Codelco suspends Andes Norte over emerging seismic risk: when stopping becomes an engineering decision

Did 1,782 workers return to Andes Norte?

No. Codelco said it had created reassignment options for 1,782 people, or 72% of a target workforce of 2,460. Gradual reincorporation was scheduled from 7 September in work such as Project Diamante, recovery of production areas and cross-functional project services. No later official notice confirming the reopening of Andes Norte was located by the evidence cut-off.

Another 678 workers remained suspended without a reassignment option, with short-term pay continuity while alternatives were assessed. People returning to other work is not the same as construction and development restarting at Andes Norte.

What changed in the evidence

On 4 August, Codelco temporarily suspended Andes Norte development and construction as a preventive measure. The company said six months of technical information had identified an emerging seismic phenomenon with characteristics different from risks historically known and managed at El Teniente.

That wording matters. Codelco did not announce an imminent collapse or establish a final cause. It described a possible emerging risk associated with greater mining depth, still under investigation, and a need to strengthen instrumentation, monitoring and technical analysis.

A known hazard can still change

El Teniente has long managed mining-related seismicity in panel-caving operations. The emerging issue was not the existence of seismicity in the abstract, but observed behaviour that no longer appeared to fit the historical model with the same level of confidence.

Emerging risk may arise when new evidence changes the understood mechanism, frequency, magnitude, spatial pattern or interaction of a familiar hazard. Engineering does not have to pretend certainty while waiting for a final explanation. It must recognise when the confidence supporting continued exposure has weakened.

Depth, panel caving and rock-mass response

Andes Norte, Andesita and Diamante form the Teniente project portfolio intended to deepen the operation toward Teniente 9. Codelco describes roughly two billion tonnes of resources and reserves and a plan supporting 137,000 tonnes per day for more than 50 years.

Greater depth can alter stress, confinement and stored strain energy, but depth alone does not determine rockburst risk. Geology, structures, extraction sequence, geometry, support, production rate and stress redistribution interact. Panel caving is not inherently unsafe; it requires the rock mass response to remain understood and controlled.

Seismic event is not a synonym for rockburst

A seismic event is an energy release detected in the rock mass. A rockburst involves violent damage to an excavation associated with energy release. Not every event damages workings, and not every recorded change supports an exact forecast of where and when damage will occur.

Monitoring informs interpretation and decisions; it does not promise deterministic prediction. This distinction prevents two errors: treating every event as a catastrophe and treating a lack of exact prediction as a reason to ignore a changing trend.

A sensor does not protect anyone by itself

Seismic arrays, convergence measurements, extensometers, mapping and inspections can reveal behaviour, but raw data may contain noise, communication failures or changes in sensitivity. Competent validation and interpretation stand between a reading and an operational decision.

Monitoring without decision criteria is observation. A mature chain connects data, validation, trend, threshold, action, restriction or stop work, analysis and re-entry criteria. The practical question is simple: when a metric leaves its expected envelope, who has the authority to act?

Stop work as a risk control

Stopping is neither automatic proof of excellence nor an admission of failure. When uncertainty exceeds confidence in available controls, suspension can become the barrier that prevents continued exposure while knowledge is rebuilt.

A pause buys time. Engineering must convert that time into understanding and control: strengthen instrumentation, challenge assumptions, test scenarios, revise sequences or support, define exclusion zones and establish triggers. Otherwise, a pause is only delay.

Safe re-entry needs positive evidence

Re-entry is not the automatic opposite of stopping. It should demonstrate what changed: whether data were validated, the model was revised, additional controls were installed and verified, and withdrawal and response triggers are understood.

A quiet period alone does not prove that the rock mass returned to its former state. Re-entry needs accountable technical approval, communication to workers and contractors and enhanced surveillance. The aim is not zero-risk language; it is a defensible basis for controlled exposure.

Critical controls and information integrity

Critical controls are those whose absence or failure materially increases the likelihood or consequence of a major event. Depending on the scenario, they may include mine design, extraction sequence, ground support, monitoring, exclusion zones, withdrawal protocols and re-entry authorisation.

The 2025 fatal rockburst at El Teniente forms part of the wider learning and governance context, but it must not be presented as having the same cause as the 2026 Andes Norte decision. One transferable lesson is that technical information must reach decision-makers without its uncertainty being filtered away. Information integrity can itself be a critical control.

Workers, contractors and the cost of stopping

The suspension affected contractor companies and people assigned to development and construction. Safety decisions have human consequences: uncertainty, reassignment, contract changes and pressure on households and communities. Mature management does not pretend that stopping is free.

It prevents economic pressure from erasing technical evidence while managing the impact responsibly. Codelco's options for 1,782 people and temporary pay continuity for 678 belong in that conversation. Schedules can be renegotiated; physics cannot.

Chile's framework and a Brazilian comparison

Chile's DS 132 provides the regulatory context for underground mine safety, ground support, inspections and corrective responses. The 4 August suspension was announced as Codelco's preventive decision; this article does not claim that Sernageomin ordered it.

Brazil's NR-22 does not apply in Chile. As a comparison for Brazilian readers, however, section 22.13 links rock-mass stability to monitoring and requires work to stop when geotechnical assessments indicate potential instability. Different jurisdictions converge on a technical principle: continuity cannot remain the default when the ground indicates otherwise.

Executive questions for any critical risk

This logic extends beyond mining to mechanical integrity, dams, structures, energy, automation and climate.

• What is normal behaviour, and what change is technically significant?
• Who validates the data and owns the interpretation?
• Which thresholds trigger attention, restriction and stop work?
• Who can stop exposure?
• Are critical controls verified in the field?
• Is there a formal re-entry standard?
• Can new evidence reach senior leadership without distortion?
• Can production pressure override the technical decision?

Conclusion: monitoring sees; control decides

As 1,782 workers were scheduled to begin moving into other work, Andes Norte remained suspended. That distinction captures the case. Reassignment can mitigate human impact; it cannot replace technical understanding of the deep-seismicity phenomenon under study.

When risk changes behaviour, engineering's first obligation is to admit that it may not yet know enough. Monitoring sees. Understanding interprets. Control decides before evidence becomes consequence.