Dear Reader,

in issue 44 of our heatbeat Research Newsletter we present 2 examples for how considering the substation behavior can help to improve the network operation as a whole. While the first paper focuses on actively shaping the load profile of a single very large customer substation, the second paper has a focus on evaluating the measurement data of a large number of small substations. Together, both papers provide interesting insights into the holistic optimization of district heating network operation.

Network Peak Shaving with Storage and Forecasting at a Large Customer Substation

In addition to integrating more renewable heat sources, improving the efficiency of district heating networks in operation and avoiding the use of peak load boilers is another central challenge in the ongoing heat transition. In the paper "Peak shaving at system level with a large district heating substation using deep learning forecasting models" , Trabert et al. present an interesting use case for improving the network operation.

The paper is a case study based on an existing network which includes 2 heat supply plants, a combined heat and power plant with around 135 MW of installed thermal power and a peak load heating plant at a separate location with around 70 MW of installed thermal power. In this network, one industry customer accounts for around 7 % of the total heat load in the system. Thus, the load of this one customer has a significant effect on the system as a whole. To improve the system operation, the paper investigates how utilizing a tank storage and heat load forecasting at the customer site can help to shave off peaks in the network and thus avoid the use of the peak load boilers.

For the results, it's interesting to see the influence of different optimization intervals (up to 24 hours), the storage capacity (up to 7 hours of the customer's heat load), and the forecasting methodology. The authors show how including a weather forecast improves the peak shaving compared to only using preceding measurement data, and how much better a perfect weather forecast would be compared to an actual weather forecast with deviations to the actual weather. The authors are able to show that the economic optimum for the case study was a storage capacity to buffer between 1 and 2 hours of operation, and that a good forecasting method can support the peak shaving efforts of this large customer with beneficial effects on the entire network.

Using Smart Meter Data to Evaluate Substation Performance

While the above paper showed that a single large customer can be optimized to have a positive influence on the network operation, we know that faulty and inefficient customer substations can also have a negative effect on the network operation. And the increasing digitalization of substations and smart metering offers increasing opportunities to detect such faults and thus support improving the network operation. In this context, the paper "Contextual operational energy performance indexing of district heating consumers" by Søndergaard et al. describes an interesting use case of defining efficiency metrics for substations and using them to cluster and rank these substations by their efficiency.

As a foundation for this study, the authors utilize smart meter data of several hundred buildings including the heat consumption, supply and return temperatures, and the volume flow in hourly resolution for 3 years. As one metric, the authors define an "Energy Efficiency Parameter" with which they compare the heat consumption per building area and in relation to the outdoor air temperature for similar building groups (based on building type and year of construction). In addition, they define a second metric as the "Cooling Efficiency Parameter", which is calculated as the difference between the substation's supply and return temperatures, thus meaning how efficient the substations cools down the primary network medium when drawing heat from the network.

Based on these metrics, the paper defines 4 efficiency clusters for the buildings and shows examples of tracking the daily building performance, both in comparison to other buildings and to its own historical average. Furthermore, the authors show how tracking these metrics and comparing them against thresholds can help to detect faults in the substation behavior. While similar results are possible with slightly different choices of metrics, we think that this investigation is a good example for using smart meter data to evaluate the substation behavior, detect faults, and thus build a foundation for continuous optimization of network operations.

The next issue of our newsletter will be published on July 3, 2024.