Data analysis and modeling of users consumption profiles in District Heating Systems

District heating represents an optimal solution to increase the efficiency of the heating sector, which accounts for a large share of the total primary energy consumption in Europe. However, the heat load demand that is generated by served buildings has a strong influence on the efficient operation of the generation plants and the network. When Combined Heat and Power units are used, the heat load shape has a close connection also with the amount of electricity that can be produced, thus having a strong impact on the system’s economic sustainability. If renewable energy sources are integrated within the network, the variability of heat load can compromise their actual contribution to the overall production.Finally, the characterization of the heat load of many buildings through physical models is prevented by both the difficulty to gather the large quantity of necessary input data and the required computational time. This research work focuses on the analysis of the heat load consumption profiles of a large stock of buildings connected to a real district heating in Italy. A numerical model for predicting the heat load profile by using the smallest possible set of input variables is proposed. The model aims at finding numerical relationship among variables that are quick and simple to collect and significant for the definition of the heat load profile. Data gathered from 85 buildings are used as data-base for developing and testing the model. Two distinct preliminary procedures have independently assessed that time-of day and outdoor temperatures are by far the two most important variables in defining the heat load profile of each studied building. The role of time-of-day depends primarily on the control unit set-ups of each building; since many buildings must be analyzed and no information was available about these schedules, a profiling algorithm has been built and tuned to automatically identify the different operation modes set by each control unit. In the successive phase this identification is used as a basis for applying to each clusterization a linear regression model between the outdoor temperature and the heat load. The capability to distinguish among each different profile state allows the linear regression model to be much more effective when applied to each clusterization w.r.t. its application to the whole set of data. The outdoor temperature variable is also corrected by an effective time constant that is calculated to consider the effect of thermal inertia on the evolution of the heat load profile. The model has been tested over a one year period at both single building and simulated network levels. The final results confirm that the proposed model is capable of forecasting with sufficient accuracy the heat load profile of the simulated network of buildings both in terms of detailed heat load shape and yearly energy consumption. The last chapter reports three possible applications of the developed model. In the first one the impact of the implementation of an advanced control function within the district heating substations is simulated. The model proves to be reliable and suggests that the implemented function could guarantee significant energy savings by avoiding the supply of heat to buildings when excessive outdoor temperatures occurr. An optimization procedure is developed as a second application that is aimed at reducing the maximum peak load of the whole network by shifting the switchon/ switch-off schedules of the simulated buildings. This application proves that such advanced control could help reduce significantly the maximum power peaks. Finally, the proposed data analysis and modeling capability is used to simulate the effect of adoption of heat pumps as an alternative heat generation technology for substituting traditional natural gas boilers. It is shown that the heat pump solution is characterized by a primary energy consumption that is considerably lower than that of the natural gas boiler considered for comparison. The final results show that (1) the demand-side has a strong impact on the efficiency and sustainability rate of district heating systems; (2) an accurate numerical model can be used to describe and predict the heat load of a large building stock at detailed time-step, provided that sufficient data is collected for training the model; (3) the model can be successfully implemented in applications regarding demandside management, advanced control simulation and local-scale energy planning opportunities.