Skip to content
Meet us at CIGRE Paris Session 2026, August 23–28
Eneryield
All news
Insights

Understanding SAIDI and SAIFI — Grid Reliability Metrics

The two numbers every operator answers for

SAIDI (System Average Interruption Duration Index) measures how many minutes of outage the average customer experiences per year. SAIFI (System Average Interruption Frequency Index) measures how often the average customer is interrupted. Regulators track them, benchmark them and in many markets tie revenue to them.

The definitions are simple: SAIDI is the total customer interruption minutes in a period divided by the number of customers served; SAIFI is the total number of customer interruptions divided by customers served. Divide one by the other and you get CAIDI — the average duration of a single interruption. Together they are the standard reliability scorecard for distribution system operators worldwide.

What actually drives them

SAIFI is driven by how many faults strike your network; SAIDI by how many strike and how long each takes to locate, isolate and repair. Traditional levers work the duration side: faster crew dispatch, automated switching, better fault localization. All valuable — but they start after the fault.

Prediction attacks both indices at the root

A fault that is predicted and prevented never enters either index. That is what makes fault prediction categorically different from faster restoration:

  • SAIFI falls because evolving faults — degrading insulation, vegetation contact, water ingress in cables — are repaired on schedule before they become interruptions.
  • SAIDI falls twice: fewer interruptions, and the faults that do occur arrive with predicted location and root cause, cutting localization time.

At Vaasan Sähköverkko in Finland, 64 faults have been predicted in advance over five years of continuous service — with warnings up to three weeks ahead. Weeks to inspect, plan and fix during working hours instead of restoring at 3 a.m.

The practical takeaway

If your improvement program only contains restoration measures, you’re managing the consequences of faults. Adding prediction — running on the standard measurement infrastructure you already have — is the most direct way to move both indices at once.

See IntelliView® in action

A 30-minute demo on real grid data — and a conversation about your network.