Randvoorwaarden en omstandigheden voor de uitrol van thermische netwerken

As part of its climate targets, the European Union aims to reduce CO2 emissions, improve the efficiency of the use of energy and invest in technologies that use renewable energy sources. Originally, that implied electrification of the heating sector. In recent years, however, the district heating concept has gained attention to diversify the means to reach the climate targets and as such lower the overall cost to reach those targets. District heating networks distribute heated water from various sources through underground pipes to buildings or industries. The network is able to connect to a variety of heat sources that are in line with the European climate targets, including (but not limited to) heat pumps, heat from industrial processes that would otherwise be wasted and geothermal heat. The potential of district heating networks has been well established for a multitude of regions. However, district heating networks compete with individual heating options such as individual heat pumps at the demand side location. Due to numerous decisive factors, it is difficult to obtain general recommendations whether to invest in district heating or individual heating. For example, the presence of strategic sources can give district heating an edge over individual heating. These strategic heat sources have a low marginal cost because they originate from natural renewable energy sources or sources that would otherwise be wasted. As a counter example, district heating require substantial investments. The investment costs depend on the annual heat demand per length of the network, i.e. the linear heat demand density. For a low linear heat demand density, the investment costs can be very high, making individual heating more attractive. The relation between context specific parameters and the decision whether to invest in district heating or individual heating can be considered as the decision boundary. The aim of this thesis to learn more about the decision boundary by developing a methodology that is able to quantify the decision boundary. The methodology involves an energy system planning model to evaluate the decision between district heating and individual heating for a single context, as well as a parameter analysis that creates the decision boundary between district heating and individual heating by evaluating the energy system planning model for multiple contexts. To make the energy system planning model efficient and effective, the model design is specifically tailored to the methodology for finding the decision boundary. To obtain general recommendations for investments in district heating, the model makes rational decisions based on the economic optimum (also considering externalities such as a tax on CO2 emissions). To handle multiple decisive factors, the model reduces computational power requirements by making abstraction of the exact configuration of the district heating network. As there are various context specific parameters that play a role in the decision whether to invest in district heating or individual heating, it is impossible to evaluate the energy system planning model for every possible context. Therefore, the decision boundary between district heating and individual for various contexts is predicted from a smaller number of samples. Typically, as part of a scenario analysis, a few samples are designed from assumptions and insights to obtain qualitative results with very little calculations. Here, as part of a parameter analysis, the selection of a larger number of samples is optimised to iteratively improve the prediction of the decision boundary and as such obtain quantitative results. The prediction of the decision boundary is based on machine learning techniques and is visualised in a way that allows us to study the decision boundary more clearly. The methodology and the decision boundary are shown off in various theoretical and practical case studies. Theoretical case studies highlight the sensitivity and performance of the methodology. The sensitivity is the highest for the decision boundary, implying the need for the most samples and robust results near the decision boundary. The performance is limited by the curse of dimensionality (i.e. the required computational power increases strongly with every parameter), implying the need to reduce the number of parameters. One trick to reduce the number of parameters is to run the simulation twice; once with a low resolution and a high number of parameters to determine the most important parameters and once with a high resolution and a low number of parameters to actually obtain a prediction of the decision boundary. Practical case studies are oriented towards the interpretation of the decision boundary between district heating and individual heating. Various topics are touched upon including the use of strategic heat sources and sector coupling between the heating and cooling sector. These case studies show both qualitatively and quantitatively that the linear heat demand density and the availability of strategic heat sources play a significant role in the affordability of district heating networks.