What is a "normal" frequency range?

🙂 49.900 - 50.100 Hz- Business as usual

😕 49.750 - 49.900 / 50.100 - 50.250 Hz– System is a bit out of balance

😣 49.500-49.750 / 50.250 - 50.500 Hz– System is seriously out of balance

About Baltic Grid Desynchronisation

On Saturday 8th of Feb, 2025, the electricity system of the Baltic States will disconnect from the Russian power grid and on Sunday it will synchronise with the grid of mainland Europe.

Timeline of Events

• On Saturday 8th of Feb, 2025, some time after 07:00 EET

  Lithuania will cut the connection with Belarus and RussiaLatvia will cut the connection with Russia

• On Saturday 8th of Feb, 2025, some time after 08:20 EET

  Estonia will cut the connection with Russia

• On Saturday 8th of Feb, 2025, by 09:00 EET

  All three countries will have cut the connections to Russia and Belarus

• Island mode for 33 hours

  Next, there will be a period of 33 hours where the Baltic power grid will operate in an 'island mode', balancing the grid on its own.

• On Sunday 9th of Feb, 2025, by 18:00 EET

  Lithuania will synchronise the Baltic system with the grid of mainland Europe (via Poland).

What can we expect?

Unless something goes awfully wrong, the regular electricity users will not notice a thing.

However, as electricity system geeks, we hope to see some periods of abnormal grid frequency during this process. Also, Elering's CEO Kalle Kilk mentioned on the "Esimene stuudio" talk-show that the Transmission System Operators (the parties responsible for maintaining a stable grid) are planning to run some tests driving the frequency very high or very low -- to test if emergency reserves activate as they are supposed to. Such tests are difficult to carry out when the Baltic grid is synchronised with a larger grid, so the "Island mode" presents a perfect opportunity for this.

We set up this web application to observe the grid frequency in near real-time with per-second sampling.

For more information:

About this page

This page was put together by a group of volunteers from Sympower so we could observe the Baltic Grid Desynchronisation Event by following the grid frequency near real-time and with per-second granularity. The original idea was to build it for ourselves, but then we decided to share it with the rest of the world as it may also interest others.

We do not have the frequency data readily available as Sympower is not offering any services in the Baltic markets (because the market so far has been too small and the grid has not needed many reserves). However, we have development offices in Tartu and Tallinn, so we have knowledgeable people with appropriate hardware and skills available.

Tech Stack

Development timeline, from idea to finish

• Thu (Jan 30th, 2025)

  Steve (from Tartu) asks in the Estonian office Slack channel if we have any frequency meters in Sympower Tartu office to see what will happen during the desynchronisation weekend.Then follows a bit of a technical discussion with Erkki (from Tallinn) and Neeme (from Tartu) about which power/frequency meters we have available and which are suitable for the purpose. The discussion dies off as there is no easy solution: either some code has to be written or a different model of meter is needed.

• Mon (Feb 3rd, 2025)

  As procuring a new meter will take too much time, Neeme decides to spend an hour hacking together the necessary code - to see if we can, in practice, read the frequency with the available meter.The experiment is successful and as we have access to the raw data, we can think what we can do with this data.The experiment is successful and as we have access to the raw data, we can think what we can do with this data.After some discussion among Neeme and Tõnis (from Tartu), it is decided to build a small public web application (using Sympower tech stack, AWS, Kotlin, Typescript, React) to give access to the frequency data.After informing the rest of the Sympower Team about the Baltic Desynchronisation Event, even more volunteers (Frank from Spain and Rik from The Netherlands) enthusiastically offer to help make the website happen.Tõnis proposes a technical design for the backend.We have a quick video call (Neeme, Tõnis, Rik) to discuss the status and plans.We settle on a solution for the backend: store data in Timestream, cache it in memory and serve it via a JSON-over-HTTP API.Neeme deploys the code to publish the frequency data collected from a wall socket in the Tartu office via MQTT, visualised in internal Grafana.Frank creates a source code repository for the front-end and pushes the first version of the front-end application.Tõnis creates a source code repository for the back-end and pushes the first version of the back-end application together with the API specification. For now, all data is stored in memory.There is discussion among Jaak (Estonian living in Sweden), Tõnis and Rik whether we should collect more telemetry (not just frequency). For example, we could also collect voltage.

• Tue (Feb 4th, 2025)

  Discussion continues about the extra telemetry and the physics of the alternating-current electricity system.Tõnis deploys the back-end microservice with mocked input data so front-end developers can integrate with the API.Neeme investigates what telemetry does the power meter support and tests which are available in practice. Due to the way the power meter is connected (wall socket) we have access only to voltage of a single phase and frequency. We decide that voltage of a single phase is not particularly interesting so we focus on frequency data. There is also some discussion about the accuracy of the frequency measurement but the power meter manual does not say anything about it, so it remains unclear. Probably not too accurate as the main function of this meter is power measurement, not frequency measurement.Frank and Rik get the data flowing between back-end and front-end, visualised on a graph.Tõnis implements data persistence (to Timestream database) in the back-end.

• Wed (Feb 5th, 2025)

  Frank and Rik present a new version of the front-end, with improved visualisation and data fetching logic.Discussion among Neeme, Tõnis and Jaak about the web application URL. Should we go with https://baltic-freq-event.sympower.net, https://baltic-frequency.sympower.net, https://baltic-grid.sympower.net or https://tartu-frequency.sympower.net. We let Katarzyna (Polish, living in Netherlands) from the marketing team decide -- she chooses https://baltic-grid.sympower.net/.Discussion among Tõnis, Neeme and Rik about API redesign and how to aggregate the old data while also serving non-aggregated fresh data.Tõnis deploys a new version of back-end that has real frequency data instead of random noise.Tõnis redesigns the API aggregation logic.

• Thu (Feb 6th, 2025)

  Jaak deploys the front-end, the web application is now publicly available at https://baltic-grid.sympower.net/!Frank continues improving the front-end, replacing the gauge visualisation with a better one.

• Fri (Feb 7th, 2025)

  Fine-tuning continues.Neeme prepares content for 'about' pages. Kat (Katarzyna) polishes texts.Public announcement about the application.

About Sympower

Sympower was founded in Estonia in the summer of 2015 by Simon Bushell (British/Dutch) and Georg Rute (Estonian). They started in Estonia because they could set up the first "office" in Georg's place in Tallinn. The original idea was to provide (balancing) reserve services to Transmission System Operators using IoT technology to steer a portfolio of electricity-consuming assets.

Given that the company was born in Estonia, the first market Sympower intended to enter was the Estonian market -- but it turned out that the Estonian reserve markets were almost nonexistent -- the market was too small, and the prices were too low to build a viable business.

Sympower chose to enter the Finnish market instead and has since expanded to Sweden, Norway, Netherlands, and Greece.

Sympower’s headquarters is located in Amsterdam, with large engineering offices in Tartu and Tallinn.

In case you want to learn more: