A comparative study of benders decomposition and ADMM for decentralized optimal power flow

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

11 Scopus citations

Abstract

Aiming for solving optimal power flow (OPF) problems decentrally, this work offers a comparative analysis between methods representing two distinct families of decomposition techniques: Benders decomposition (BD) from cutting-plane methods and consensus-based alternating direction method of multipliers (ADMM) from dual decomposition-based algorithms. Within the scope of the study, the computational performance and their communication requirements of these methods are compared. The results demonstrate a relative advantage of BD in terms of computational performance, as ADMM mostly requires a considerably larger number of iterations under the same convergence criterion. ADMM in turn has a completely distributed architecture, which allows regulators to withhold local information and realizes a higher parallelization potential.

Original languageEnglish
Title of host publication2020 IEEE Power and Energy Society Innovative Smart Grid Technologies Conference, ISGT 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728131030
DOIs
StatePublished - Feb 2020
Event2020 IEEE Power and Energy Society Innovative Smart Grid Technologies Conference, ISGT 2020 - Washington, United States
Duration: 17 Feb 202020 Feb 2020

Publication series

Name2020 IEEE Power and Energy Society Innovative Smart Grid Technologies Conference, ISGT 2020

Conference

Conference2020 IEEE Power and Energy Society Innovative Smart Grid Technologies Conference, ISGT 2020
Country/TerritoryUnited States
CityWashington
Period17/02/2020/02/20

Keywords

  • ADMM
  • Benders decomposition
  • Decentralized optimization
  • Optimal power flow

Fingerprint

Dive into the research topics of 'A comparative study of benders decomposition and ADMM for decentralized optimal power flow'. Together they form a unique fingerprint.

Cite this