Five years in service: 64 faults predicted in advance on a Nordic distribution network

The challenge: from reacting to outages to preventing them

A Nordic distribution system operator wanted to move from reacting to outages to preventing them — using the measurement infrastructure already in its network. Together with our digitalization partner, the goals were concrete: reduce the number of supply interruptions, sharpen fault localization, isolation and restoration, and shorten the interruptions that do occur.

Fault prediction from existing relay recordings

Eneryield IntelliView® was commissioned on two rural distribution feeders, reading disturbance recordings from the network’s existing relays — no additional sensors, no new field hardware. The explainable-AI approach was central from the start: every prediction shows how the conclusion was drawn, so the control room can verify the reasoning instead of trusting a black box.

Five years of results

The deployment has been in continuous service since October 2020 — the longest-running Eneryield IntelliView® installation. Across those five years, the two feeders alone produced more than 80,000 disturbance recordings, roughly 44 every day. Out of that stream:

• 64 faults were predicted in advance
• Warnings arrived up to three weeks before the interruption
• Predicted root causes agreed with operator findings in 88% of cases

Why it matters for distribution reliability

This is the reference that answers the question every utility asks: does it keep working outside a pilot? Five years of unbroken operation, on ordinary relay data from ordinary rural feeders, with warnings long enough to turn callouts into planned work.