EN
Back to the archive

The encyclopedia · Engineering & Operations · Strategic decision · 2015–2018

Chuquicamata's scenario plan cushions bad years without losing profit

Codelco's Chuquicamata used scenario-based scheduling to make underground copper extraction robust to conveyor failures.

Codelco

the move

Chuquicamata, Codelco's copper mine in Chile, was moving from open-pit to underground extraction. The bottleneck was an overland conveyor whose failures made extraction capacity unpredictable, so a plan that looked optimal on average could fail badly in a bad year.

Researchers built a deterministic model representing the plan the mine could implement without the study, then extended it to a multistage stochastic model that generates capacity scenarios and penalizes deviations from an ex ante profit target.

Simulations showed the stochastic schedule achieved the same expected profit as the deterministic one, but earned on average 5% more in scenarios below the 10th percentile of earnings. Setting the target 2% below the deterministic expectation raised that cushion to 9%.

why it works

  • Equipment failure is a tail risk; average-case plans get punished exactly when the conveyor breaks.
  • Penalizing deviations from a profit target forces the optimizer to build slack into the schedule.
  • The stochastic plan traded nothing on expectation, so there was no cost to robustness.
  • Scenario modeling made the risk of each extraction schedule visible to mine planners.
the payoffOptimize for bad scenarios, not just the averageclever

what transfers

Plan for variability explicitly: the robust plan kept expected profit while lifting worst-decile earnings by 5-9%.

what came after

The stochastic framework became part of how Chuquicamata evaluated long-term extraction schedules under uncertainty. The study was published in Interfaces and highlighted how accounting for conveyor reliability changes mine sequencing decisions that run for decades.

references

spotted an error? The archive wants to know.

same kind of clever