District heating utilities still rely on intuition

Originally published by energate 

This interview was originally published in German by energate, an independent information and communications service provider for energy topics with the largest dedicated energy editorial team in the German-speaking region. This English translation has been prepared by Gradyent for the convenience of international readers.  

Berlin (energate) – As municipal heat planning advances across Europe, district heating systems are set to expand significantly. Yet many utilities still operate without live operational data and maintain unnecessarily high supply temperatures. Alexander Schultz, Business Development Lead at Gradyent, explains how real-time Digital Twins help operators gain system-wide visibility, unlock flexibility, and optimize district heating systems end-to-end.

energate: Why are Digital Twins becoming increasingly widespread in district heating?

Schultz: District heating is becoming increasingly strategic for municipal utilities. Heat transition plans are moving from planning into execution. At the same time, district heating systems are rapidly decarbonizing, electrifying, and becoming more interconnected. More sustainable and often decentralized energy sources are being integrated, while networks continue to expand and densify.

As a result, operational complexity is increasing across the entire energy system, particularly from a hydraulic perspective. Utilities are no longer only producing electricity; through sector coupling, they increasingly depend on electricity to operate assets such as large-scale heat pumps and power-to-heat installations. The volatility of electricity markets has therefore become a key factor influencing heat production costs.

Managing this new reality requires more than traditional rule-based optimization tools. Today's district heating systems must balance volatile energy markets, decentralized generation assets, hydraulic constraints, and security of supply simultaneously. Without combining physics-based modelling, calibrated with machine learning, and running real-time operational data, these challenges cannot be addressed effectively.

Historically, production dispatch optimization often ignored what was happening in the distribution network itself. Today, however, an end-to-end view of the entire district heating system is becoming essential.

energate: And that approach is no longer sustainable?

Schultz: Perhaps in a network with high operating temperatures and a single production site. But the moment multiple heat sources are integrated, operators need to consider a growing number of variables when making operational decisions. These include electricity market prices, weather forecasts, fuel costs, asset availability, and historical operating patterns.

Utilities that are further along in their heat transition journey increasingly realize that traditional ways of operating are no longer sufficient. Many have invested heavily in metering infrastructure over recent years, yet much of the available operational data remains underutilized.

We see these challenges across Europe and across all system sizes. Gradyent is currently involved in more than 50 district heating systems, ranging from approximately 50 GWh annually to our latest project in Warsaw, which supplies around 11 TWh of heat every year.

energate: How did Gradyent secure such a large contract in Warsaw?

Schultz: Warsaw's district heating system is operated by the Veolia Group. Over the past three years, we have worked closely with Veolia on several projects, including real-time optimization in Łódź and Poznań. Through these projects, we were able to demonstrate that the value created by Digital Twins is both measurable and operationally relevant. That trust ultimately led to this new large-scale deployment in Warsaw.

The Poznań team wanted a tool that would view the system as a whole, not separate, unconnected buckets. Their goal was to save operators time by eliminating the need to spend hours coordinating across teams using spreadsheets.

Today, Gradyent's real-time Digital Twin automatically considers all these factors. The platform continuously forecasts and optimizes the district heating system end-to-end, combining weather forecasts, electricity market prices, demand forecasts, asset performance, network constraints, and learned operational patterns into one physics-based source of truth.

This enables operators to move from reactive operations and manual coordination to proactive, system-wide optimization with visibility across the entire value chain.

energate: Is Germany further ahead in this regard, or are these manual coordination processes still common? 

Schultz: To be honest, they are still quite common. 

Many operations teams have an enormous amount of experience and deep knowledge of their systems. What they often lack are the tools needed to bring together all relevant operational data for decision-making. As a result, operators continue to rely heavily on their experience and intuition and often work with unnecessarily large safety margins.  

Rather than replacing operator expertise, a Digital Twin complements it. It provides operators with system-wide control and allows them to understand exactly what is happening throughout the full system. Decisions can then be based on operational reality instead of assumptions. 

This allows organizations to move from reactive firefighting toward proactive, data-driven operation. Experienced operators remain essential because they are best positioned to validate the model and ensure it accurately reflects the behavior of the real system.  

energate: You mentioned intuition and hydraulics. What are the new challenges associated with lower temperatures and new generation assets? 

Schultz: Historically, district heating networks were designed around a large central generation plant supplying heat through large transmission pipes to different parts of the city and then through smaller distribution systems to end customers. 

