Heat networks will grow substantially over the next two decades. In the KOMpare project, the Öko-Institut and Fraunhofer ISE evaluated around 380 municipal heat plans from about 940 municipalities. According to those plans, the share of heat delivered through networks is to rise from around 15 per cent today to 38 per cent in 2045. "Heat pumps and heat networks form the backbone of the municipal heat transition," says Marc Stobbe, energy expert at the Öko-Institut.
For operators, that means more connections, longer pipelines and more generators that have to work together. Alongside the gas boiler, solar thermal, biomass, large heat pumps and storage are increasingly part of the mix. A network like this can hardly be operated on rules of thumb alone, because every generator needs different conditions and the most economical operating mode changes from hour to hour.
The data is there, but not together
There is no shortage of measured values in any heat network. The control systems of the heating centres and pump stations record temperatures, pressures and flow rates, the generation plants have their own controllers, and the pipeline network is recorded in a geographic information system. On top of that come the customers' heat meters, which will be switched to remote reading by the end of 2026 and will then, for the first time, deliver regular readings from every heat transfer station.
As a rule, these sources sit in separate systems, with different service providers and at different time intervals. As long as that remains the case, generation and consumption cannot be compared directly. Network losses then only appear in the annual balance sheet, and a faulty heat transfer station only comes to light when a customer complains.
Data quality is a well-known problem in the industry, and not only for heat networks. In a 2025 study by adesso and Natuvion, one third of the German energy utilities surveyed named data quality problems as the biggest surprise in their IT projects. No other obstacle was mentioned more often.
A shared data foundation
Our approach therefore does not start with another application, but with consolidation. We connect heating centres, pump stations, storage and meter data via open industrial standards and store all values in a shared, documented data set that belongs to the operator. Generation and consumption then sit in the same time grid, and all further analyses draw on the same figures.
Predictive maintenance for pumps, boilers and heat transfer stations
On that basis, models learn how the plants behave in normal operation and report deviations at an early stage. For pumps and boilers, that means, for example, changed runtimes, temperatures or power consumption that announce themselves over weeks. For heat exchangers, fouling shows up in a growing temperature difference.
Heat transfer stations are particularly revealing. A station that is set incorrectly or is faulty returns water that is too warm to the return line, raising the temperature of an entire branch. This increases heat losses and reduces the efficiency of solar thermal systems and heat pumps. The meter data shows which stations are conspicuous, so technicians can drive there specifically. Many interventions can thus be scheduled for the summer months before they turn into a breakdown during the heating season.
The digital twin of the network
The next step is a digital twin, a computational image of the network that is continuously reconciled with the measured values. It combines the pipeline data from the geographic information system with a hydraulic and thermal model and the current values from the control system and meters.
If calculated and measured values diverge, that points to a problem in the network, such as a leak, a sticking valve or a bottleneck. The twin also makes it possible to model changes before they are implemented in the real network, for example a lower supply temperature, the connection of a new housing development or an additional storage tank. Established simulation tools can be connected for the hydraulic calculation; the data foundation stays the same.
Automating the operating mode
A reliable data foundation also makes it possible to automate operation itself to a greater extent. A load forecast based on weather data, day of the week and consumption history shows how much heat will be needed in the coming hours. From that follow when the storage should be charged, which generator should be switched on next and how far the supply temperature can be lowered without customers receiving too little heat.
A step-by-step introduction makes sense. Initially the proposals run alongside existing operations as recommendations, so the team can compare them with its own experience. Only once the results are convincing are individual setpoints passed automatically to the control system. Responsibility for the network remains with the operator.
This also creates the basis for the reports that heat network operators will in future have to deliver regularly, for example on the share of renewable heat. If generation and consumption are recorded together anyway, that share can be calculated continuously instead of being compiled once a year.
Getting started
A single network is enough to begin with. In a four-week analysis we record which data is available at what quality, where the biggest gaps are and what closing them would achieve. On that basis, the operator decides whether and in what order the data foundation, early detection, digital twin and automation follow.
Kategorie / Category: Blogposts
