Get ready for the 2027 storm season by fixing the storm report first.

When the last member is back on and the mutual-aid crews have gone home, the storm is over for almost everyone except the people who have to report on it. Indices have to be calculated, major event days identified, and an after-action report written for the board, the commission or both. That work often takes weeks, mostly because the timeline has to be pieced together from systems that were never designed to agree.

Scrutiny of that work increased through 2026, and there is every reason to plan for more of it in 2027. This article covers what regulators are asking for now, the measures behind the numbers, and what to put in place before the next storm season.

What 2026 tells us about post-storm scrutiny

Two 2026 examples show the direction.

In July 2026, after severe thunderstorms over the Fourth of July holiday, the Michigan Public Service Commission opened an investigation and directed three utilities to file reports by August 27, 2026. The commission asked for the deployment of resources including the allocation of crews, detailed evaluations of public and customer communication efforts including how accurate restoration estimates were, and any other pertinent issues or lessons learned ¹. Inaccurate restoration estimates were a central complaint.

In Rhode Island, a storm report filed in April 2026 follows a familiar structure: an executive summary, incident anticipation, the storm and its impact, restoration, communications during and after the event, technology issues, and a conclusion ². The underlying settlement requires a description of the storm, a summary of the damage, the number and duration of outages, and the planning and restoration activities ².

Whether or not your utility answers to a state commission, expect boards, councils and members to ask similar questions after the next major storm: what happened, when, how the utility responded, how long it took, how accurate the estimates were, and what will change.

The indices behind the numbers

Most storm and annual reliability reporting relies on indices defined in IEEE 1366, the IEEE Guide for Electric Power Distribution Reliability Indices ³⁴. The core indices count sustained interruptions, those lasting more than five minutes. Interruptions of five minutes or less are momentary ⁴.

• SAIFI (System Average Interruption Frequency Index): the average number of sustained interruptions per customer over a period ⁴⁵.
• SAIDI (System Average Interruption Duration Index): the average minutes of sustained interruption per customer served over a period ⁴⁵.
• CAIDI (Customer Average Interruption Duration Index): the average time to restore service to interrupted customers, calculated by dividing duration by frequency, that is, SAIDI divided by SAIFI ⁴⁵.
• MAIFI (Momentary Average Interruption Frequency Index): the average number of momentary interruptions per customer ⁴.

Because these indices depend on customer counts and durations, every interruption record needs a reliable start time, restoration time and number of customers affected. When a feeder is restored in steps, each step needs its own time and customer count, or customer minutes will be wrong.

For scale, EIA noted in August 2026 that customers in the mainland United States generally experience about two hours of electricity interruptions per year ⁶.

Major event days, in plain terms

A severe storm can dominate a year's numbers and hide what is happening on ordinary days. IEEE 1366 handles this with the major event day: a day on which daily SAIDI exceeds a statistically derived threshold ⁴⁶.

The threshold comes from what is known as the 2.5 beta method. Daily SAIDI values from the previous five years ⁶ are transformed with a natural logarithm, and the threshold is set at 2.5 standard deviations above the mean of those values, which isolates rare, high-impact days ³. Any day in the reporting period above the threshold is a major event day.

Major event days are not thrown away. Utilities commonly report performance both including and excluding major events ⁴. A June 2026 Lawrence Berkeley National Laboratory report adds a planning point: excluding major events helps track day-to-day reliability, but resilience planning is about exactly those larger events ³. Expect more interest in how utilities perform on their worst days, not only on typical ones.

Why the timeline is hard to rebuild

Almost every item in a storm report depends on an accurate timeline, and that timeline is spread across systems.

Outage start times come from the OMS, AMI last-gasp messages or member calls, and they do not always agree. Device operations sit in SCADA. Crew assignments may be in the OMS, a dispatch board or on a whiteboard. Switching steps are on paper orders or in a separate application. Field observations travel by radio and phone, and some are written down hours later. Member notifications come from an IVR or messaging platform. Mutual-aid crews often work outside the utility's own tools.

Each source has its own clock and identifiers. When a feeder is restored in several steps, or a second fault sits behind the first, matching customers to times becomes manual work, and errors flow straight into SAIDI, CAIDI and the major event day calculation. Restoration estimates suffer the same way: if the field picture reaches the people giving estimates late, the estimates will be wrong.

What to set up before the 2027 storm season

Over the next 12 to 24 months, plan to treat each outage as a single incident record that collects its history as the work happens:

1. One identifier from first report to restoration. Calls, meter events, crew assignments, switching steps, field notes and restoration steps attach to the same incident.
2. Timestamps at the moment of action. Each step records when it happened and who did it, not when someone remembered to type it up.
3. Step restoration recorded as it occurs. Customer counts and times for each partial restoration are captured on the record, so customer minutes do not need to be estimated later.
4. Restoration estimates tied to field status. Estimates should update from the same record crews are working in, and the history of each estimate should be kept so its accuracy can be reviewed.
5. A report template agreed in advance. Decide now which sections your board or commission will expect, and map each to the records that feed it.
6. A dry run. Before the 2027 season, produce a storm report from a recent event and time how long it takes. Then remove the steps that depend on memory.

Where GridSteward fits

Amplyra GridSteward carries one outage record through prediction, crew assignment, field work and switching, restoration and evidence, so the timeline is assembled as the storm is worked. Its Outage Response workspace includes the live outage map, crew status, customer notifications, restoration estimates and IEEE 1366 reliability indices, and outage analytics break restoration performance down by stage, cause, crew and feeder. Reports carry their review and approval status with them. In a pilot, results are measured against your current process rather than assumed.

To see a storm timeline built in one record, Book a guided tour.

← All insights