Distributed Optimal Power Flow in Unbalanced Distribution Grids With Non-Ideal Communication

Inaolaji A., Savasci A., Paudyal S., Kamalasadan S.

IEEE Transactions on Industry Applications, vol.59, no.5, pp.5385-5397, 2023 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Volume: 59 Issue: 5
  • Publication Date: 2023
  • Doi Number: 10.1109/tia.2023.3283236
  • Journal Name: IEEE Transactions on Industry Applications
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Applied Science & Technology Source, Business Source Elite, Business Source Premier, Communication Abstracts, Compendex, Computer & Applied Sciences, INSPEC, Metadex, Civil Engineering Abstracts
  • Page Numbers: pp.5385-5397
  • Keywords: distributed optimization, Distribution grids, imperfect communication, non-ideal communication, optimal power flow
  • Abdullah Gül University Affiliated: No


The use of distributed optimization for solving optimal power flow (OPF) problems in distribution networks is increasing as the number of controllable devices and the span of service areas increase. However, the performance of the distributed optimization algorithm largely depends on the availability and quality of communication links between the computing entities. In this paper, we evaluate the performance of the alternating direction method of multipliers (ADMM) for solving the OPF problem for volt-VAr optimization (VVO) under non-ideal communication. The overall problem is formulated by considering the unbalanced operation of multi-phase distribution networks. The algorithm's performance is first analyzed, assuming there is an ideal communication between computing entities, which serves as a benchmark for further analysis. Then, two non-ideal communication models, namely (1) sporadic communication failure and (2) communication under a noisy transmission, are applied to the information exchange architecture of the ADMM. Extensive numerical experiments are conducted with two test systems, the IEEE 123-bus feeder and a 2522-bus network scaled down from the IEEE 8500-bus system; these are used to evaluate the performance of the algorithm. The results indicate that the ADMM-based distributed VVO algorithm has deteriorated performance under high levels of communication failure and high levels of noise injection. Additionally, although the centralized-VVO counterpart has some computational superiority when the overall problem is convex and no communication impairments exist, the distributed-VVO has superior immunized solution performance under realistic communication constraints.