When two or three decentralized energy sources are added—for example industrial waste heat or river-source heat supplied through large-scale heat pumps—the existing hydraulic design may no longer be suitable. 

Many of these new heat sources require lower temperature levels. Heat pumps, for example, achieve significantly higher efficiency and a better coefficient of performance (COP) when network temperatures are reduced. 

energate: What temperature levels are we talking about—80°C, 90°C, or lower? 

Schultz: In Flensburg, for example, we are supporting a project that uses dynamic temperature control to reduce the percentage of operating hours above 95°C from ten percent to three percent. Even this change results in a substantial efficiency improvement. 

Incidentally, 95°C corresponds to the requirements defined under Germany’s Federal Funding for Efficient Heating Networks (BEW) scheme. 

We also work with Danish customers operating at supply temperatures of around 75°C, which is remarkable. In such networks, even a reduction of a single degree can unlock significant value. 

One important consideration is that the same amount of heat must still reach customers. As temperatures decrease, flow rates increase, it is critical to understand how much hydraulic capacity the system can actually accommodate. 

An alternative approach—which is not advisable but still sometimes practiced—is simply to continue lowering temperatures until customers begin to complain. 

energate: What do you recommend instead? 

Schultz: We model the actual heat demand at every individual endpoint within the network and calculate heat transport times throughout the system. This enables us to define a dynamic supply temperature setpoint for each generation site. 

Instead of relying on a static outdoor-temperature-based heating curve, the system generates target setpoints every 15 minutes, tailored to the current load conditions and hydraulic requirements. 

There are also situations in which our Digital Twin recommends higher heat output than traditional operating methods would have prescribed. 

On average, this approach allows supply temperatures to be reduced by five to ten degrees Celsius, resulting in a similar reduction in distribution heat losses. 

energate: How does your implementation process work in practice, for example in Warsaw? 

Schultz: Warsaw is a special case because of the district heating system’s scale and complexity. Generally, we begin by obtaining the customer’s GIS data, which provides a geographical inventory of all pipes, their specifications, and elevation profiles. We then model the physical characteristics of the entire network pipe by pipe, using historical data on supply temperatures, pressure levels, differential pressures, flow rates, and other operational parameters. 

Typically, we use one year of historical data. 

Next, we work closely with the operations team. For example, our model may calculate a certain pressure level while a sensor reports a different value. We then investigate the discrepancy. 

One common example is equipment that still appears in network documentation but has been out of service for years. 

Following these updates, the model is recalibrated repeatedly until the calculated values align with historical operating data. Once this level of accuracy has been achieved, we establish live interfaces and transition into operational deployment. 

energate: I imagine inaccurate network documentation is quite common? 

Schultz: Actually, it is almost universal. 

Many utilities also store relevant information across multiple isolated data silos. Creating transparency and establishing a complete overview is an important part of our project work. 

At the outset, utilities often lack a complete picture of their network. By the end of the process, they have one—and that visibility proves extremely valuable. 

energate: How would you assess the current state of technical infrastructure, such as smart substations and metering systems? 

Schultz: Many utilities are actively digitizing their operations, including through LoRaWAN-based communication networks. 

However, communication quality can sometimes be problematic. For example, wireless connections in basement substations may be so weak that data cannot be transmitted for several days. 

The more data available, the more granular our optimization can become. Nevertheless, most German utilities already possess significantly more data than is required to start live optimization today.  

energate: How long does it take to build a complex model such as Warsaw’s, and how long for simpler projects? 

Schultz: Typically, implementation takes between five and twelve months from project kickoff to go-live. 

The timeframe largely depends on the application. Temperature optimization projects can be completed considerably faster than production scheduling and dispatch optimization projects, where every generation asset must be modelled in detail and integrated with intraday and day-ahead electricity trading systems. 

energate: Which reference customers does Gradyent have in Germany? 

Schultz: We have been working with Stadtwerke Flensburg for several years. They are one of Germany’s frontrunners, with a district heating connection rate exceeding 90%, and they will soon commission a large-scale heat pump. 

Our collaboration with Iqony Fernwärme began with heat transition planning and the analysis of smart meter data. We are now operating live optimization in the Essen-Süd network and have expanded into additional networks. 

Other reference customers we are permitted to mention include FUG, Stadtwerke Böblingen, SaarLorLux, and Stadtwerke Kamp-Lintfort. 

However, our journey started in the Netherlands, where seven of the country’s ten largest district heating networks are currently supported by our solutions.