District heating is a common means of providing residential heating in urban areas in Northern, Eastern, and Central Europe. Shifting external parameters, such as fuel price fluctuations and policy implementations encouraging decarbonization, can lead to significant cost increases that take years to resolve, as production facilities are built to operate for multiple decades. In this article, a mixed-integer linear optimization model minimizing total network costs is developed to assess how flexible networks are to such changes and how ideal investment pathways depend on existing infrastructure. Utilizing temporal and technology aggregation methods, the operation of a given network is simulated over a multi-decade timeframe, with the option to invest in new production facilities every five years as existing facilities reach end-of-life. The model is applied to two Nordic district heating networks located in Turku, Finland, and Gothenburg, Sweden. Progressive expansion of power-to-heat utilization in the form of heat pumps and electric boilers is shown to be economically advantageous in both networks as existing facilities are decommissioned. The effect of path dependence becomes apparent under varying simulated scenarios, such as in Turku, where significant investments are postponed until the decommissioning of a single large combined heat and power plant. Air-source heat pumps are found to be uncompetitive in the colder Nordic climate compared to heat pumps utilizing industrial waste heat at a constant temperature, but may be incorporated in cases where access to such waste heat sources is constrained. • An optimization model is built to determine long-term district heating investments • Technology and temporal aggregation enable efficient multi-decade optimization • Two Nordic district heating networks are analysed as comparative case studies • Optimal investment pathways are strongly influenced by the existing infrastructure • Gradual electrification via power-to-heat expansion is cost-optimal in most cases
Högnabba et al. (Sun,) studied this question.