Nordic grid’s low-inertia hours nearly quadrupled versus 2025

If a large generator suddenly disconnects, electricity supply drops. Inside generators still connected to the grid, heavy rotors slow slightly, releasing some of their stored motion energy. This cushions the immediate drop in the grid’s frequency—the roughly 50 Hz rhythm of alternating current—until fast reserves respond. That cushioning effect is called inertia. It comes from many machines already connected to the grid.

The Nordic grid operators estimate that stored kinetic energy every 15 minutes across Finland, Sweden, Norway and eastern Denmark. For this analysis, we counted a clock hour if at least one of its four readings fell below 150 gigawatt-seconds (GWs). Fingrid says 120–280 GWs is typical. We chose 150 GWs, near the bottom of that range, to compare the two years; Fingrid does not publish it as a rule for buying reserves. Fingrid is Finland’s transmission-grid operator. (Fingrid’s inertia explainer; kinetic-energy data)

From January through 26 September 2026, Nordic inertia dipped below 150 GWs at least once in 1,137 separate hours—17.6% of the hours with complete data. Over the same dates in 2025, that happened in 291 hours (4.6%). Such dips were nearly four times as common in 2026, though these records cannot tell us why.

The first seconds after a power loss

Moving parts delay a sudden change but cannot restore balance on their own. Operators prepare a rapid response before it is needed. One such service is Fast Frequency Reserve (FFR), which can add power or cut demand in about a second during low-inertia hours. (Fingrid on reserves; FFR product information)

Too little fast reserve does not immediately cause a blackout. After a major power loss, frequency can fall without customers noticing; at 48.8 Hz, automatic protection starts disconnecting some customers to avert a wider blackout. (Fingrid’s Nordic frequency-control overview)

A 100-megawatt (MW) FFR booking sets aside that much response capacity for an hour. Fingrid pays to keep it ready; the booking is not a record of 100 megawatt-hours (MWh) of electricity delivered. Reserve units monitor grid frequency and respond automatically at their preset trigger. Fingrid’s hourly purchase data show what was reserved, but not whether it responded or by how much. (Fingrid reserve markets; FFR market information; procured-volume data)

The order of events matters. Providers submit bids the previous evening, and Fingrid confirms the next day’s hourly purchases that evening. The kinetic-energy readings in this analysis come from the following delivery hours. A low reading at, for example, 14:15 on Wednesday could coincide with capacity Fingrid had already booked on Tuesday evening. It could not have caused that earlier booking. (Fingrid reserve markets; FFR market information)

Finland’s FFR bookings during low-inertia hours

We lined up Fingrid’s Nordic inertia readings with its Finnish FFR purchases hour by hour. An hour counted only if it had all four inertia readings and a recorded FFR value, including zero. The inertia figure covers the whole Nordic grid; the purchase figure covers Finland alone. (Fingrid data on kinetic energy; FFR purchases)

Both years cover the same dates, but some hours are missing a reading, especially in 2025. We left those hours out, which is why the table has different totals for the two years.

The gap between the years opened in late spring and persisted through summer. September ends on the 26th in both years. The comparison does not identify a cause: these two series cannot tell us which generation technologies changed the Nordic grid’s inertia or support blaming batteries, wind or any other single factor.

For 283 of the 2026 hours with low Nordic readings, Fingrid’s purchase record shows no FFR booking. That was a slightly larger share than in 2025, despite far more low readings in 2026.

The overlap is informative, but it is not a scorecard for the operator. Fingrid bases its FFR need on forecasts of inertia and the size of the largest failure it prepares for. The purchase is made ahead of time, while our comparison uses the later, measured inertia. A coincident low reading alone should not be treated as the trigger for a purchase. (Fingrid FFR product information)

The gap survives a tougher test

The low readings were not confined to isolated quarter-hours. In 940 hours of 2026, all four readings fell below 150 GWs; the same test found 224 hours over the matching dates in 2025. Fingrid had booked FFR for most of those hours in both years.

The 150 GWs mark is useful for a consistent comparison, but a different line would produce different counts. Fingrid’s typical range of 120–280 GWs does not make either endpoint a rule for buying FFR. The available records also do not show why Nordic inertia changed between these years. Explaining that would require generation and system-operation evidence beyond these two series.

Why battery storage matters

As wind and solar displace conventional generation, fewer generator rotors may be online in some hours. Lower inertia lets frequency fall faster after a sudden power loss, increasing the need for FFR. EV charging adds demand; it does not itself lower inertia. Batteries can provide FFR if they pass the grid operator’s tests. (Nordic grid operators on wind, solar and FFR; Fingrid’s FFR requirements)

Sustainable Digital Assets Inc., issuer of SDA Token (SDA), lists battery energy storage systems (BESS) among its proposed projects. If built, qualified and selected in Fingrid’s hourly market, they would supply FFR capacity during low-inertia hours. (SDA’s proposed projects; Fingrid’s FFR market)

How we checked the data: We matched Fingrid’s Nordic kinetic-energy readings with its FFR purchases for the same hour. We counted only hours with every reading present. Some records from 2025 are stamped a minute early and overlap the regular readings, so we left them out of the main count. Using them to fill gaps barely changes the result; we have not found an explanation for the shift. The data explainer defines the table, and the Fingrid datasets named below let anyone repeat the count. Please credit Fingrid if you reuse the data or charts.