Long-Term Production Scheduling

Long-term production scheduling assigns mining activities and material movements to periods over a planning horizon. It determines when selected mining units become available, how much is extracted and where the material goes, subject to physical and operating restrictions. Formal open-pit scheduling models distinguish these time-dependent decisions from an ultimate-pit calculation, which identifies an extraction envelope without scheduling its production.^[1]

The schedule translates a development strategy into a testable production trajectory. It connects future feed and product to the preparatory work, capacity and expenditure needed to deliver them. Its usefulness depends on whether the representation preserves the constraints that matter at the intended decision scale.

Define the Decisions and Their Resolution

Specify the starting state, planning horizon and decision periods before constructing the model. The starting state includes completed extraction and development, available stockpiles, commissioned facilities and activities already in progress. Material already mined must not remain available for extraction merely because the geological model has not been depleted.

Define the scheduling unit separately from the geological estimation unit. A block model can be much finer than the material the operation can mine or direct independently. MineLib explicitly treats blocks as modelling constructs and notes that simple slope precedences can leave a selected block physically inaccessible from its own level. Its models therefore cannot, by themselves, demonstrate an executable operating layout.^[1]

The choice of resolution creates a trade-off. Combining units can conceal differences in access or processing suitability; separating them can imply selectivity the operation does not possess. Choose units around the decisions to be tested, then inspect the detailed material and geometry inside them. A favourable average for a large unit is insufficient if its unfavourable portion must be mined first.

Time resolution requires the same discipline. A period total says little about the order of events within that period. If access is completed near its end, production dependent on that access cannot be assumed available throughout it. Use shorter periods or explicit timing relationships where such transitions affect feasibility. A coarse representation may remain appropriate farther into the future where detailed execution is not yet supported by information.

Translate Engineering Relationships into Constraints

Constraints describe what the schedule is permitted to do. An objective describes what is preferred among permitted outcomes. A capacity that can be exceeded by paying a penalty is a soft target; a physical or safety restriction does not become optional merely because an objective assigns it a high cost.

For open pits, precedences represent the removal needed before deeper material can be extracted. Capacity restrictions then limit activities within each period, while destination decisions link extracted material to processing or disposal. MineLib's formulations illustrate these distinct relationships and also state their simplifying assumptions, including fixed spatial precedences and known input data.^[1]

Underground scheduling needs relationships appropriate to the mining method. In the sublevel-stoping model of Nehring, Topal and Little, stope development, drilling, extraction and backfilling form linked production phases, with constraints on adjacent activity and fill exposure. These are features of the modelled system, not a prescription for every underground mine. The demonstration assumes that development serving several stopes has already been completed, an important limitation when applying the formulation to a new area.^[2]

Distinguish ordering from elapsed time. Saying that backfilling precedes nearby extraction is incomplete if the engineering design also requires a period for the fill to achieve its intended condition. Likewise, a construction milestone may precede production but still require commissioning before its full service becomes available. The necessary lag comes from the relevant design and evidence; the scheduling model should not invent it.

Represent Capacity at the Right Interface

A production target must be checked against the activities that deliver it. Ore tonnes alone do not describe waste movement, development effort or transport demand. For each potentially binding resource, identify the work imposed by a scheduled unit and the usable capacity in the same period and on the same measurement basis.

Nehring and colleagues' conceptual example separately constrains development, production drilling, backfill and shaft haulage. It uses predefined activity profiles, and the proposed formulation links successive phases without significant delays. This illustrates why a total ore-production ceiling cannot represent all underground constraints, and why those timing assumptions need checking before using such a model.^[2]

Capacity may depend on the material. Navarra and colleagues' geometallurgical framework represents processing time through material-specific throughput and operating mode. Its proposed modes also have different recoveries and costs. The framework omits the time and cost of changing modes; these effects require separate consideration if they are material to the operating plan.^[3]

An engineering implication is that equal feed tonnages need not consume equal plant time. A schedule that substitutes slower-processing feed for faster-processing feed should recheck treatment capacity even if its tonnes remain unchanged. A proposed mode change also needs the equipment and operating conditions assumed by the model.

Check shared resources at their common interface. Two production areas can each satisfy their local transport limit while collectively exceeding a shared route or handling facility. Conversely, treating every local maximum as simultaneous demand can make an otherwise workable schedule unnecessarily restrictive. The model needs the actual concurrency assumptions and an explanation of how shared capacity is allocated.

