Eneryield was selected for the EPRI Incubatenergy Labs 2021 cohort — one of 20 startups chosen globally — and paired with the New York Power Authority for a 16-week utility demonstration.
The project applied machine learning to historical measurement data from the Y-49 Long Island Sound Cable, a 23-mile, 693 MW submarine transmission link, to detect anomalies that precede faults.
The results: incipient cable faults predicted two months in advance at 80% confidence, and 24 hours ahead at close to 99% confidence. The project was covered by T&D World and led to our first transatlantic reference.
What it means for utilities: submarine and underground cables are among the hardest assets on any grid to inspect — and the most expensive to repair unplanned. Demonstrating incipient-fault prediction on a live transmission link, using historical data alone, showed that warning time on critical cable assets is a data problem that AI fault prediction can solve without new sensors.
Read the full story in our NYPA case study.