AI in high-hazard operations: illustrative cases

These cases are illustrative, not real incidents. Each draws on operational patterns seen across several facilities, and each ends with the questions M-OPS OMI would have asked. All names of companies, sites, equipment and people in them are invented; any resemblance to a real organisation, site or event is coincidental.

Predictive maintenance and asset integrity

The Corlinda caseIllustrative case: a predictive maintenance model performs as designed, every procedure is followed, and a cryogenic cold box fails after an AI-supported deferral.

Subsurface and wells

The Vinmora caseIllustrative case: an AI model trained on an adjacent field rates hole collapse as low risk and losses as high; the team mitigates losses and the hole collapses.

Control room and process control

The Linmara caseIllustrative case: an AI soft sensor passes its acceptance tests on normal data, then reports sweet gas while H₂S breaks through on a sour gas plant.
The Quillmere caseIllustrative case: a chemical site writes an AI policy and register, while a vendor's model update changes control room alarm ranking outside management of change.

Updated 7 October 2026.