In which hours is the grid at capacity, when do batteries need to feed in, and how do we prevent renewable energy from being curtailed at midday, with the result that we have to fire up expensive gas plants in the evening?
That is a complex question, and it is exactly the right one if you want to understand where the energy transition faces its next technical test. The energy transition has shifted its most expensive bottleneck from copper to information. In the transmission grid, new lines solve the problem. In the distribution grid, nobody solves it without knowing what is happening in their own grid right now.
The regulatory trend provides a tailwind. With Redispatch 2.0, grid operations were switched to a schedule-based procedure. Acute congestion or surplus capacity is no longer answered by spontaneously ramping down wind and PV farms, but through a forecast-based congestion avoidance process. The logic behind it is compellingly simple: better forecasting replaces expensive reaction.
Redispatch 3.0 follows this path to its logical conclusion. In future, even small and very small installations are to be included in forecasting and grid optimisation processes — installations that today simply fly under the radar. This requires a larger and better integrated data basis as the foundation for the AI algorithms meant to orchestrate it all. Data integration and AI are not buzzwords here but the explicit target architecture that the Federal Ministry for Economic Affairs, Fraunhofer IEE and the transmission system operators agreed on long ago.
An honest interim assessment
This is a point where it pays to be precise, not least because a lot of nonsense is currently being said about redispatch. At transmission level, the job is largely done. The major north-south corridors will come online in the coming years, and costs are falling anyway: for 2025, the provisional costs of congestion management stood at around 2.7 billion euros, and the transmission system operators have revised their forecasts for subsequent years downwards by billions. Grid expansion is working. That is the good news, and it should not be talked down. Our conviction is that the stability of the power grid will not suffer as a result, provided it is done well. The industry association BDEW calls digitalisation in the distribution grid an “imperative necessity” and stresses that in practice Redispatch 2.0 only works with close real-time data exchange across voltage levels.
Integrating renewables therefore confronts us with challenges and, through Redispatch 3.0, opens up opportunities at the same time. Because the more interesting news is this: the relevant and growing bottleneck is moving into the distribution grid. The share of distribution grids in redispatch volumes for renewables rose within a single year from 20 per cent (2023) to 26 per cent (2024). This is no longer about roughly a hundred large power plants controlled centrally, but about hundreds of thousands of distributed installations, storage units and controllable loads that have to be forecast, integrated and coordinated.
That also shifts the actual constraint. In the transmission grid it was and remains physical: as long as the line is missing, no forecast, however good, will help. In the distribution grid the binding constraint is a different one, namely data integration. Anyone who fails to bring together the asset, status and schedule data of their grid level cannot participate in Redispatch 3.0 at all, and leaves existing grid capacity unused.
The thesis is therefore not “digitalisation instead of grid expansion”. It is this: data integration is the necessary condition for the distribution grid to be able to play its part in congestion management at all.
What does digitalisation have to deliver for distribution grids?
In line with the findings of the Fraunhofer research project on Redispatch 3.0, a modern and networked energy system has to be about setting targeted incentives for grid-friendly behaviour, and thereby above all encouraging the contributions of a growing number of distributed installations to the system, particularly in providing ancillary services. But for that to happen — and here the circle closes back to the opening question — grid utilisation has to become more predictable.
So how can a real-time-capable and resilient digitalisation concept be advanced as the precondition for grid operations that are not merely negative and reactive but, at best, positively grid-supportive?
The answer does not begin with yet another tool, but with the data foundation beneath it. The question of when the distribution grid is at capacity can today rarely be answered in real time, because the necessary information is scattered across dozens of systems: asset master data here, measured values there, weather and generation forecasts, schedules and market data each in their own silo. As long as this data is not brought together, every statement about utilisation remains an estimate with a time lag.
That consolidation is the real engineering achievement, and it is more demanding than the forecasting model that ultimately runs on top of it. Vertical data integration means moving data across all voltage levels and across the boundaries of operations, markets and technology into a consistent, queryable and real-time-capable layer.
This is precisely where control-f comes in: building an integrated, streaming-capable data platform that makes the distribution grid legible in the first place. Not as another isolated solution, but as the foundation on which forecasting, congestion management and the connection to industry-wide platforms such as DA/RE or Connect+ can be built reliably.
About control-f. control-f GmbH is a values-driven AI company based in Konstanz. Since 2022, the data boutique has been building big data platforms for industrial telemetry data and helping companies in German-speaking Europe make complex data landscapes usable. Its clients include large corporations and mid-sized companies from plant engineering, automotive and the energy industry. Managing directors Simon Deussen (machine learning engineer and founder) and Daniel Tremer (formerly Specialist Data Science & AI Projects at Porsche AG) focus on building stable data architectures as the foundation for analytics, software solutions and AI applications such as predictive maintenance.
Kategorie / Category: Blogposts