Decentralized optimization approaches for using the load flexibility of electric heating devices

Abstract

Electric heating devices can provide the needed load flexibility for future energy systems with high shares of renewable energies. To exploit these flexibilities, the literature often suggests centralized scheduling-based optimization. However, centralized optimization has crucial drawbacks regarding complexity, privacy and robustness while uncoordinated decentralized optimization approaches yield non-optimal results for the entire system. In this paper, we develop two novel coordinating decentralized optimization approaches, PSCO and PSCO-IDA. Furthermore, we define an optimization procedure to generate a solution pool with diverse schedules for the coordinating approaches. The results show that all investigated approaches for coordinated decentralized optimization lead to lower surplus energy and thus to higher self-consumption rates of locally generated renewable energy compared to the uncoordinated approach. Moreover, using solution pools generated by our optimization procedure strongly improves the Iterative Desync Algorithm (IDA), an effective and privacy-preserving algorithm for decentralized optimization. A comparison of the different decentralized optimization approaches reveals that PSCO-IDA leads to an average improvement of 10% compared to IDA while PSCO leads to similar results with reduced communication effort. All decentralized approaches have significantly reduced runtimes compared to centralized optimization. Our study reveals the strong advantages of coordinated decentralized optimization approaches for using flexible electrical loads.