Thermal request optimization in district heating networks using a clustering approach

In this paper, a method for the optimization of the thermal load of a district heating network acting on the building demand profiles is proposed. The aim of the optimization is to increase the system performance by minimizing the thermal peaks. The simplest way to implement such optimization consists in anticipating the time of the heating systems are switched on, so that the maximum requests do not occur at the same time. Proper constraints are posed in the optimization, so that the thermal comfort in the buildings is not affected. Various practical results can be obtained: (1) reduction of the primary energy consumption, (2) connection of additional buildings to the network despite the presence of hydraulic bottlenecks, and (3) shifting the thermal production of cogeneration plants in order to increase competitiveness on the electricity market. The optimization of the thermal request profile is here performed along with a cluster analysis, in order to catch buildings with similar characteristics. With this aim operating data of various thermal substations connected to a distribution network of a large district heating are analysed. Results show that the clustering approach allows achieving a large peak reduction. The application to six distribution networks indicates that an average reduction of the thermal peak load of about 14% can be obtained, while this reduction is about 8% when optimization is performed without clustering.