Skip to content

Decision story

When the Spinning Stops

How much fast reserve should the grid hold when there is less spinning mass to hold it up?

Rates lift payments. Payments eat buffers. Watch where that pressure concentrates.

Scroll · into the decision

With reduced motion, camera moves cut instead of gliding, the illustration shows each state without animating, and the beat narration is announced in a live region.

The grid does not need to be kept spinning; it needs to be caught faster

This story ships as a directed film. Open /stories/how-much-fast-reserve/film.

A real trip on 1 June 2026, the grid as a machine of spinning wheels, and a year of hours by spinning mass. The method, data and schema are at /stories/how-much-fast-reserve/method.

Methodology

Directed film over one frozen evidence pack. Fingrid's 10 Hz frequency archive for August 2025 – July 2026 is median-filtered over 0.9 s and scanned for falls of more than 100 mHz within 5 s; each fall is classified by its shape, and 19 of 33 are trips. Each event is paired on UTC with that hour's kinetic energy and fast-reserve purchase. Hours are clock-hour means of Fingrid's real-time kinetic-energy estimate. Published thresholds are quoted from the Nordic operators. A one-bus frequency model fitted to the operators' design points draws the design-case curves and drives the illustration's dial; it is never quoted as a figure.

Limitations

Which unit tripped is not in the open data. Frequency is measured at Finnish substations and local oscillation is not separated. Kinetic energy is itself an operator estimate. Finland's fast reserve is one share of a Nordic need. The model is one-bus and draws shapes only. No event in the window came near 49.0 Hz; the story is about margin, not a near-miss. Not affiliated with, and not endorsed by, Fingrid or any Nordic operator.

Data references

  • [observed] 1 June 2026, 17:44 Finnish time: the frequency fell to 49.651 Hz in 5.5 s — Fingrid 10 Hz frequency, median-filtered over 0.9 s. The hour held 169.6 GWs of kinetic energy and no fast reserve was bought.
  • [calculated] 19 trips in a year; none fell below 49.6 Hz — 33 sustained falls, classified by shape; 19 are lost generation.
  • [modelled] The largest trip at 194 and 100 GWs, with and without 300 MW of fast reserve — One-bus model fitted to the operators' design points; drawn as a shape, never quoted, and about 1.7× deeper than the real trips as a design case.
  • [calculated] 873 hours below 150 GWs in the year; 584 of them in May–July 2026 — Hourly means of Fingrid's real-time estimate of Nordic kinetic energy. Our yearly counts are within 6% of the operators' published ones for 2020–2024.
  • [calculated] Fast reserve bought in 91% of hours at 120–140 GWs, and in none above 200 GWs — Fingrid's procured volume per hour against that hour's kinetic energy.
  • [published] Floor 49.0 Hz; largest trip 1,450 MW; about 300 MW of fast reserve at 100 GWs; about 20 GWs of mass per 0.1 Hz — Nordic TSOs 2025 and FFR design 2024; Ørum et al. 2017 via Energinet.
  • [illustrative] Wheels on a shaft, a plant that trips, a dial against the floor — The dial follows the modelled trip to scale; the wheels' slow-down is exaggerated.

Sources