Return Temperature Prediction in District Heating Networks Using Primary Side Measurement Data

District heating networks are essential for decarbonizing the heating sector, yet their efficient operation is often hindered by high return temperatures, which indicate reduced efficiency or substation faults. Modeling return temperature is therefore critical but challenging due to complex building-side dynamics, limited data availability, and the high effort required for detailed physical simulations. This study proposes a measurement-based surrogate modeling approach for predicting primary side return temperature at substations in district heating networks. Our method is based on a simplified thermodynamic formulation of return temperature as a function of supply temperature, heat flow, and a scaling coefficient. A polynomial regression model is used to estimate this coefficient from primary side measurement data that are typically available to network operators, without relying on secondary side or building-specific information. Model performance is evaluated using residual analysis as well as statistical and goodness-of-fit metrics. The results demonstrate that our simple approach produces unbiased and accurate predictions of return temperature indicating that patterns arising from the system’s behavior are captured despite the model’s simplicity. The proposed approach provides a transferable and data-efficient alternative for scalable district heating network analysis under realistic data availability constraints. Our findings complement other conceptual contributions, highlighting the relevance of relatively simple statistical models to generate testable hypotheses for high-effort physical simulations.