Choosing mine planning optimisation software: what actually matters

Two different problems wear the same word

"Optimisation" in mine planning covers two jobs that are not the same job.

Strategic planning decides what to mine over years or decades: pit limits, phase design, cut-off grade, long-term sequencing and the NPV that falls out of those choices. Tactical planning decides what to mine next week or next month: short-term scheduling, blending, equipment assignment and haulage. A strategic solver is not a slower tactical one. The maths differs, the constraints differ, and the input data differs.

Most sites need both. The strategic plan sets the destination, the tactical plan keeps production on track.

What we hear from planning teams

Two patterns come up repeatedly when we talk to planners and geologists.

The first is that solve time, not judgment, decides how much detail a model can carry. Every model is a careful trade: how much operational reality it can hold and still solve in time, still be trusted, still be defended in a review. When the solver is the bottleneck, detail gets left out to keep runs feasible, and the model sits further from the operation you already understand better than it does.

The second is that expertise outruns the tooling. Geologists and planners hold deep knowledge of the orebody and how the operation really runs. Too often that knowledge cannot reach a viable recommendation while the decision is still live. The judgment is there. The path from judgment to a defensible answer is what runs out of time.

What solve speed actually buys you

Speed matters because good planning is iterative. A model that takes four hours to run leaves you one answer and no time to test it, however many questions you still have. That is where value leaks out of a plan, and it is the tool's ceiling doing it, not yours.

But a fast number on its own tells you nothing. Before you trust any solver, ask three things about the result, not just the clock:

• Does it respect your hard constraints? A plan that returns in seconds but violates slope angles, minimum mining width or plant capacity is not a fast plan, it is a wrong one you now have to catch by hand.
• How good is the answer? Some strategic problems do have provably optimal methods. Ultimate pit limits, for one, can be solved to a certified optimum, and that is worth having. But multi-period production scheduling with capacity, precedence and blending is a different animal, and at real scale it is rarely solved to proven optimality, so tools fall back on heuristics or report an optimality gap. For that work, what you need is a high-quality feasible plan and enough runs to see whether the shape of the answer holds as your assumptions move. Ask a vendor which of the two they are doing, and treat a blanket optimality claim across both with caution.
• Can you run it enough times to trust it? The real output of strategic work is not one plan. It is the comparison across commodity prices, cost curves and plant configurations that tells you which option is robust. A solver you can only afford to run twice cannot give you that.

Judge speed by how many honest scenarios it lets you run before a decision is due, not by a single headline time.

Whether the tool fits your workflow

A planning tool that cannot read your block model or write back to your planning package becomes an island, and the cost of getting data in and out repeats on every run. Before you evaluate any product, map your current data flow and check it against these.

• Block model formats. Does it accept CSV, OMF, and the exports from the packages your team already runs? If you have to reformat by hand, that tax compounds across every scenario.
• Schedule exports. Results need to feed the short-term scheduling tools and spreadsheets your operations team already trusts. A plan nobody downstream can open does not get used.
• Geometry and domains. Strategic work needs pit shells, phase shells and geotechnical domains to go in cleanly.
• Provenance. Mines are long-life assets, and you will need to know which assumptions produced a given result years after the engineer who ran it has moved on. Ask how the tool records inputs, parameters and model versions.
• Data residency. If you upload client block models, confirm where the data sits and who can reach it. For Australian operations this is often a contractual line, not a preference.

Cloud or desktop

Desktop gives you control over hardware and data and keeps files local. Cloud removes installation, scales compute on demand, and makes it far easier for several people to view, rerun and compare scenarios rather than waiting on one engineer's workstation. For strategic studies, where the work is scenario-heavy and collaborative, that difference is the point. The decision usually comes down to data policy and how many people need to touch the plan.

Where Cognomine fits

Cognomine is a strategic mine planning solver. Give it a block model and a set of scenario assumptions, and it returns long-term schedules with the economic sensitivities that go with them, while respecting the hard constraints you set: slope, width, capacity, precedence. Large raw models are re-blocked to a workable size before loading, which is standard for strategic work and matches the aggregated model your team would build for a study like this anyway.

It is built for one thing above all: running many scenarios fast enough that you actually run them. Studies that take days on conventional workflows return in minutes, which changes how a planning team works. The runs you always wanted to test, price down, costs up, plant reconfigured, come back onto the table, and you walk into the decision knowing which plan holds across all of them rather than which one the clock allowed. Your read on the orebody carries into the model at the detail it deserves and still returns an answer while the decision is live.

It uses metaheuristic methods, so it does not hand you a certified optimum, and it does not pretend to. For multi-period scheduling and option studies, that trade is the right one: a high-quality feasible plan you can generate twenty times over is worth more than a single slow run you cannot afford to repeat. Every result carries its inputs and assumptions with it, so it stands up when someone reopens the study later.

If your day is dominated by short-term dispatch and daily truck routing, a dedicated tactical tool will serve you better. If your work is life-of-mine scheduling, pit and phase studies, or strategic option analysis, that is what Cognomine is for.

A short checklist

• Decide whether you need strategic, tactical, or both, and buy accordingly.
• Confirm the tool reads and writes the formats your team already uses.
• Model your real constraints in a trial, not just the vendor's demo dataset.
• Judge speed by how many scenarios you can run, and check that every result respects your hard constraints.
• Ask how inputs, parameters and versions are recorded.
• Verify data residency and access control if client data is involved.

Choosing planning software is less about which tool has the most features and more about which one matches the decisions you actually have to make.