Stochastic net load optimization in distributed integrated energy systems - a forecast based scheduling approach

The increasing integration of decentralized and volatile producers and consumers across sectors (electricity, heating and mobility) introduces significant operational challenges for distributed energy systems. This work presents a systematic stochastic optimization approach for generating day-ahead operation schedules in sector-integrated energy systems based on probabilistic net load forecasts, for local applications that require low data, low computing power and enable a high level of data security. The proposed method enables more robust decision-making under uncertainty, overcoming the limitations of deterministic point forecasts and optimizations or the requirements for large amounts of data. The approach is demonstrated using the energy system of a logistics facility, focusing on the electrical preconditioning of refrigerated trailers under varying daily preconditioning frequencies. The study outlines how probabilistic net load forecasts can be transformed into representative scenarios and implemented as stochastic net load inputs within the energy system model. A comparative analysis between the scenario-based stochastic optimization and three deterministic optimization variants highlights the advantages of the proposed approach. The stochastic method achieves up to 25 % lower total operating costs and reduces daily peak power exceedances by 30 to 66 % compared to deterministic scheduling. Furthermore, the analysis of regret costs indicates average daily reductions between 21 % and 69 %, depending on the number of reefers to be preconditioned, demonstrating enhanced robustness, cost efficiency, and operational reliability under forecast uncertainty.