Distributed Discrete Optimization for Autonomous Restoration in DER-Rich Power Distribution Network

Srayashi Konar

doi:10.7273/000006284

Traditionally, advanced distribution management system (ADMS) uses centralized approach for restoration and topology reconfiguration of a power distribution system. However, with increasing penetration of renewable generation with enhanced control variables, energy storage systems, and changing load types, centralized coordination of distributed energy resources for restoration is becoming challenging. Though centralized optimization provides optimal solutions, it also suffers from scalability issues since computation of set points is performed in the ADMS based on all the data collected from sensors in the system. Centralized approach is also prone to single point of failure since it is dependent on network-wide communication for data-delivery. Distributed optimization facilitates parallel computation and distributed decision making while providing near optimum solutions. Advances in sensors, digital automation with remotely controllable switches supports the fully autonomous distributed reconfiguration. Moreover, distributed optimization involves communication only among neighboring agents thereby reducing communication overhead resulting in improved robustness to communication failures and hence enhanced resiliency. This work presents distributed algorithms for optimal restoration for three-phase unbalanced distribution system.The primary contributions of this dissertation are categorized in four sections. First, a decomposable centralized restoration algorithm is developed for an unbalanced power distribution grid. The model can capture effects of grid forming and grid following distributed energy resources in restoration of distribution grid. Second, an open loop model predictive control (MPC) based centralized restoration algorithm is proposed for realization of flexible boundary networked microgrid system in the absence of substations in a feeder. The same formulation is extended for MPC based sequential / multi-step restoration of a distribution grid. The algorithm enables restoration of distribution grid by forming multiple islands around grid forming generators. The sizes of these islands are increased in successive time steps of restoration till the entire network is energized. Third, a novel penalty driven alternating direction method of multipliers (PD-ADMM) based distributed algorithm is proposed for restoration of distribution grid. A detailed scenario assessment is carried out to assess the efficiency of the proposed algorithm. Fourth, a novel distributed discrete proximal atomic coordination algorithm (D^2PAC) is proposed for restoration of distribution grid. Along with performance analysis a comprehensive comparison is carried out to highlight the superiority of D^2PAC over ADMM. The superiority of the proposed algorithms is verified in IEEE 33-bus, 69-bus, 123-bus, and 1069-bus systems.

Distributed Discrete Optimization for Autonomous Restoration in DER-Rich Power Distribution Network

Files and links (1)

Abstract

Metrics

Details