Keep Extraction, Storage and Processing Consistent

Stockpiles connect periods, allowing extraction and processing to occur at different times. Their representation must account for inventory and reclaim composition. Moreno and colleagues show that permitting independent retrieval of original blocks from a mixed stockpile can create an optimistic bound that is infeasible under a homogeneous-mixing model.^[4]

A simple conservation check, consistent with their inventory balance, is:

S_t = S_(t-1) + A_t - R_t

Here S_t is closing stock, A_t additions during period t, and R_t reclaimed material; all are dry tonnes. The expression assumes no material losses or other transfers. Initial stock must be specified; additions, reclaim and closing stock must be non-negative, and storage capacity respected. Where reclaim occurs before additions, also impose R_t <= S_(t-1). Moreno and colleagues' basic model uses this ordering; it is a modelling assumption, not a universal operating rule.^[4]

The inventory of each tracked component, measured as mass, also needs checking. Conserving total tonnes cannot demonstrate that the required grade is available at reclaim. Rehandling consumes resources and incurs costs, while material left at the horizon needs an explicit treatment rather than silently becoming saleable product.^[4]

Closing balances alone do not establish that storage limits are respected at every point within a period; check inventory peaks against the assumed receipt and reclaim sequence. These distinctions also matter when diagnosing a feed shortfall. More material in storage does not necessarily solve the problem if it cannot be reclaimed, processed or combined into an acceptable feed within the required period. The remedy must address the particular restriction.

Compare Value on a Consistent Basis

Economic scheduling values the timing of production and its associated costs. State what the objective includes before interpreting its result. Nehring and colleagues explicitly limit their net present value (NPV) calculation to activities represented in the model; their formulation excludes taxation and depreciation. Such an objective is not automatically a complete project valuation.^[2]

For a fair comparison, use the same valuation date, currency convention, price assumptions and treatment of expenditure. Check whether mining costs occur when material is extracted, processing costs when it is treated, and product receipts at the assumed commercial point. If stockpile processing is deferred, assigning all revenue to extraction can reward a sequence that cannot deliver those receipts.

Also inspect the planning horizon. A schedule can appear better by excluding costs needed for the planned continuation beyond the final modelled period or by using early high-value material while leaving an expensive continuation. Compare terminal conditions: remaining inventory, unfinished development, residual stockpiles and infrastructure commitments. Their treatment should be consistent across alternatives and with the larger mine plan.

Examine Uncertainty and Information Timing

Navarra and colleagues illustrate a two-stage approach in which extraction decisions are shared across geological scenarios while later processing decisions can vary by scenario. This distinction makes explicit which commitments are made before uncertainty is resolved and which responses remain available afterward. Its outcome is conditional on the scenario representation and the assumed opportunities to adapt.^[3]

Use this distinction when assessing any scenario study. Revaluing a fixed schedule under changed assumptions tests exposure. Reoptimizing independently for every future describes what could be achieved with advance knowledge of that future. A practicable response must respect when information arrives and what decisions have already become irreversible.

Stress tests should inspect the cause of failure, not only a change in NPV. A lower grade can affect product output; slower development can remove access to otherwise suitable feed. Each can produce a similar revenue shortfall but demand a different response. Record the periods, locations and constraints affected so that further investigation or contingency planning has a clear target.

Validate the Schedule Before Handing It Down

Begin with accounting checks: no unit is mined twice; all extracted material has a destination; opening and closing inventories balance; and period totals reproduce the reported production profile. Then review spatial stages and activity timelines with the responsible disciplines. Check the beginning and end of the horizon as well as the steady production periods.

Treat model feasibility and operational feasibility as separate tests. A mathematically valid solution can still depend on unrealistic within-period sequencing, selective recovery from a mixed stockpile or unavailable construction access. These are reasons to improve the representation and rerun the comparison, not reasons to waive the missing restriction.

Finally, disaggregate the near-term portion sufficiently to test the transition to execution. Record the capacities, milestones and feed requirements passed to detailed planners, together with their permitted flexibility. When detailed scheduling cannot meet them, identify whether the conflict arises from data, aggregation or a genuine strategic limitation. A controlled revision should retain the previous baseline and explain how the corrected schedule changes future access, output and commitments.