The integrated provision of energy among various energy sectors plays an important role in the process of decarbonisation of large energy systems. An important pillar is thereby the decarbonisation of the heat sector, where nowadays still a large percentage of heat supply originates from high-emission fossil fuels like coal or oil. In Central Europe, combined heat and power (CHP) plant applications, e.g. in local district heating networks, represent established methods to provide both electricity and heat at the same time, lowering overall fuel demands and lowering concomitant emissions. Heat pumps, converting electricity into heat, are also increasingly adopted by commercial (and household) customers. However, the optimal marketing and production scheduling of the heat and power-providing portfolios under price uncertainty is a challenging and often complex task. The importance of proper uncertainty handling is underscored even more if the optimal dispatch of flexible technologies like storages needs to be considered. In this paper, we propose an enhanced multi-stage stochastic programming model for coordinated bidding in two sequential markets, namely the one-hour and the fifteen-minute electricity products in the German (day-ahead) spot market. Our study develops and applies a stochastic mixed-integer linear programming model for a virtual power plant, acting as a price taker in the mentioned electricity markets. The model determines the optimal bidding strategies for a heterogeneous portfolio of small gas-fired motor-CHP units, heat pumps, electric storage heaters and battery storage systems. Thereby, we introduce a novel approach to construct piece-wise linear bidding curves for these markets, choosing their supporting points based on the simulated price paths. For the evaluation of the benefits of decision-making by help of stochastic modelling and optimization with different scenario numbers, we develop a new concept, the Benefit of Stochastic Optimization (BSO) and reflect and contrast our results with the computational burden of stochastic simulation, using the example of a real-world portfolio. We find that stochastic optimisation is a valuable optimisation method that may inform and improve individual marketer's decision-making processes. However, the observable additional II benefits, i.e. compared to deterministic point forecasts, are limited in the investigated cases, while computational expensiveness increases significantly when adding further scenarios.
