guideline requires to be explicitly controlled. Phasing reduces commercial risk by increasing the duration of a safety risk, and the two are not in the same currency.
13. node: Risk. It commits the whole capital against figures no one has reproduced on this haul profile. If the gain lands at the bottom of the range or below it, the operation has converted a fleet it cannot un-convert on a business case that did not hold.
14. node: Action. Conclude the enterprise agreement consultation and make the retraining offer before conversion begins, not alongside it. The research identifies workforce transition rather than technology as where implementations fail, and the obligation is a precondition rather than a parallel workstream.
15. node: Constraint. Any period in which autonomous and manned trucks share haul roads requires a documented interaction control and a revised traffic management plan approved before operation. Mixed operation is permitted but is a controlled state rather than a transitional convenience.
16. node: Risk. The throughput result from a mixed pit will not answer the question the pilot was run to answer, because the autonomous trucks are constrained by the manned traffic around them. The pilot can demonstrate that the system works; it cannot demonstrate what it is worth.
17. node: Risk. A three-year conversion assumes the mine plan holds for three years. Two pits have sequences that would be costly to alter once conversion has started in them.
18. node: Risk. One hundred and twenty-eight operator roles change in a single step, against an enterprise agreement requiring consultation and an offer of retraining. Concentrating that into one change is the workforce transition the research identifies as where implementations most often fail.
19. node: Risk. Segregation constrains the mine plan: trucks cannot be moved between pits to meet a schedule change, so flexibility the operation currently relies on is lost for the duration.
20. node: Action. Approve the traffic management plan and interaction control for each two-month changeover window before any autonomous truck operates in that pit, and stand the control down explicitly when the pit's segregation is complete rather than letting it lapse.
21. node: Constraint. The operation has an enterprise agreement obligation to consult on changes affecting operator roles, and to offer retraining before redeployment. No option may commence conversion before that consultation has concluded.
22. node: Risk. It requires the first pit to be one where full conversion is practical, which is not necessarily the pit where the business case is strongest. The option may prove the technology on the least representative ground available.
23. node: Action. Measure throughput in the converted pit against the same pit's manned baseline for the preceding twelve months, not against fleet averages. The purpose of converting one pit fully is to obtain a comparison the published figures cannot provide.
24. node: Industry guideline. The Global Mining Guidelines Group's guideline for the implementation of autonomous systems sets out implementation as a staged process with defined readiness criteria, and treats the interaction between autonomous and non-autonomous equipment as a hazard to be controlled explicitly rather than managed by operator judgement. The mixed-fleet criterion below exists because of that treatment.
25. node: Action. Select the first pit on how representative its haul profile is of the other two, as well as on how practical full conversion is there, and record the reasons, so that the measured result can be read across to the pits that follow.
26. node: Research. Published work on implementing autonomous haulage trucks reports safety benefits arising from reduced human exposure to the haulage environment and from removing fatigue and operator error as accident mechanisms, alongside management challenges in workforce transition and in the new hazards the systems introduce. Both halves of that finding are represented in the options below.
27. node: Industry evidence. Reported results from operations that have converted include throughput increases of the order of 15 to 20 per cent and unit cost reductions of around 15 per cent, with some operators reporting higher figures. These are operator-reported results from specific sites rather than controlled comparisons, and the operations reporting them differ from this one in ore body, haul profile and fleet age. No controlled study is cited for the range; it is used here only as a starting assumption to be tested.
28. node: Review trigger. If the measured throughput gain in the converted pit falls below 10 per cent, do not extend to the second pit. The business case for the remaining conversion rests on the reported range holding here, and this is where that is tested.
29. node: Research. Work on technology innovation and implementation barriers in mining identifies workforce transition and the integration of autonomous with existing systems as the barriers on which implementations most often fail, rather than the technology itself. The roles criterion and the consultation constraint carry that finding into this decision.
30. node: Review trigger. If any interaction event occurs during a changeover window, halt and review the interaction control before proceeding. The whole argument for segregation is that the operation spends as little time as possible in that state.
31. node: Assumption. The published throughput gains are attainable here at the lower end of the reported range. The haul profile at this operation is steeper than at the operations reporting the highest figures, and no adjustment for that has been calculated.
32. node: Review trigger. If the mine plan changes such that inter-pit fleet movement becomes necessary before conversion is complete, reassess. Segregation trades flexibility for hazard removal, and that trade is only sound while the plan holds.
33. node: Assumption. Mine plan stability over the conversion period. Two of the three pits have a planned sequence that would be difficult to alter once conversion has begun in them, and a change to the plan mid-conversion would strand equipment configured for a specific pit.
34. node: Assumption. Each pit can be given its own haul route and tipping point at the crusher and waste dumps, so that a converted pit's trucks never share road with another pit's manned trucks. Where that is not possible, segregation weakens to a controlled interface at the point where the routes converge.
35. node: Pilot six trucks in one pit alongside the manned fleet
36. node: Convert pit by pit over three years, trucks converted progressively within each pit alongside manned trucks
37. node: Convert the whole fleet in a single campaign
38. node: Convert one pit fully, with autonomous and manned operation segregated by pit rather than mixed on shared roads
39. node: Recommendation. Convert one pit fully, segregated from manned operation, then extend pit by pit on the measured result. It produces a throughput figure measured on this operation's own haul profile before most of the capital is committed, which no other option does, while committing about 30 per cent of the capital the single campaign requires. It spends about six months in mixed operation, in three two-month changeover windows, against thirty for the phased option; that is two months more than the single campaign, accepted in exchange for the measurement and consultation rounds a third the size. Segregation removes the hazard between windows where phasing controls it for longer, at the cost of inter-pit fleet flexibility, which the review triggers watch.