DEED-ADMM: A Scalable Distributed Algorithm for Economic Dispatch in Multi-Energy Systems With Energy Storage
Shanying Zhu, Tao Ding, Cailian Chen, Mo–Yuen Chow, Xinping Guan
Abstract
Multi-energy systems with energy storage can coordinate various energy carriers to facilitate the integration of large amounts of distributed energy sources and promote the overall efficiency of energy use, which needs distributed dispatch with the requirement of security and privacy. This paper studies the distributed economic dispatch based on information from neighboring agents only. In order to handle the non-convexity due to the complementarity constraint of energy storage, it is proved that simultaneous charging and discharging is suboptimal for the multi-energy systems. Based on this, an equivalent convex problem is reformulated. A scalable distributed algorithm based on parallel ADMM and dynamic consensus mechanism, termed DEED-ADMM, is then proposed. It is shown that DEED-ADMM is scalable in terms of per-agent energy consumption and computational complexity with centralized methods. Moreover, under general convex cost functions, convergence properties of DEED-ADMM are theoretically analyzed by adopting the Lyapunov-based approach. It is proved that the primal problem and the dual problem can be simultaneously solved. Finally, case studies demonstrate the effectiveness of the proposed algorithm. Note to Practitioners—This paper is motivated by the problem of coordinating various energy carriers as well as energy storage in multi-energy systems to promote the overall efficiency of energy use. The coupling among different energy carriers and the complementarity constraint of non-simultaneous charging and discharging of battery storage make the problem non-convex. Existing distributed approaches require stringent assumptions on the cost functions, or suffer from a heavy computational burden. To address the above challenges, a fully distributed algorithm is developed, which is scalable and suitable for large-scale systems. Moreover, it is the first distributed algorithm that solves the economic dispatch problem and the dual problem simultaneously in multi-energy systems with general convex cost functions. Practitioners can easily adjust the coefficients of the proposed algorithm to guarantee convergence for the IEEE 30-bus or even 116-bus systems, as long as the economic dispatch problem is feasible. Our future work will focus on designing resilient mechanisms under potential attacks and considering more practical situations such as power loss.