A data-driven method for heat loss estimation from district heating service pipes using heat meter- and GIS data

District heating (DH) pipes are often older than 20 years and of unknown condition, imposing a strong demand for detailed knowledge about their current state to help cost-effective ongoing renovation and reduce water and energy losses. Here, we develop a scalable method that exploits hourly resolved remotely read heat meter data, pipe coordinates, and soil temperature measurements to continuously estimate the heat loss of service pipes. The method does not require the measured or simulated flows in the full system, making it possible to omit the usual step of establishing a complete thermohydraulic network model which is currently hindering implementation on larger scale. By exploiting open-source soil temperature data, the estimated heat loss is normalized to output the transmission coefficient, U, in W/(m∙K), allowing for direct intercomparison between pipes of different length and assessment of timely evolution across seasons. Estimated heat losses are validated by complementary techniques, using simulation of data sets and highly time-resolved data from a test area with additional measurement equipment installed. Finally, a full-scale application is demonstrated over a year on 1089 service pipes, including detection of a spontaneously emerged leakage, revealed through the model sensitivity towards the consequential changes in heat loss coefficient, U.