Distributed Optimization

February 2022

Optimization of electric power systems has traditionally been centralized with data collected and processed at a control center. For instance, system operators collect all necessary data and then centrally solve optimal power flow problems to schedule generator setpoints. The rapid adoption of distributed energy resources increasingly challenges the scalability of the communicating, modeling, and computing tasks associated with centralized optimization. Distributed optimization addresses these challenges by allowing many local controllers to cooperatively solve large optimization problems using local computations informed by communication with neighboring controllers. Distributed optimization has several potential advantages over centralized approaches, including enhanced robustness against the failures of individual controllers, resilience against failures of supporting cyber systems, faster parallel computations, and increased privacy with limited data communicated to a subset of other controllers. Ongoing research efforts are working to improve distributed optimization algorithms by speeding up convergence rates, enhancing optimization accuracy, extending modeling flexibility, and enhancing robustness to both noisy data from nonideal communications and malicious data from cyberattacks. Demonstration projects are also illustrating the applicability of distributed optimization in practical settings, the need for distributed computing devices, and options for implementation architectures.

Active Committees/Task Forces of Interest

Working Group on Computational Challenges and Solutions for Implementing Distributed Optimization in Power Systems, part of the Analytic Methods for Power Systems (AMPS) Committee, Computing and Analytical Methods (CAMS) Subcommittee.

Working Group on Multi-Agent Systems (MAS), part of the Analytic Methods for Power Systems (AMPS) Committee, Intelligent Systems Applications (ISA) Subcommittee.

Technical Reports & Applicable Papers or Presentations
IEEE Xplore Articles:
Power & Energy Magazine (P&E) Articles:
Other Available Material
  • A.K. Srivastava and D.K. Molzahn, “Distributed Optimization for Electric Power Systems: Needs, Algorithmic Developments, and Use Cases,” 2022 Half-Day Tutorial (March 1-2, 2022). [Tutorial registration available here]
The National Renewable Energy Laboratory (NREL) has dedicated significant effort to distributed optimization for power systems: https://www.nrel.gov/grid/distributed-optimization-control.html IEEE Symposium on Multi-agent System Coordination and Optimization (IEEE MASCO)