Forecasting District Heating Consumption based on Customer Measurements

Kaisa Kontu, Tingting Fang, and Risto Lahdelma


District heating, forecasting model, production management, customer measurements


The target of this paper was to develop a forecasting model for district heating (DH) consumption using hourly heat consumption data from individual customers. The starting point of this research was that more accurate heat consumption data is now available directly from the customers and almost in real time. Earlier forecasting models have been based on overall production data for entire heating networks or sections. The forecasting model was constructed based on linear regression where outdoor temperature predicts heat consumption. Accuracy of the model was improved by adding the weekly rhythm of heat consumption to the model. The forecasting models were developed for nine different customers, four pairs of similar customers, and all customers summed together. The results show that the accuracy of the forecasting models varies when considering different customers and increases when the social component is also taken into account. The smallest relative errors are reached in the model where all customers are summed together. The temporal differentiation of the heat consumption of single customers implies that the overall consumption of a large set of customers taken in aggregate is rather stable.

Important Links:

Go Back