Deep mine systems · Mechanica Labs

Hover over a number to see what it is. Click any part of the mine to look inside.

A dashed outline marks a stope waiting for the fill beside it to cure. The ring shows a filled stope gaining strength.

Fig. 1 A deep mine drawn as one system. Ore leaves the stopes, goes up the shaft and through the processing plant. Tailings go to the tailings facility or to the paste plant, which pumps them back underground as backfill. A filled stope has to cure before the stope beside it can be mined. Hover over a number to see what each part is, and click a part to see it in detail. “New realization” mines a different version of the same orebody.

An orebody nobody has seen

Before a stope is mined, the orebody has been sampled only along drill holes. Between the holes, its grade and the other properties that matter have to be estimated. Geostatistical simulation produces many versions of the orebody, called realizations. Every realization agrees with the drill holes. They differ between the holes, and that difference is the geological uncertainty.

Realization 1

Realization 2

Realization 3

low grade → high grade
Drill-hole samples
Orebody outline

Fig. 2 Three realizations of one vertical section through an orebody. All three agree with the five drill holes, where the grade was measured. Between the holes they differ.

Geometallurgical models then link these properties to how the ore will behave in the plant. They predict how fast it can be processed, how much of the metal can be recovered, and what the tailings will be like. So the uncertainty does not stay in the ground. It travels with the ore.

Where the uncertainty is absorbed

Whether that uncertainty reaches production depends on the path the ore takes. A large stockpile mixes ore from many places and smooths out the variation. A small bin, a busy hoist or a plant already running at capacity passes it straight on.

Grade (%)

Ore leaving the stopes ±0.00

Ore entering the plant ±0.00

Day

Stockpile 3 days of feed

Middle 80% of realizations
Median

Fig. 3 Thirty realizations of the orebody above, mined in the same sequence. The top chart is the grade of the ore leaving the stopes each day. The bottom chart is the grade entering the plant after the stockpile. A larger stockpile absorbs the day-to-day swings, but the slower differences between realizations still reach the plant.

The backfill loop

As shallow deposits run out, mines go deeper, where the rock is under higher stress and harder to hold up. In open stoping, a mined-out stope is filled with cemented backfill, which then acts as an artificial pillar. The stope next to it can only be mined once the fill is strong enough.

The fill is made from the plant’s own tailings, dewatered and mixed with water and binder. This closes a loop. The geology of one stope changes the tailings, the tailings change how fast the fill gains strength, and the fill decides when the next stope can be mined. Backfill is also a large operating cost in some mines, and a delay in delivering it delays production.

Three questions

The work asks how geological and geometallurgical uncertainty moves through a deep mine whose parts depend on each other, and how much of the mine a model has to represent before its operating decisions can be trusted. It has three parts.

1. Which uncertainties reach production?

Realizations of the orebody are run through a model of the whole mine over time. Some variations are absorbed on the way. Others change the production shortfall, the amount of metal recovered or the timing of extraction. Changing the capacity of storage, haulage and the plant shows when each happens.

2. How much detail does a reliable decision need?

Each part of the mine that turns out to matter is modelled twice, once in detail and once simply. For backfill, the detailed model lets strength depend on the recipe and on the tailings, and the simple one uses a fixed curing time. Operating policies are optimized under each model and then tested, unchanged, on realizations and models they were not chosen on. A simplification is acceptable when it changes the prediction but not the decision.

3. What is coordination worth?

The extraction sequence, underground haulage, plant operating mode, tailings allocation and backfill recipe can be decided together or one at a time. Both approaches are tested on new realizations, to find out when deciding them together improves production and operating return, and whether that improvement holds up under conditions it was not developed for.

From simulation to a working mine

A first version of the model already exists. It links geological realizations, material flows, backfill and later extraction in one simulation. A paper on simulating coupled mine, processing, tailings and backfill systems has been accepted at APCOM 2027, the international symposium on computers and operations research in the mineral industry, held in Montréal in May 2027.

Mine data and operating conditions come from Mitacs projects on backfill and on underground fleet management, and from collaborations with deep-mine operators. The plan runs over four years. The first two cover how uncertainty propagates and how much detail a model needs. In the third year the results are combined into coordinated operating policies, which are tested first against historical mine data and then alongside the decisions operators actually make. In the third and fourth years, where it is feasible, selected schedules and policies are put into practice with industrial partners and compared with current operating practice.