Shell operates one of the world's largest energy infrastructure networks, including refineries, chemical plants, offshore platforms, and pipelines across six continents. Unplanned equipment failures at these facilities created significant safety risks, production losses, and environmental hazards. Traditional maintenance strategies relied on fixed schedules or reactive repairs after failures occurred — intervening either too early (wasting serviceable component life) or too late (resulting in catastrophic failures). The diversity of equipment types across manufacturers and operating conditions made it infeasible to develop bespoke models for every asset class.
Data-quality challenges compounded the problem: sensor calibration drift, missing timestamps, and inconsistent tagging across facilities eroded the reliability of historical training datasets. Shell's assets spanned geopolitical environments with widely divergent regulatory regimes, from North Sea decommissioning obligations to Gulf of Mexico well-integrity mandates. Any algorithmic maintenance recommendation required passage through formal management-of-change processes, with hazard identification workshops involving operations supervisors and process-safety engineers.
Shell deployed an AI-powered predictive maintenance platform in partnership with C3 AI, scaling to monitor over 10,000 pieces of critical equipment — including control valves, pumps, and compressors — across upstream, manufacturing, and integrated gas assets globally. The technical infrastructure ingests 20 billion rows of data weekly from more than 3 million sensors and runs nearly 11,000 machine learning models in production.
The platform applies semi-supervised anomaly detection across equipment families, enabling knowledge transfer from data-rich assets to sparsely instrumented counterparts. A robust data-ingestion pipeline incorporates automated quality checks — including sensor-drift correction, outlier flagging, and missing-value imputation. The system generates remaining-useful-life estimates with calibrated uncertainty bounds, allowing maintenance planners to optimise intervention schedules based on risk-adjusted cost functions. Integration with SAP automates work-order generation and spare-parts procurement when predicted failure windows approach.
In 2021, Shell and C3 AI signed a five-year strategic agreement to accelerate enterprise AI deployment. Shell has since commercialised its AI predictive maintenance applications through the Open Energy AI initiative (OAI), an open ecosystem co-founded with C3 AI, Baker Hughes, and Microsoft to advance AI adoption across the energy sector.
This is an industry case study based on publicly available information. Shell is not a Pertama Partners client.
We help enterprises across Southeast Asia design and deliver AI transformation programs. Let’s talk about what’s possible for your team.
