Market design for future district heating systems

District heating is envisioned to play an important role in future carbon-neutral energy systems, by facilitating the integration of renewable and excess heat, and providing flexibility to the electrical power system. With the integration of renewable and excess heat, there will be increased fluctuations in the available amount and the generation cost of heat. To provide heat reliably at a reasonable cost, flexibility on the demand side is needed. Attractive sources of demand-side flexibility exist both at the end-consumer and in thermal storage systems. While the increased number of small heat providers and the increasing demand-side flexibility lead to a more complex operation of the system, these developments may at the same time pave the way for liberalized heat markets. This thesis concerns the design of market mechanisms to support the more complex operation of future district heating systems, with three future system characteristics in mind: the presence of many small generators, the fact that generators and flexible loads are distributed over the system, and the presence of thermal storage systems. First, we investigate an alternative to direct market participation of excess heat producers, where these actors are sent a price signal, based on which they self-schedule their production. In a realistic case study we quantify how suboptimal this self-scheduling is in comparison to market participation, in terms of total generation cost. We find that the self-scheduling method may be suitable under low excess heat penetration, while we advise that more sophisticated pricing signals and/or other market setups are used when excess heat covers a significant share of the heat load. Second, we propose a market mechanism that takes heating network constraints into account. We provide insight into the price formation in this network-aware market design, by including peer-to-peer trades that reveal the network losses caused by each market participant. In an illustrative case study, we show that the proposed market design can result in schedules and prices that are beneficial for network operation and reduce its cost, by effectively promoting more local heat consumption. Our third contribution sheds light on a more fundamental issue in market design for non-merchant storage. To cope with time-linking constraints that are present in such markets, it is common to make simplifying assumptions on the end-of-horizon storage level. Using illustrative examples and formal proofs, we analyze which market properties hold under such assumptions, as well as in their absence. In particular, we find that market inefficiencies may even arise given perfect decisions on the end-of-horizon storage level, because prices in subsequent market horizons may fail to reflect the value of stored energy. Finally, we propose a method for restoring market efficiency in a perfect foresight setting.