Production and guidance
Ophir
What should each mine dig, feed and shut down over the next 13 weeks to deliver the most payable copper, and what guidance can the board stand behind?
Payable copper, kt a week
- Re-plan, median
- 10th to 90th percentile
- Plan as booked
- Recommended change
Week
03Next quarter’s guidanceBoard view 1.8 kt above the median
35 kt36 kt37 kt38 kt39 kt
01Payable copper, next 13 weeks+1.1 kt
Plan as bookedAdded by the re-plan
- 01
Mill reline, week 6Decision now
Booked into the highest-grade week of the quarter
- Exposure
- 1,300 tPayable copper
- Expected effect
- +600 to 900 tPayable copper
Move the reline to week 9Maintenance planner, before Friday’s schedule lock
- 02
Plant feed, weeks 2 to 4Awaiting approval
Transitional ore recovering below plan
- Exposure
- 700 tPayable copper
- Expected effect
- +300 to 450 tPayable copper
Blend the high-grade stockpile into the feedPlant metallurgist, from Monday
- 03
Next quarter’s guidanceAwaiting approval
The board’s 39 kt sits above the forecast’s 90th percentile
- Exposure
- 39 ktBoard view
- Expected effect
- 80%Chance of landing in the range
Guide 36 to 38.5 ktChief financial officer, before the quarterly update
- Payable copper at risk
- 2,000 t
- Recovered by the re-plan
- 900 to 1,350 t
- Guidance range
- 36 to 38.5 kt
What your systems can’t see.
- The mine plan, the plant and the maintenance schedule live in different systems, so nobody sees what a change in one costs the others in metal.
- The block model, the mill and the assays drift apart, and the gap is found at month end.
- Guidance comes from the annual plan as a single number, with no honest view of the odds of meeting it.
- Shutdowns are booked to suit the maintenance calendar, which is rarely the week that loses the least metal.
It reads what you already keep.
Each system remains the authority for its own data. The engine reconciles identities, units and dates at each decision cutoff and flags any conflict.
- 01Short-term and annual mine plans
- 02Block model extract
- 03Fleet shift data
- 04Plant historian and assays
- 05Maintenance schedule and downtime log
- 06Budget versions
Inside the engine.
Forecasts are settled before the options are scored. Permissions and hard constraints always take priority.
| Component | Responsibility | Effect on the decision |
|---|---|---|
| Machine learning | Forecast payable metal by week and quarter as a range, the drift between block model, mill and assays, throughput and recovery by ore type, equipment availability and failure risk, and how long each shutdown will run. | Sets the forecast used to compare the options |
| Rules and constraints | Enforce the pit sequence and geotechnical limits, stockpile capacity, plant and tailings limits, fleet and crew availability, contracted concentrate deliveries and the approved budget, and provide the fallback when a model cannot be used. | Removes actions that are not allowed or not feasible |
| Optimisation | Choose the mining sequence, plant feed and shutdown timing for the next 13 weeks that deliver the most payable metal within those limits. | Chooses the set of actions to take |
What it tells you to do.
Each recommendation names the action, its target, owner and deadline, the constraints it respects and the evidence that confirms completion.
- 01Re-sequence mining for the next 13 weeks
- 02Change the plant feed blend by ore type
- 03Move a planned shutdown to the week it costs the least metal
- 04Set the quarter’s guidance as a range, with the odds of meeting it
- 05Write the approved plan back to the planning and maintenance systems
How value is measured
Payable metal is measured against your own forecast, and forecast error at one, three and twelve months. Uplift counts only where a recommendation was adopted, against a counterfactual agreed with your technical services lead.
Built into the way you already work.
The engine runs in a dedicated AWS account or your own infrastructure, connects to your systems and uses models calibrated on your records.
- Scope
- Every operation, re-planned every week
- Connected systems
- Mine plans, block model, fleet, plant historian, assays and maintenance
- Approval
- Your planning lead signs off every plan before it is written back
We prove the case before you commit.
Can 12 months of your plans, fleet, plant and downtime records show how much metal was lost to causes you control, and could a range forecast have beaten your own?
What we need to see
- 12 months of budget and forecast versions
- Fleet shift data and daily plant and metallurgical accounting
- A downtime log, and one live connection to the fleet system or plant historian
- A named owner for the mine plan, with the authority to change it
When we would advise stopping
- Planned and actual records cannot be reconciled period by period
- Most of the lost metal comes from causes no plan can control
- No named owner has the authority to change the plan
Assess this decision in your operation.
Start with a baseline review on a slice of your records, or go straight to the opportunity assessment of the full decision, before a pilot begins.
Book a live walkthrough